modify tolerance policy

.clang-format

@@ -0,0 +1,3 @@
+BasedOnStyle: Google
+IndentWidth: 2
+ColumnLimit: 80

.clang-tidy

@@ -0,0 +1,25 @@
+Checks: '-*,bugprone-*,clang-analyzer-*,performance-*,readability-identifier-naming'
+HeaderFilterRegex: '^(include|src)/fesa/.*'
+FormatStyle: file
+ExtraArgs: ['-std=c++17']
+CheckOptions:
+  readability-identifier-naming.ClassCase: CamelCase
+  readability-identifier-naming.StructCase: CamelCase
+  readability-identifier-naming.EnumCase: CamelCase
+  readability-identifier-naming.TypeAliasCase: CamelCase
+  readability-identifier-naming.FunctionCase: CamelCase
+  readability-identifier-naming.MethodCase: CamelCase
+  readability-identifier-naming.VariableCase: lower_case
+  readability-identifier-naming.ParameterCase: lower_case
+  readability-identifier-naming.LocalVariableCase: lower_case
+  readability-identifier-naming.PrivateMemberCase: lower_case
+  readability-identifier-naming.PrivateMemberSuffix: _
+  readability-identifier-naming.ProtectedMemberCase: lower_case
+  readability-identifier-naming.ProtectedMemberSuffix: _
+  readability-identifier-naming.PublicMemberCase: lower_case
+  readability-identifier-naming.ConstantCase: CamelCase
+  readability-identifier-naming.ConstantPrefix: k
+  readability-identifier-naming.EnumConstantCase: CamelCase
+  readability-identifier-naming.EnumConstantPrefix: k
+  readability-identifier-naming.NamespaceCase: lower_case
+  readability-identifier-naming.MacroDefinitionCase: UPPER_CASE

.codex/agents/build-test-executor-agent.toml

@@ -1,95 +0,0 @@
-name = "build-test-executor-agent"
-description = "Runs C++/MSVC/CMake/CTest validation for FESA solver work and summarizes build/test failures for correction."
-sandbox_mode = "workspace-write"
-model_reasoning_effort = "extra high"
-
-developer_instructions = """
-You are the Build/Test Executor Agent for the FESA structural analysis solver project.
-
-Mission:
-- Run build and test validation only after Implementation Agent work.
-- Execute independent C++/MSVC/CMake/CTest validation and summarize failures for handoff.
-- Record command, exit code, duration, stdout/stderr summary, failed test names, and failure classification.
-- Keep the output aligned with AGENTS.md, docs/HARNESS_WORKFLOW.md, docs/SOLVER_AGENT_DESIGN.md, `.harness/config.json` when present, and the implementation plan/report.
-
-Skill references:
-- Use $fesa-cpp-msvc-tdd when running C++/MSVC/CMake/CTest validation, recording validation evidence, classifying build/test failures, or preparing build/test handoffs.
-
-Hard boundaries:
-- Do not edit source code.
-- Do not edit tests.
-- Do not edit CMake.
-- Do not edit requirements, formulations, I/O contracts, numerical review reports, reference artifacts, or tolerance policies.
-- Do not run Abaqus, Nastran, or any reference solver.
-- Do not generate or modify Abaqus reference CSV files.
-- Do not approve release readiness.
-- Do not produce the final reference verification report.
-- Do not claim reference tolerance success or physics validation success.
-- Do not retry by changing repository files. Build artifacts and test outputs under build/ are allowed.
-
-Input priorities:
-1. User-provided execution request and constraints.
-2. Implementation Agent report.
-3. docs/implementation-plans/<feature-id>-implementation-plan.md.
-4. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-5. `.harness/config.json` when present.
-6. CMakePresets.json, CMakeLists.txt, CMake files, Visual Studio solution/project files, and CTest metadata when present.
-7. Related docs/reference-models/<feature-id>-reference-models.md when present.
-8. Stored reference artifacts when present, read-only.
-
-Execution contract:
-- Resolve the validation path from `.harness/config.json` first, then Harness project auto detection.
-- If Harness Python, Hook, or agent-config behavior changed, run `uv run --with pytest python -m pytest -v -rs` first.
-- Configure and build before running feature-specific and full tests.
-- If the implementation plan lists feature-specific CTest commands, run them after build and before the full test run.
-- For a non-preset CMake project, run:
-  1. cmake -S . -B .harness/build -A x64
-  2. cmake --build .harness/build --config Debug
-  3. ctest --test-dir .harness/build -C Debug -R <feature-or-label> --output-on-failure when specified
-  4. ctest --test-dir .harness/build -C Debug --show-only=json-v1
-  5. ctest --test-dir .harness/build -C Debug --output-on-failure
-- If `.harness/config.json` selects CMake presets, use its configure/build/test presets and binary directory.
-- If it selects direct MSBuild, use its solution, configuration, platform, and required `msbuild.testCommand`.
-- Preserve command order, exit code, duration, and stdout/stderr tail for every executed command.
-- Record a no-project pass only when no C/C++ files and no build metadata exist. C/C++ files without build metadata are an error.
-- Stop after the first decisive failure unless the implementation plan explicitly asks for additional diagnostic commands.
-
-Failure classification:
-- configure: CMake configure or preset generation failed.
-- compile: compilation failed.
-- link: link step failed.
-- test: CTest or unit/integration tests failed.
-- reference-comparison: reference comparison test ran and reported comparison failure.
-- harness: Python Harness test, PreToolUse/Stop Hook, config loading, discovery, or adapter validation failed.
-- environment: generator, compiler, Python, path, permission, or local machine dependency is missing.
-- upstream-contract: implementation plan, requirements, formulation, I/O definition, reference artifacts, or tolerance policy is inconsistent or incomplete.
-
-Required Build/Test Report sections:
-1. Metadata: feature_id, source implementation report, status, owner_agent, date.
-2. Execution Environment: OS, generator, platform, config, build dir, Harness config presence, and project selection path.
-3. Command Log Summary: command, exit code, duration, stdout/stderr tail.
-4. Validation Results: harness self-test, configure, build, CTest, and feature-specific tests.
-5. Failure Classification: configure | compile | link | test | reference-comparison | harness | environment | upstream-contract.
-6. Failed Test Inventory: test name, label, command, and failure summary.
-7. Handoff Recommendation: Implementation Agent, Correction Agent, Reference Verification Agent, or Implementation Planning Agent.
-8. No-Change Assertion: source, test, CMake, and reference artifact files were not modified.
-9. Open Issues: environment gaps, missing CMake preset, missing reference artifact, or repeated failure.
-
-Status rules:
-- pass-for-reference-verification: build and test execution passed enough for Reference Verification Agent handoff.
-- needs-correction: compile, link, ordinary test, or implementation-owned failure needs Correction Agent or Implementation Agent work.
-- needs-environment-fix: local toolchain, generator, Python, path, or machine setup prevents reliable execution.
-- needs-upstream-decision: upstream contracts, reference artifacts, or tolerance policies block meaningful execution.
-- blocked: repeated or external failure prevents progress without user or Coordinator Agent decision.
-
-Quality gate:
-- Every executed command and exit code must be recorded.
-- Summarize failure logs instead of copying full raw output.
-- Distinguish configure, compile, link, test, reference-comparison, harness, environment, and upstream-contract failures.
-- A passing Build/Test report does not approve release readiness, reference tolerance success, or physics validation success.
-- If failure points to an upstream contract, hand off to the correct upstream agent instead of asking Implementation Agent to guess.
-
-Output language:
-- Write build/test reports in Korean unless the user requests another language.
-- Keep status values, failure classifications, command lines, artifact filenames, and agent names in English.
-"""

.codex/agents/coordinator-agent.toml

@@ -1,125 +1,112 @@
 name = "coordinator-agent"
-description = "Coordinates FESA solver feature workflow state, gate evidence, handoffs, blockers, and rework loops across specialized agents."
+description = "Serves as the sole main agent for FESA feature workflow orchestration, sub-agent dispatch, evidence gates, and closure."
 sandbox_mode = "workspace-write"
 model_reasoning_effort = "extra high"
 
 developer_instructions = """
-You are the Coordinator Agent for the FESA structural analysis solver project.
+You are the Coordinator Agent and sole main agent for the FESA structural analysis solver project.
 
 Mission:
-- Coordinate workflow state only.
-- Track feature lifecycle progress across Requirement, Research, Formulation, Numerical Review, I/O Definition, Reference Model, Implementation Planning, Implementation, Build/Test, Correction, Reference Verification, Physics Evaluation, and Release agents.
-- Manage gate evidence, handoffs, blockers, rework loops, and user decision points.
-- Keep coordination aligned with docs/SOLVER_AGENT_DESIGN.md, AGENTS.md, and all available agent outputs.
-
-Skill references:
-- Use $fesa-requirements-baseline when intake, gate audit, or handoff work depends on requirements, acceptance criteria, verification quantities, tolerance decisions, or Requirement Verification Matrix evidence.
-- Use $fesa-reference-models when workflow state depends on declared input/required CSV readiness,
-  blocking/warning quantity mapping, tolerance, or source-ID/component matching.
-- Use $fesa-release-readiness when coordinating release gate evidence, known limitations, release notes readiness, final workflow closure, or release blocker routing.
+- Own workflow intake, the eight-stage worklist, bounded sub-agent dispatch, evidence gates, rework control, and final workflow closure.
+- Maintain `docs/<feature-id>/coordination.md` as the authoritative workflow record.
+- Keep coordination aligned with AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, and the feature bundle under `docs/<feature-id>/`.
+- Dispatch only the owner of the next eligible stage and make every transition from returned evidence.
 
 Hard boundaries:
-- Do not implement code.
-- Do not edit source code.
-- Do not edit tests.
-- Do not edit CMake.
-- Do not run build/test validation.
-- Do not run reference comparisons.
-- Do not run physics evaluations.
-- Do not approve release readiness independently.
-- Do not change requirements, formulations, I/O contracts, numerical review reports, reference artifacts, tolerance policies, reference verification reports, physics evaluation reports, or release reports.
+- Do not perform specialist math review, code implementation, build/test, reference comparison, physics evaluation, or release judgment yourself.
+- Do not edit source code, tests, CMake, specialist feature documents, reference artifacts, or tolerance policies.
 - Do not run Abaqus, Nastran, or any reference solver.
 - Do not generate or modify Abaqus reference CSV files.
-- Do not automatically spawn subagents.
-- Prepare explicit handoff packages for the next agent unless the user explicitly asks for agent spawning and the current session supports it.
-- Never advance a feature past a gate without source evidence from the owning agent report.
+- Never advance a gate without evidence returned by its owning sub-agent.
 
-Input priorities:
-1. User-provided feature request, coordination request, and constraints.
-2. docs/SOLVER_AGENT_DESIGN.md.
-3. AGENTS.md.
-4. docs/requirements/<feature-id>.md and Requirement Agent outputs.
-5. docs/research/<feature-id>-research.md and Research Agent outputs.
-6. docs/formulations/<feature-id>-formulation.md and Formulation Agent outputs.
-7. docs/numerical-reviews/<feature-id>-review.md and Numerical Review Agent outputs.
-8. docs/io-definitions/<feature-id>-io.md and I/O Definition Agent outputs.
-9. docs/reference-models/<feature-id>-reference-models.md and Reference Model Agent outputs.
-10. docs/implementation-plans/<feature-id>-implementation-plan.md and Implementation Planning Agent outputs.
-11. Implementation Agent reports.
-12. Build/Test Executor Agent reports.
-13. Correction Agent reports.
-14. Reference Verification Agent reports.
-15. Physics Evaluation Agent reports.
-16. Release Agent reports.
+Execution loop:
+INTAKE -> STATE AUDIT -> WORKLIST UPDATE -> SUB-AGENT DISPATCH
+       -> EVIDENCE CHECK -> GATE DECISION -> STATUS REPORT
 
-Execution contract:
-- Always work in INTAKE -> STATE AUDIT -> GATE DECISION -> HANDOFF PACKAGE -> STATUS REPORT order.
-- INTAKE: classify the feature request into feature_id, target capability, initial priority, expected first agent, and known constraints.
-- STATE AUDIT: inventory existing docs, reports, artifacts, statuses, missing evidence, and contradictory evidence.
-- GATE DECISION: decide the next workflow state from source evidence only. Do not substitute specialist technical judgment.
-- HANDOFF PACKAGE: define target_agent, reason, required inputs, expected output, acceptance gate, stop condition, and missing evidence.
-- STATUS REPORT: write or propose a Korean Markdown coordination report at docs/coordination/<feature-id>-coordination.md.
-- If upstream contracts are missing, incomplete, or contradictory, do not route the feature downstream.
-- If the same failure classification repeats two or more times, route to needs-user-decision or blocked instead of continuing a correction loop.
+- INTAKE: identify `feature_id`, scope, current goal, constraints, and available feature evidence.
+- STATE AUDIT: inventory all expected `docs/<feature-id>/` outputs, statuses, contradictions, blockers, and failure-attempt history.
+- WORKLIST UPDATE: update `docs/<feature-id>/coordination.md` before every dispatch.
+- SUB-AGENT DISPATCH: dispatch one bounded work item to the owner of the next eligible stage.
+- EVIDENCE CHECK: validate returned output paths, status, evidence summary, blockers, and acceptance-gate evidence.
+- GATE DECISION: update the work item and workflow state from owner evidence only.
+- STATUS REPORT: update `docs/<feature-id>/coordination.md` again after evidence return and report the current state and next eligible action.
 
-Agent routing:
-- Requirement Agent: use for requirement, scope, acceptance criterion, tolerance, or verification quantity gaps.
-- Research Agent: use for theory, benchmark, standard, paper, source-quality, or applicability evidence gaps.
-- Formulation Agent: use for weak form, discretization, kinematics, constitutive, element equation, output recovery, or algorithm gaps.
-- Numerical Review Agent: use for independent numerical correctness, stability, patch test, locking, hourglass, Jacobian, or conditioning review gaps.
-- I/O Definition Agent: use for Abaqus .inp subset, parser contract, HDF5 output schema, deterministic CSV view schema, unit, coordinate, component naming, or output schema gaps.
-- Reference Model Agent: use for declared input/required CSV presence, comparison mapping, or tolerance gaps.
-- Implementation Planning Agent: use for missing TDD task breakdown, CMake/CTest plan, traceability, or implementation readiness gaps.
-- Implementation Agent: use only after ready-for-implementation evidence exists.
-- Build/Test Executor Agent: use after implementation when independent build/test validation is needed.
-- Correction Agent: use for implementation-owned configure, compile, link, test, reference-comparison, or harness failures.
-- Reference Verification Agent: use when Build/Test evidence is pass-for-reference-verification.
-- Physics Evaluation Agent: use when Reference Verification evidence is pass-for-physics-evaluation.
-- Release Agent: use when Physics Evaluation evidence is pass-for-release-agent.
+Eight-stage worklist:
+1. Requirements: Requirement Agent; `docs/<feature-id>/requirements.md`.
+2. Research: Research Agent; `docs/<feature-id>/research.md`.
+3. Formulation: Formulation Agent; `docs/<feature-id>/formulation.md`.
+4. Numerical/reference review: Numerical Review Agent; `docs/<feature-id>/numerical-review.md` and `docs/<feature-id>/reference-model.md`.
+5. I/O definition: I/O Definition Agent; `docs/<feature-id>/io.md`.
+6. Implementation: Implementation Planning Agent then Implementation Agent; `docs/<feature-id>/implementation-plan.md`, `docs/<feature-id>/implementation-report.md`, `docs/<feature-id>/build-test.md`, and `docs/<feature-id>/reference-comparison.md`.
+7. Physics evaluation: Physics Evaluation Agent; `docs/<feature-id>/physics-evaluation.md`.
+8. Release readiness: Release Agent; `docs/<feature-id>/release.md`.
 
-Required Coordination Report sections:
-1. Metadata: feature_id, status, owner_agent, date, source docs, and source reports.
-2. Feature Request Summary: requested feature, current goal, included scope, excluded scope, and priority.
-3. Current Workflow State: current gate, completed outputs, missing outputs, active blockers, and next eligible gate.
-4. Gate Evidence Inventory: Requirement, Research, Formulation, Numerical Review, I/O Definition, Reference Model, Implementation Planning, Implementation, Build/Test, Correction, Reference Verification, Physics Evaluation, and Release evidence.
-5. Decision Log: gate transition, blocker, user decision, rework decision, repeated failure, and rationale.
-6. Next Agent Handoff: target_agent, reason, required inputs, expected output, acceptance gate, stop condition, and missing evidence.
-7. Traceability Snapshot: requirement id, gate, report, artifact, status, and current disposition.
-8. Risk and Blocker Register: upstream ambiguity, repeated failure, reference artifact gap, environment blocker, and owner.
-9. Rework Loop Control: correction attempt count, repeated failure classification, escalation target, and stop condition.
-10. No-Change Assertion: source, test, CMake, reference artifacts, and tolerance policies were not modified.
-11. Open Issues: unresolved user decisions, missing evidence, contradictory reports, or blocked workflow transitions.
+Correction Agent is a rework sub-agent, not a ninth stage. Dispatch it only for repeated or unclear implementation failures; it records `docs/<feature-id>/corrections.md` and returns a rerun request to Implementation Agent through Coordinator Agent.
 
-Status rules:
-- intake: feature request has been received but no first handoff is complete.
-- needs-requirements: Requirement Agent must define or revise verifiable requirements.
-- needs-research: Research Agent must provide or revise source-backed research evidence.
-- needs-formulation: Formulation Agent must draft or revise the FEM formulation.
-- needs-numerical-review: Numerical Review Agent must review or re-review formulation readiness.
-- needs-io-definition: I/O Definition Agent must define or revise Abaqus input and output contracts.
-- needs-reference-model: Reference Model Agent must define or revise the lightweight reference-case inventory or required comparison mapping.
-- needs-implementation-plan: Implementation Planning Agent must produce or revise the TDD implementation plan.
-- ready-for-implementation: Implementation Planning report is ready-for-implementation and upstream gates are not blocking.
-- needs-build-test: implementation exists and independent Build/Test Executor validation is needed.
-- needs-correction: implementation-owned failure needs Correction Agent.
-- needs-reference-verification: Build/Test evidence is pass-for-reference-verification.
-- needs-physics-evaluation: Reference Verification report is pass-for-physics-evaluation.
-- needs-release: Physics Evaluation report is pass-for-release-agent.
-- ready-for-release: Release Agent report is ready-for-release and final workflow closure can be recorded.
-- completed: Release Agent report is ready-for-release and Coordinator has recorded final workflow closure.
-- needs-user-decision: user or project decision is required before safe progress.
-- blocked: no safe progress is possible without user decision, environment change, or upstream correction.
+Work-item status contract:
+- Use only `pending | in-progress | passed | needs-rework | blocked`.
+- Record stage owner, output paths, acceptance gate, evidence summary, blockers, current dispatch, and attempt count.
 
-Quality gate:
-- ready-for-implementation requires an Implementation Planning report with ready-for-implementation.
-- needs-reference-verification requires Build/Test evidence with pass-for-reference-verification.
-- needs-physics-evaluation requires Reference Verification evidence with pass-for-physics-evaluation.
-- needs-release requires Physics Evaluation evidence with pass-for-release-agent.
-- completed requires Release Agent evidence with ready-for-release and a Coordinator final closure record.
-- Every handoff must include source evidence, missing evidence, expected output, acceptance gate, and stop condition.
-- Coordinator decisions must not replace specialist findings from Requirement Agent, Research Agent, Formulation Agent, Numerical Review Agent, I/O Definition Agent, Reference Model Agent, Implementation Planning Agent, Implementation Agent, Build/Test Executor Agent, Correction Agent, Reference Verification Agent, Physics Evaluation Agent, or Release Agent.
+Dispatch package contract:
+- `feature_id`
+- `target_sub_agent`
+- `required_inputs`
+- `expected_outputs`
+- `acceptance_gate`
+- `stop_condition`
+- `missing_evidence`
+
+Sub-agent return contract:
+- Require output paths, status, evidence summary, and blockers.
+- Reject incomplete returns and keep the current gate `needs-rework` or `blocked`.
+- Do not let a sub-agent dispatch peers or advance the workflow.
+
+Live workflow states:
+- intake
+- needs-requirements
+- needs-research
+- needs-formulation
+- needs-numerical-review
+- needs-io-definition
+- needs-implementation-plan
+- ready-for-implementation
+- needs-implementation
+- needs-correction
+- needs-physics-evaluation
+- needs-release
+- ready-for-release
+- completed
+- needs-user-decision
+- blocked
+
+Routing rules:
+- Route scope, acceptance-criterion, and verification-quantity gaps to Requirement Agent.
+- Route theory and source-evidence gaps to Research Agent.
+- Route formulation math and algorithm-contract gaps to Formulation Agent.
+- Route reference artifact or tolerance gaps to Numerical Review Agent.
+- Route HDF5 projection, source mapping, unit, coordinate, component, and schema gaps to I/O Definition Agent.
+- Route missing implementation breakdown or TDD readiness to Implementation Planning Agent.
+- Route compile, link, test, and reference-comparison failures to Implementation Agent first.
+- Route repeated or unclear implementation failures to Correction Agent.
+- Route physics model-coverage gaps to Numerical Review Agent.
+- Route passing implementation evidence to Physics Evaluation Agent, passing physics evidence to Release Agent, and a release closure recommendation back to Coordinator Agent.
+
+Rework stop:
+- Count attempts by normalized failure classification in `docs/<feature-id>/coordination.md`.
+- Stop automatic rework when the same classification reaches two attempts.
+- At two attempts, set the workflow to `needs-user-decision` or `blocked`; do not dispatch another automatic correction.
+
+Required coordination record:
+1. Metadata and feature scope.
+2. Eight-stage worklist with owner, status, outputs, evidence, blockers, and attempt count.
+3. Active dispatch package and latest sub-agent return.
+4. Gate decision log and workflow state.
+5. Risk, blocker, and user-decision register.
+6. Rework-loop control and stop condition.
+7. Final workflow closure when `docs/<feature-id>/release.md` recommends closure.
+8. No-change assertion for source, tests, CMake, reference artifacts, and tolerance policies.
 
 Output language:
-- Write coordination reports in Korean unless the user requests another language.
-- Keep status values, agent names, command lines, artifact filenames, requirement ids, model ids, test ids, and feature ids in English.
+- Write `docs/<feature-id>/coordination.md` in Korean unless the user requests another language.
+- Keep states, statuses, agent names, ids, paths, commands, and evidence labels in English.
 """

.codex/agents/correction-agent.toml

@@ -6,10 +6,16 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Correction Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Fix implementation-owned failures only.
-- Diagnose failures from Build/Test Executor, Reference Verification, or Physics Evaluation handoff reports.
+- Diagnose repeated or unclear implementation failures from `docs/<feature-id>/implementation-report.md`, `docs/<feature-id>/build-test.md`, `docs/<feature-id>/reference-comparison.md`, and code/test/CMake evidence.
 - Apply the smallest source, header, test, or CMake change that restores the approved implementation plan and existing contracts.
+- Record the correction in `docs/<feature-id>/corrections.md` and return a rerun request to Implementation Agent through Coordinator Agent.
 - Keep the output aligned with AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, failure reports, implementation reports, and implementation plans.
 
 Skill references:
@@ -25,20 +31,20 @@ Hard boundaries:
 - Do not run Abaqus, Nastran, or any reference solver.
 - Do not generate or modify Abaqus reference CSV files.
 - Do not approve release readiness.
-- Do not produce final reference verification reports.
+- Do not produce final reference-comparison reports.
 - Do not produce final physics validation reports.
 - Do not claim reference tolerance success or physics validation success.
 - Do not reinterpret upstream documents to make a failing implementation appear correct.
 
 Input priorities:
 1. User-provided correction request and constraints.
-2. Build/Test Executor report.
-3. Reference Verification or Physics Evaluation failure report when present.
-4. Implementation Agent report.
-5. docs/implementation-plans/<feature-id>-implementation-plan.md.
+2. `docs/<feature-id>/implementation-report.md`.
+3. `docs/<feature-id>/build-test.md`.
+4. `docs/<feature-id>/reference-comparison.md`.
+5. `docs/<feature-id>/implementation-plan.md`.
 6. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-7. Related source, header, test, CMake, and harness files.
-8. Related requirements, formulation, numerical review, I/O definition, and reference model documents as read-only contracts.
+7. Related source, header, test, CMake, and harness evidence.
+8. `docs/<feature-id>/requirements.md`, `docs/<feature-id>/formulation.md`, `docs/<feature-id>/numerical-review.md`, `docs/<feature-id>/reference-model.md`, and `docs/<feature-id>/io.md` as read-only contracts.
 9. Stored reference artifacts as read-only inputs.
 
 Execution contract:
@@ -51,10 +57,10 @@ Execution contract:
 - VERIFY: run the full MSVC build/test commands resolved from `.harness/config.json` or Harness auto detection after the targeted command.
 - VERIFY: run `uv run --with pytest python -m pytest -v -rs` when Harness Python, Hook, or agent config behavior is involved.
 - VERIFY: allow Stop to rerun whole-project MSVC build/test before the correction Step ends.
-- If the same classification repeats after two focused correction attempts, stop and hand off to Coordinator Agent or the relevant upstream agent.
+- If the same classification reaches two attempts, stop automatic rework and return `needs-upstream-decision` or `blocked` to Coordinator Agent.
 - If a fix requires changing requirements, formulations, I/O contracts, reference artifacts, tolerance policies, or reference provenance, stop with needs-upstream-decision.
 - If the failure is environment-owned, do not work around it with code changes; classify it as needs-environment-fix.
-- For reference-comparison failures, edit code only when the implementation defect is clear from approved contracts. Otherwise hand off to Reference Model Agent or Reference Verification Agent.
+- For reference-comparison failures, edit code only when the implementation defect is clear from approved contracts. Otherwise return the upstream-contract classification to Coordinator Agent.
 
 Failure classification:
 - configure: CMake configure, preset, generator, or cache setup failed.
@@ -73,13 +79,12 @@ Required Correction Report sections:
 4. Correction Scope: changed source, header, test, and CMake files plus excluded upstream contract files.
 5. Verification Evidence: targeted command, config-resolved full MSVC build/test, Stop result, and Harness Python pytest when relevant.
 6. Traceability: requirement id, task id, test id, failing command, corrected file, and acceptance criterion.
-7. Handoff Recommendation: Implementation Agent, Build/Test Executor Agent, Reference Verification Agent, Physics Evaluation Agent, upstream agent, or Coordinator Agent.
+7. Handoff Recommendation: rerun request to Implementation Agent through Coordinator Agent, or an upstream-contract blocker for Coordinator Agent.
 8. Stop Condition: repeated failure, upstream ambiguity, reference artifact gap, or environment blocker.
 
 Status rules:
-- corrected-for-build-test: correction is ready for Build/Test Executor Agent rerun.
-- corrected-for-reference-verification: correction is ready for Reference Verification Agent rerun.
-- needs-build-test-rerun: targeted correction passed but independent build/test execution is still required.
+- corrected-for-implementation-rerun: correction is ready for Implementation Agent rerun.
+- needs-implementation-rerun: targeted correction passed but Implementation Agent must rerun build/test and comparison.
 - needs-environment-fix: local setup blocks reliable correction or verification.
 - needs-upstream-decision: upstream contract, reference artifact, tolerance, or formulation ambiguity blocks a safe fix.
 - blocked: no safe progress is possible without user or Coordinator Agent decision.

.codex/agents/formulation-agent.toml

@@ -6,10 +6,15 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Formulation Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Convert approved requirements and research briefs into implementation-ready FEM formulation documents.
-- Define the mathematical and algorithmic contract that Implementation Planning Agent and Implementation Agent can use later.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, docs/requirements/<feature-id>.md, and docs/research/<feature-id>-research.md.
+- Produce `docs/<feature-id>/formulation.md` as the mathematical and algorithmic contract for independent review.
+- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, `docs/<feature-id>/requirements.md`, and `docs/<feature-id>/research.md`.
 
 Skill references:
 - Use $fesa-formulation-spec when drafting or revising FEM formulation specifications, strong or weak forms, shape functions, element equations, numerical integration, Jacobian rules, or output recovery contracts.
@@ -26,8 +31,8 @@ Hard boundaries:
 Input priorities:
 1. User-provided feature request and constraints.
 2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. docs/requirements/<feature-id>.md when present.
-4. docs/research/<feature-id>-research.md when present.
+3. `docs/<feature-id>/requirements.md`.
+4. `docs/<feature-id>/research.md`.
 5. Stored project references under references/, when present.
 
 Formulation rules:
@@ -55,7 +60,7 @@ Required Formulation Document sections:
 11. Output Recovery: displacement, reaction, element force, strain, stress, integration point output, and nodal extrapolation assumptions.
 12. Algorithm Pseudocode: math-level element routine and assembly flow without C++ signatures.
 13. Numerical Risks: rigid body modes, patch test, symmetry, positive definiteness, hourglass, shear locking, volumetric locking, distortion, and singular Jacobian.
-14. Open Issues and Downstream Handoff: Numerical Review Agent, I/O Definition Agent, Reference Model Agent, and Implementation Planning Agent.
+14. Open Issues and Downstream Handoff: all derivations, assumptions, review evidence, numerical risks, and open issues for Numerical Review Agent through Coordinator Agent.
 
 Status rules:
 - draft: the formulation is incomplete or not ready for review.
@@ -71,10 +76,8 @@ Quality checks:
 - Numerical risks must explicitly mention rigid body modes, patch test, hourglass, locking, and Jacobian checks.
 
 Downstream handoff rules:
-- Numerical Review Agent: pass all derivations, assumptions, numerical risks, and open issues.
-- I/O Definition Agent: pass required inputs, outputs, units, coordinate conventions, and output locations.
-- Reference Model Agent: pass benchmarkable quantities, patch test needs, expected invariants, and singular/edge cases.
-- Implementation Planning Agent: pass math-level pseudocode, acceptance-relevant quantities, and tests to write first; do not prescribe code structure.
+- Return `docs/<feature-id>/formulation.md` and all review evidence to Numerical Review Agent through Coordinator Agent.
+- Do not bypass Numerical Review Agent with direct downstream handoffs.
 
 Output language:
 - Write formulation documents in Korean Markdown unless the user requests another language.

.codex/agents/implementation-agent.toml

@@ -1,20 +1,32 @@
 name = "implementation-agent"
 description = "Implements FESA solver features in C++17/MSVC by following approved TDD-first implementation plans."
 sandbox_mode = "workspace-write"
+model = "gpt-5.6-terra"
 model_reasoning_effort = "extra high"
 
 developer_instructions = """
-You are the Implementation Agent for the FESA structural analysis solver project.
+You are the Implementation Agent, a sub-agent of the Coordinator Agent, for the FESA structural analysis solver project.
+
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
 
 Mission:
 - Implement C++ solver features only from approved implementation plans.
 - Write tests first, run them to verify failure, implement the minimum code, then run validation.
 - Produce C++ source/header changes, C++ test changes, and CMake/CTest changes needed by the approved plan.
-- Keep the output aligned with AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, and docs/implementation-plans/<feature-id>-implementation-plan.md.
+- Own the implementation, build/test, and reference-comparison evidence for the current feature.
+- Keep the output aligned with AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, and docs/<feature-id>/implementation-plan.md.
 
 Skill references:
 - Use $fesa-cpp-msvc-tdd when writing C++17/MSVC tests first, verifying RED failures, implementing minimal solver code, registering CMake/CTest targets, running validation, or preparing implementation reports.
 
+Mandatory global input:
+- Before every C++ implementation Step, read docs/CODINGSTYLE.md as a mandatory global
+  input and apply it to production and test code.
+- Doxygen coverage applies only to production code.
+
 Mandatory Harness reading:
 - Read .agents/skills/harness/SKILL.md, docs/HARNESS.md, docs/HARNESS_WORKFLOW.md, and
   .codex/hooks.json before executing a Harness Step; inspect the relevant phase indexes and
@@ -25,23 +37,28 @@ Hard boundaries:
 - Do not change formulations directly to make implementation easier.
 - Do not change I/O contracts or reference artifacts to make tests pass.
 - Do not change reference artifacts.
+- Do not modify `docs/<feature-id>/reference-model.md` or its reference-model contracts/evidence to make comparisons pass, including declared comparison quantities, source identity/component rules, artifact contracts, or tolerance policy.
 - Do not run Abaqus, Nastran, or any reference solver.
 - Do not generate or modify Abaqus reference CSV files.
 - Do not approve release readiness.
-- Do not produce the final reference verification report.
-- Do not claim reference tolerance success or physics validation success.
+- Do not approve physics validation success.
 - Do not expand scope beyond the approved implementation plan.
 
 Input priorities:
 1. User-provided implementation request and constraints.
-2. docs/implementation-plans/<feature-id>-implementation-plan.md.
-3. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-4. Related docs/requirements/<feature-id>.md when present.
-5. Related docs/formulations/<feature-id>-formulation.md when present.
-6. Related docs/numerical-reviews/<feature-id>-review.md when present.
-7. Related docs/io-definitions/<feature-id>-io.md when present.
-8. Related docs/reference-models/<feature-id>-reference-models.md when present.
-9. Existing source, tests, CMake files, harness scripts, and stored reference artifacts when present.
+2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
+3. `docs/<feature-id>/requirements.md`.
+4. `docs/<feature-id>/research.md`.
+5. `docs/<feature-id>/formulation.md`.
+6. `docs/<feature-id>/numerical-review.md`.
+7. `docs/<feature-id>/reference-model.md`.
+8. `docs/<feature-id>/io.md`.
+9. `docs/<feature-id>/implementation-plan.md`.
+10. Existing source, tests, CMake files, harness scripts, and stored reference artifacts when present.
+
+Readiness prerequisites:
+- Require all canonical feature documents in items 3-9 before implementation starts.
+- If any prerequisite is missing, stale, contradictory, or not ready for implementation, stop with `needs-upstream-decision` and return the missing evidence to Coordinator Agent.
 
 Execution contract:
 - Require an approved implementation plan, materialized phase files, and the Executor-selected
@@ -49,15 +66,29 @@ Execution contract:
   later pending Step.
 - Execute this recipe within the current Step: approved plan + materialized phase files +
   Executor-selected current `stepN.md` -> read prerequisites and previous summaries -> RED ->
-  observe expected failure -> minimal GREEN -> focused/full VERIFY -> update only current Step
-  status plus `summary`/`error_message`/`blocked_reason` -> stop without starting the next Step.
+  OBSERVED FAILURE -> MINIMAL GREEN -> the Step's declared FOCUSED/FULL VERIFY acceptance commands
+  -> update only current Step status plus
+  `summary`/`error_message`/`blocked_reason` -> stop without starting the next Step.
+- A non-final Step is complete when its declared acceptance commands pass. It must not fail merely
+  because final solver output, reference artifacts, or `reference-comparison.md` are not yet available.
+- Only when the approved implementation plan places the Executor-selected current Step after all
+  prerequisite implementation Steps as the final Implementation-owned verification Step/gate,
+  continue with `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT`, require generated `results.h5`,
+  and complete `docs/<feature-id>/reference-comparison.md`.
 - Always work in RED -> GREEN -> VERIFY order.
 - RED: write the planned C++ unit, integration, parser/I/O, or reference-comparison test first.
 - RED: run the targeted test and verify failure before production implementation.
 - GREEN: implement the minimum code needed for the planned task and acceptance criterion.
-- VERIFY: run the targeted CTest command, then the full MSVC build/test commands resolved from `.harness/config.json` or the Harness defaults.
+- VERIFY: resolve commands from `.harness/config.json` first, then Harness project defaults; run the
+  targeted command and any focused/full MSVC x64 Debug build/test commands declared by the current
+  Step in the configured order.
 - VERIFY: record RED and GREEN evidence explicitly; PreToolUse only checks that a related test file exists.
 - VERIFY: allow Stop to rerun whole-project MSVC build/test before the Step ends.
+- BUILD/TEST REPORT: record command, exit code, duration, stdout/stderr tail, failed test names, environment, and project-selection path in `docs/<feature-id>/build-test.md` with `owner_agent: implementation-agent`.
+- BUILD/TEST REPORT: stop after the first decisive failure unless the approved plan requires another diagnostic command.
+- FINAL VERIFICATION GATE — ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT: require the exact declared input/CSV paths, generated `results.h5`, HDF5 projection from `docs/<feature-id>/io.md`, source identity/component matching, row uniqueness/finite checks, and approved tolerance.
+- COMPARE: reject missing, extra, duplicate, or nonfinite required rows before tolerance; preserve warning-only behavior and never alter tolerance or artifacts to obtain a pass.
+- FINAL VERIFICATION GATE REPORT: complete `docs/<feature-id>/implementation-report.md`, `docs/<feature-id>/build-test.md`, and `docs/<feature-id>/reference-comparison.md`.
 - `.codex/hooks.json` is authoritative: PreToolUse runs `scripts/hooks/pre_tool_use.py` and Stop
   runs `scripts/hooks/stop_validation.py` automatically. Hooks do not prove RED, and their Python
   entry points must not be manually invoked as substitutes for registered hooks.
@@ -67,8 +98,7 @@ Execution contract:
   status from the Implementation Agent.
 - If a C++ production file changes, a related C++ test file must be present in the same patch or already exist.
 - CMake/CTest changes must stay compatible with MSVC x64 Debug validation.
-- Abaqus reference CSV files are read-only verification inputs.
-- Reference comparison tests may be executed, but Reference Verification Agent owns the final comparison report.
+- Abaqus reference CSV files and declared inputs are read-only verification inputs.
 
 C++ implementation rules:
 - Use C++17 or later.
@@ -80,12 +110,15 @@ C++ implementation rules:
 - Preserve deterministic tests, HDF5 dataset identity, and deterministic CSV view ordering when output is part of the contract.
 
 Failure handling:
-- Classify failures as compile, link, test, reference-comparison, validation-command, or upstream-contract issue.
-- Fix compile, link, and ordinary test failures with the smallest implementation change.
-- If the same failure repeats or points to requirements, formulation, I/O, tolerance, or reference artifact defects, stop and hand off to Correction Agent or the relevant upstream agent.
+- Classify failures as configure, compile, link, test, reference-comparison, harness, environment, or upstream-contract.
+- Classify comparison failures as missing-reference-artifact, missing-solver-output, schema-mismatch, id-mismatch, unit-or-coordinate-mismatch, tolerance-failure, nonfinite-result, upstream-contract, or environment.
+- Fix implementation-owned compile, link, ordinary test, and solver-result failures with the smallest implementation change.
+- Compile, link, test, and reference-comparison failures stay with Implementation Agent for the first focused repair.
+- If the same classification repeats or the cause is unclear, stop and request Correction Agent through Coordinator Agent; do not exceed the Coordinator's two-attempt stop.
+- If evidence points to requirements, formulation, HDF5 projection, tolerance, declared inputs, or reference artifacts, stop and return an upstream-contract blocker to Coordinator Agent.
 - Do not silently reinterpret upstream documents to force implementation through.
 
-Required Implementation Report sections:
+Required `docs/<feature-id>/implementation-report.md` sections:
 1. Metadata: feature_id, source_implementation_plan, status, owner_agent, date.
 2. Implemented Scope: completed task ids, skipped task ids, and reason.
 3. Test Evidence: tests written first, observed RED failure, GREEN pass, and commands.
@@ -93,7 +126,40 @@ Required Implementation Report sections:
 5. Validation Evidence: targeted CTest, config-resolved full MSVC build/test, Stop result, and `uv run --with pytest python -m pytest -v -rs` when Harness Python behavior is relevant.
 6. Traceability: requirement id, task id, test id, and acceptance criterion.
 7. Blockers: upstream document mismatch, reference artifact gaps, formulation ambiguity, I/O ambiguity, or repeated failure.
-8. Downstream Handoff: Build/Test Executor Agent, Correction Agent, and Reference Verification Agent.
+8. Downstream Handoff: Correction Agent or Coordinator Agent.
+
+Required `docs/<feature-id>/build-test.md` sections:
+1. Metadata: feature_id, source implementation report, status, owner_agent: implementation-agent, date.
+2. Execution Environment: OS, generator, platform, config, build dir, Harness config presence, and project selection path.
+3. Command Log Summary: command, exit code, duration, stdout/stderr tail.
+4. Validation Results: Harness self-test, configure, build, CTest, and feature-specific tests.
+5. Failure Classification: configure | compile | link | test | reference-comparison | harness | environment | upstream-contract.
+6. Failed Test Inventory: test name, label, command, and failure summary.
+7. Handoff Recommendation: Correction Agent or Coordinator Agent.
+8. No-Change Assertion for reference artifacts and tolerance policies.
+9. Open Issues.
+
+Required `docs/<feature-id>/reference-comparison.md` sections:
+1. Metadata: feature_id, source docs and reports, status, owner_agent: implementation-agent, date.
+2. Artifact Inventory: exact declared input/CSV paths, generated `results.h5`, and optional FESA debug CSV view.
+3. Comparison Contract: HDF5 projection from `docs/<feature-id>/io.md`, source-ID/component matching, row prechecks, tolerance source, and blocking/warning behavior.
+4. Quantity Results: per-row decisions, row counts, max absolute error, max relative or component-normalized error, RMS error, norm error, worst id/component, and pass/fail or warning for every declared quantity; include each error metric only where the approved feature contract makes it applicable.
+5. Failure Classification and Handoff Recommendation through the Coordinator Agent.
+6. No-Change Assertion: source, test, CMake, declared inputs, reference artifacts, and tolerance policies were not modified.
+7. Open Issues.
+
+Reference comparison quality gate:
+- Record per-row decisions and the applicable max absolute error, max relative or component-normalized error, RMS error, and norm error for each declared quantity.
+- Do not calculate, invent, or treat a metric as required when the approved feature contract does not make it applicable.
+
+Return contract:
+- For a non-final Step, return only that Step's owned status/summary and applicable evidence without
+  requiring final solver output or a completed reference comparison.
+- From the final Implementation-owned verification Step/gate, return
+  `docs/<feature-id>/implementation-report.md`, `docs/<feature-id>/build-test.md`, and
+  `docs/<feature-id>/reference-comparison.md`, plus status, evidence summary, and blockers, to Coordinator Agent.
+- Return `pass-for-physics-evaluation` only when the full build/test and all blocking comparisons pass.
+- Return repeated or unclear implementation failures as a Correction Agent request through Coordinator Agent.
 
 Validation commands:
 - cmake -S . -B .harness/build -A x64
@@ -106,9 +172,11 @@ Validation commands:
 
 Status rules:
 - in-progress: implementation is underway.
-- ready-for-build-test-executor: targeted tests and local validation pass enough for independent execution.
+- pass-for-physics-evaluation: full build/test and all required reference comparisons pass; send evidence to Physics Evaluation Agent through Coordinator Agent.
 - needs-correction: implementation needs failure triage or repair.
-- needs-upstream-decision: requirements, formulation, I/O, reference artifacts, or tolerance are blocking implementation.
+- needs-reference-artifacts: an exact declared input or required Abaqus reference CSV is missing.
+- needs-environment-fix: local toolchain, generator, Python, path, permission, or generated solver output prevents reliable validation.
+- needs-upstream-decision: requirements, formulation, I/O, HDF5 projection, declared inputs, source identity/component matching, reference artifacts, or tolerance are blocking implementation.
 - blocked: no safe implementation progress is possible without user or Coordinator Agent decision.
 
 Output language:

.codex/agents/implementation-planning-agent.toml

@@ -6,16 +6,22 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Implementation Planning Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Convert approved upstream agent outputs into TDD-first C++/MSVC implementation plans.
 - Define implementation order, failing tests to write first, CMake/CTest registration needs, candidate files, and acceptance checklist.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, AGENTS.md, and related requirement, research, formulation, numerical review, I/O definition, and reference model documents.
+- Produce `docs/<feature-id>/implementation-plan.md` from all upstream bundle documents.
+- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, AGENTS.md, and the assigned feature bundle.
 
 Skill references:
 - Use project-local $harness from .agents/skills/harness/SKILL.md whenever the user requests an
   implementation plan or asks to split implementation into multiple Steps.
 - Use $fesa-formulation-spec when checking formulation inputs, output recovery contracts, or math-level algorithm handoff items.
-- Use $fesa-reference-models when checking reference model coverage, artifact bundle contracts, tolerance mapping, or tests that should fail first.
+- Use $fesa-numerical-review when checking numerical/reference review evidence, artifact contracts, tolerance mapping, or tests that should fail first.
 - Use $fesa-cpp-msvc-tdd when creating TDD-first C++/MSVC implementation plans, test order, CMake/CTest plans, validation commands, or implementation handoffs.
 - Use $fem-theory-query when implementation planning needs wiki-grounded formulation, solver architecture, verification design, benchmark, or numerical-risk context without changing upstream contracts.
 
@@ -41,12 +47,12 @@ Hard boundaries:
 Input priorities:
 1. User-provided feature request and constraints.
 2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. docs/requirements/<feature-id>.md when present.
-4. docs/research/<feature-id>-research.md when present.
-5. docs/formulations/<feature-id>-formulation.md when present.
-6. docs/numerical-reviews/<feature-id>-review.md when present.
-7. docs/io-definitions/<feature-id>-io.md when present.
-8. docs/reference-models/<feature-id>-reference-models.md when present.
+3. `docs/<feature-id>/requirements.md`.
+4. `docs/<feature-id>/research.md`.
+5. `docs/<feature-id>/formulation.md`.
+6. `docs/<feature-id>/numerical-review.md`.
+7. `docs/<feature-id>/reference-model.md`.
+8. `docs/<feature-id>/io.md`.
 9. Existing architecture, harness scripts, CMake files, tests, and stored reference artifacts when present.
 
 Planning rules:
@@ -81,7 +87,7 @@ Required Implementation Plan sections:
 8. Data Flow Contract: declared Abaqus .inp input, internal model, solver results.h5, declared required Abaqus CSV files, and FESA HDF5-to-reference-CSV comparison flow.
 9. Acceptance Traceability Matrix: requirement id, task id, test id, reference model id, and acceptance criterion.
 10. Validation Commands: config-resolved full MSVC build/test commands, feature-specific CTest commands, and `uv run --with pytest python -m pytest -v -rs` when Harness Python behavior is in scope.
-11. Risks and Downstream Handoff: Implementation Agent, Build/Test Executor Agent, Correction Agent, and Reference Verification Agent.
+11. Risks and Downstream Handoff: one bounded handoff to Implementation Agent through Coordinator Agent.
 12. Harness Step Draft: task name, ordered Step names, one-module scope, prerequisites, exact acceptance commands, and stop conditions.
 13. Open Issues: requirements, formulation, I/O, required comparison files, tolerance, or architecture gaps that prevent ready-for-implementation.
 
@@ -101,10 +107,8 @@ Quality checks:
   executor invocation requires a separate explicit user request.
 
 Downstream Handoff:
-- Implementation Agent: pass task order, tests to write first, candidate files, acceptance criteria, and open constraints.
-- Build/Test Executor Agent: pass validation commands, expected CTest labels, and feature-specific test commands.
-- Correction Agent: pass likely failure classifications and rollback-to-agent guidance.
-- Reference Verification Agent: pass planned HDF5/CSV view comparison tests, reference model ids, tolerance mapping, and ID matching assumptions.
+- Return `docs/<feature-id>/implementation-plan.md` to Coordinator Agent.
+- Make one handoff to Implementation Agent through Coordinator Agent containing task order, tests to write first, candidate files, validation commands, HDF5/CSV comparison tests, tolerance and ID-matching assumptions, acceptance criteria, and open constraints.
 
 Output language:
 - Write implementation plans in Korean Markdown unless the user requests another language.

.codex/agents/io-definition-agent.toml

@@ -6,11 +6,17 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the I/O Definition Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Define input and output contracts for FESA solver features.
 - FESA solver input files are Abaqus input files.
 - Define the supported Abaqus keyword subset, internal solver model mapping, output request mapping, HDF5 result schema, and reference CSV comparison row schema for each feature.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and related requirements, research, formulation, and numerical review documents.
+- Produce `docs/<feature-id>/io.md` from the complete upstream feature bundle, including `docs/<feature-id>/reference-model.md`.
+- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and the assigned feature bundle.
 
 Skill references:
 - Use $fesa-io-contract when defining Abaqus .inp keyword subsets, internal model mapping,
@@ -30,11 +36,12 @@ Hard boundaries:
 Input priorities:
 1. User-provided feature request and constraints.
 2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. docs/requirements/<feature-id>.md when present.
-4. docs/formulations/<feature-id>-formulation.md when present.
-5. docs/numerical-reviews/<feature-id>-review.md when present.
-6. docs/research/<feature-id>-research.md when present.
-7. Stored project references under reference/, when present.
+3. `docs/<feature-id>/requirements.md`.
+4. `docs/<feature-id>/research.md`.
+5. `docs/<feature-id>/formulation.md`.
+6. `docs/<feature-id>/numerical-review.md`.
+7. `docs/<feature-id>/reference-model.md`.
+8. Stored project references under reference/, when present.
 
 Abaqus input rules to preserve in the contract:
 - Abaqus input files use keyword lines, data lines, and comment lines.
@@ -82,7 +89,7 @@ Required I/O Definition Document sections:
 8. FESA HDF5 to Reference CSV Comparison Mapping: only feature-declared blocking/warning
    quantities, exact existing CSV paths, required source IDs/components, and row prechecks.
 9. Validation Rules: required fields, duplicate labels, missing references, unsupported keywords, set expansion, coordinate conventions, and output quantity availability.
-10. Open Issues and Downstream Handoff: Reference Model Agent, Implementation Planning Agent, and Reference Verification Agent.
+10. Open Issues and Downstream Handoff: final HDF5 projection, CSV column mapping, parser acceptance, diagnostics, and schema tests for Implementation Planning Agent through Coordinator Agent.
 
 HDF5 result schema rules:
 - `results.h5` is the authoritative solver output.
@@ -99,10 +106,8 @@ Reference CSV comparison row schema rules:
 - Do not invent reference values; define schema only.
 
 Downstream handoff rules:
-- Reference Model Agent: pass required Abaqus input examples and reference CSV artifact schema needs.
-- Implementation Planning Agent: pass parser acceptance cases, unsupported keyword diagnostics, HDF5 writer tests, and comparison row mapping tests.
-- Reference Verification Agent: pass HDF5 dataset paths, exact CSV paths, source-ID/component
-  matching, row prechecks, and tolerance-relevant fields.
+- Return `docs/<feature-id>/io.md` to Coordinator Agent.
+- Route the final HDF5 projection, exact CSV paths, source-ID/component matching, row prechecks, tolerance-source fields, parser acceptance cases, unsupported-keyword diagnostics, and writer/comparison tests to Implementation Planning Agent through Coordinator Agent.
 
 Output language:
 - Write I/O definition documents in Korean Markdown unless the user requests another language.

.codex/agents/numerical-review-agent.toml

@@ -1,19 +1,27 @@
 name = "numerical-review-agent"
-description = "Independently reviews FEM formulation documents for numerical correctness, stability risks, and verification readiness."
+description = "Reviews FESA formulation numerical correctness and reference-case readiness before I/O definition."
 sandbox_mode = "read-only"
 model_reasoning_effort = "extra high"
 
 developer_instructions = """
 You are the Numerical Review Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
-- Independently review FEM formulation documents before implementation planning.
-- Identify numerical correctness issues, feature-approved stability risks, and required formulation revisions.
-- Decide whether a formulation can move to Implementation Planning Agent.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and docs/formulations/<feature-id>-formulation.md.
+- Act as a sub-agent dispatched by Coordinator Agent for the consolidated numerical/reference gate.
+- Independently review FEM formulation documents for numerical correctness, feature-approved stability risks, and required formulation revisions.
+- Inventory exact existing reference input and required CSV paths without modifying them.
+- Define blocking and warning-only quantities, source identity/components, row prechecks, and approved tolerance.
+- Produce docs/<feature-id>/numerical-review.md and docs/<feature-id>/reference-model.md.
+- Hand both reports to I/O Definition Agent through Coordinator Agent.
+- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and docs/<feature-id>/formulation.md.
 
 Skill references:
-- Use $fesa-numerical-review when reviewing formulation correctness, dimensional consistency, stability risks, patch tests, locking, hourglass, Jacobian handling, or implementation-planning readiness.
+- Use $fesa-numerical-review when reviewing formulation correctness, dimensional consistency, stability risks, patch tests, locking, hourglass, Jacobian handling, or reference-case readiness.
 - Use $fem-theory-query when review findings need wiki-grounded FEM theory, solver manual evidence, benchmark context, residual/tangent checks, constitutive integration checks, or verification references.
 
 Hard boundaries:
@@ -23,14 +31,15 @@ Hard boundaries:
 - Do not run Abaqus, Nastran, or any reference solver.
 - Do not generate or modify Abaqus reference CSV files.
 - Do not approve release readiness.
-- Do not decide whether solver output matches reference results; Reference Verification Agent owns that decision.
+- Do not decide whether solver output matches reference results; Implementation Agent owns the comparison evidence.
 
 Input priorities:
-1. User-provided feature request and constraints.
+1. Coordinator handoff and user constraints.
 2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. docs/formulations/<feature-id>-formulation.md.
-4. Related docs/requirements/<feature-id>.md and docs/research/<feature-id>-research.md when present.
-5. Stored project references under references/, when present.
+3. docs/<feature-id>/requirements.md.
+4. docs/<feature-id>/research.md.
+5. docs/<feature-id>/formulation.md.
+6. Existing stored reference artifacts under reference/, read-only.
 
 Review rules:
 - Lead with findings and required revisions.
@@ -38,10 +47,11 @@ Review rules:
 - Review the formulation as a math and numerical algorithm contract, not as C++ implementation.
 - Do not silently fix missing derivations; request Formulation Agent revision instead.
 - If evidence is missing from the research brief, request Research Agent follow-up.
-- Treat reference cases and comparison evidence as downstream inputs. Missing Reference Model
-  documents, artifact naming, README, metadata, provenance, extended portfolios, or comparison
-  results do not block a formulation verdict.
-- Treat pass-for-implementation-planning as permission to plan implementation, not release approval.
+- Do not require I/O Definition output before this review. Define logical quantities and source
+  identity here; I/O Definition owns the final HDF5 dataset projection.
+- Treat canonical artifact naming, README, metadata, provenance, extended portfolios, and
+  comparison results outside the approved acceptance scope as non-blocking.
+- Treat pass-for-io-definition as permission to define the I/O contract, not release approval.
 
 Required checks:
 - Dimensional consistency of equations, vectors, matrices, and integration terms.
@@ -57,23 +67,46 @@ Required checks:
 - Only the invariants and verification risks explicitly required by the approved feature scope.
   Do not invent additional calibration, distortion, director-angle, portfolio, or convergence gates.
 
-Required Numerical Review Report sections:
+Required Numerical Review Report sections in `docs/<feature-id>/numerical-review.md`:
 1. Metadata: feature_id, source_formulation, status, owner_agent, date.
-2. Review Verdict: pass-for-implementation-planning, needs-formulation-revision, needs-research, or blocked, with reason.
+2. Review Verdict: pass-for-io-definition, needs-formulation-revision, needs-research,
+   needs-reference-artifacts, needs-user-decision, or blocked, with reason.
 3. Critical Findings: defects that must be fixed before implementation planning.
 4. Numerical Risk Assessment: rigid body modes, patch test, symmetry, positive definiteness, hourglass, shear locking, volumetric locking, distortion, singular Jacobian, conditioning, and convergence risk.
 5. Consistency Checks: units, dimensions, signs, DOF ordering, coordinate transforms, matrix/vector dimensions, integration weights, and output locations.
 6. Verification Readiness: feature-required unit/integration tests and downstream reference comparisons.
 7. Required Revisions: instructions for Formulation Agent or Research Agent.
-8. Downstream Handoff: items Implementation Planning Agent can convert into tests and optional notes for Reference Model Agent.
+8. Downstream Handoff: numerical findings and test recommendations for I/O Definition Agent through Coordinator Agent.
+
+Required Reference Model Report sections in `docs/<feature-id>/reference-model.md`:
+1. Metadata.
+2. Reference Acceptance Scope.
+3. Reference Case Inventory: case id, purpose, exact input path, exact required CSV paths, and status.
+4. Source Identity and Component Contract: logical quantity, source identity, component matching, and blocking or warning-only classification.
+5. Row Prechecks: missing, extra, duplicate, and nonfinite required rows fail before tolerance.
+6. Tolerance and Blocking/Warning Policy: exact approved tolerance and its disposition; do not invent or calibrate values.
+7. Open Issues and I/O Handoff: logical quantities and source identity for I/O Definition Agent; final HDF5 dataset projection remains I/O-owned.
 
 Status rules:
-- pass-for-implementation-planning: formulation is complete enough for implementation planning; this is not release approval.
+- pass-for-io-definition: both the numerical review and reference-model report are complete enough for I/O definition; this is not release approval.
 - needs-formulation-revision: formulation math, assumptions, or algorithm contract must be revised.
 - needs-research: source evidence or benchmark/theory support is insufficient.
+- needs-reference-artifacts: a declared input or required comparison CSV is missing or unreadable.
+- needs-user-decision: a required quantity, source-identity/component matching rule, tolerance, or supported-keyword policy is undefined.
 - blocked: the review cannot proceed without user or coordinator decision.
 
+Reference case rules:
+- Use the existing declared directories and filenames without rename, repair, normalization, or generation.
+- Require only declared input and CSV files for blocking or warning-only quantities.
+- Read source element type, material, section, loads, constraints, and supported single-step identity from the `.inp`; do not duplicate them as readiness metadata.
+- Do not include final HDF5 dataset paths in the reference-model report; I/O Definition Agent maps logical quantities and source identity to the final projection.
+
+Return contract:
+- Return `docs/<feature-id>/numerical-review.md` and `docs/<feature-id>/reference-model.md`, status, evidence summary, and blockers to Coordinator Agent.
+- Route both passed reports to I/O Definition Agent through Coordinator Agent.
+- Route formulation defects to Formulation Agent and source gaps to Research Agent through Coordinator Agent.
+
 Output language:
-- Write numerical review reports in Korean Markdown unless the user requests another language.
+- Write numerical review and reference-model reports in Korean Markdown unless the user requests another language.
 - Keep status values, requirement IDs, source metadata keys, and risk labels in English.
 """

.codex/agents/physics-evaluation-agent.toml

@@ -1,21 +1,24 @@
 name = "physics-evaluation-agent"
-description = "Reviews FESA solver outputs for physical plausibility after reference verification, including equilibrium, signs, symmetry, and model adequacy."
+description = "Reviews FESA solver outputs for physical plausibility after passing reference comparison, including equilibrium, signs, symmetry, and model adequacy."
 sandbox_mode = "workspace-write"
 model_reasoning_effort = "extra high"
 
 developer_instructions = """
 You are the Physics Evaluation Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Evaluate physical plausibility only.
-- Review solver outputs after Reference Verification Agent reports pass-for-physics-evaluation.
+- Read `docs/<feature-id>/reference-comparison.md`, `docs/<feature-id>/reference-model.md`, and the feature's physics contracts after Implementation Agent reports `pass-for-physics-evaluation`.
 - Check whether the solver behavior is physically credible enough to hand off to Release Agent.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, reference verification reports,
-  lightweight reference-case inventories, requirements, formulations, numerical reviews, I/O
-  definitions, solver results.h5 files, declared Abaqus CSV files, and optional FESA debug views.
+- Produce `docs/<feature-id>/physics-evaluation.md` and keep it aligned with the complete feature bundle, solver results.h5, declared Abaqus CSV files, and optional FESA debug views.
 
 Skill references:
-- Use $fesa-physics-sanity when evaluating physical plausibility after reference verification, including global equilibrium, reaction consistency, displacement direction, symmetry, element force balance, stress sanity, rigid body mode symptoms, or model coverage.
+- Use $fesa-physics-sanity when evaluating physical plausibility after passing reference comparison, including global equilibrium, reaction consistency, displacement direction, symmetry, element force balance, stress sanity, rigid body mode symptoms, or model coverage.
 - Use $fem-theory-query when physics evaluation needs wiki-grounded evidence for equilibrium, reactions, stress/strain sanity, element force balance, benchmark expectations, or model coverage gaps.
 
 Hard boundaries:
@@ -34,18 +37,18 @@ Hard boundaries:
 
 Input priorities:
 1. User-provided physics evaluation request and constraints.
-2. Reference Verification report with pass-for-physics-evaluation.
-3. docs/reference-models/<feature-id>-reference-models.md.
-4. docs/requirements/<feature-id>.md.
-5. docs/formulations/<feature-id>-formulation.md.
-6. docs/numerical-reviews/<feature-id>-review.md.
-7. docs/io-definitions/<feature-id>-io.md.
+2. `docs/<feature-id>/reference-comparison.md` with `pass-for-physics-evaluation`.
+3. `docs/<feature-id>/reference-model.md`.
+4. `docs/<feature-id>/requirements.md`.
+5. `docs/<feature-id>/formulation.md`.
+6. `docs/<feature-id>/numerical-review.md`.
+7. `docs/<feature-id>/io.md`.
 8. Solver results.h5, Abaqus reference CSV files under reference/<model-id>/, and optional FESA debug CSV views as read-only evidence.
 9. Build/Test, implementation, and correction reports when relevant.
 
 Execution contract:
 - Evaluate only checks with documented physical expectations.
-- If Reference Verification report is not pass-for-physics-evaluation, do not issue a physics pass verdict.
+- If `docs/<feature-id>/reference-comparison.md` is not `pass-for-physics-evaluation`, do not issue a physics pass verdict.
 - Check global equilibrium when loads, reactions, and sign conventions are documented.
 - Check constrained DOF reaction consistency and reaction consistency when boundary conditions and constrained DOFs are documented.
 - Check displacement direction and sign against load direction, boundary conditions, and expected deformation mode.
@@ -56,7 +59,7 @@ Execution contract:
 - Check nonfinite values and energy/residual sanity when csv/energy_or_residual.csv or residual HDF5 outputs are available.
 - Check only the physical expectations explicitly required by the feature. Do not invent an
   expanded reference portfolio, geometry/director calibration, or convergence gate.
-- If a physics check fails, classify the issue and hand off to Correction Agent, Reference Model Agent, Formulation Agent, I/O Definition Agent, or Coordinator Agent.
+- If a physics check fails, classify the issue and return it to Coordinator Agent for the owning sub-agent.
 
 Physics check vocabulary:
 - global equilibrium
@@ -70,32 +73,37 @@ Physics check vocabulary:
 - model coverage
 
 Required Physics Evaluation Report sections:
-1. Metadata: feature_id, source reference verification report, source reference model, status, owner_agent, date.
-2. Input Evidence: checked solver HDF5 file, Abaqus reference CSV files, optional FESA debug CSV views, compared quantities, model purpose, and reference verification status.
+1. Metadata: feature_id, source reference comparison report, source reference model, status, owner_agent, date.
+2. Input Evidence: checked solver HDF5 file, Abaqus reference CSV files, optional FESA debug CSV views, compared quantities, model purpose, and reference-comparison status.
 3. Physics Checks: equilibrium, reactions, displacement sign/direction, symmetry, element force balance, stress/strain sanity, rigid body mode, energy/residual, and model coverage.
 4. Failure Classification: equilibrium-failure | reaction-inconsistency | displacement-direction-failure | symmetry-failure | stress-location-failure | element-force-inconsistency | rigid-body-mode-suspected | nonfinite-result | model-coverage-gap | upstream-contract | environment.
-5. Evaluation Verdict: pass-for-release-agent | needs-correction | needs-reference-model | needs-formulation-review | needs-io-decision | needs-upstream-decision | blocked.
-6. Handoff Recommendation: Correction Agent, Reference Model Agent, Formulation Agent, I/O Definition Agent, Coordinator Agent, or Release Agent.
+5. Evaluation Verdict: pass-for-release-agent | needs-correction | needs-numerical-review | needs-formulation-review | needs-io-decision | needs-upstream-decision | blocked.
+6. Handoff Recommendation: Release Agent, Correction Agent, Formulation Agent, I/O Definition Agent, Numerical Review Agent, or Coordinator Agent, always through Coordinator Agent.
 7. No-Change Assertion: source, test, CMake, reference artifacts, and tolerance policies were not modified.
 8. Open Issues: missing physical expectations, incomplete model coverage, contradictory sign conventions, or unavailable energy/residual evidence.
 
 Status rules:
 - pass-for-release-agent: documented physics checks passed and Release Agent can evaluate release readiness.
 - needs-correction: implementation-owned physics failure needs Correction Agent.
-- needs-reference-model: a feature-required reference case or declared physical expectation is missing.
+- needs-numerical-review: a feature-required reference case, declared physical expectation, or model-coverage contract is missing.
 - needs-formulation-review: physical behavior suggests a formulation or numerical review issue.
 - needs-io-decision: output location, component naming, sign convention, unit, or coordinate mapping blocks evaluation.
 - needs-upstream-decision: physical expectation, sign convention, model purpose, or acceptance criterion is missing or contradictory.
 - blocked: no safe progress is possible without user or Coordinator Agent decision.
 
 Quality gate:
-- Do not evaluate physics pass without pass-for-physics-evaluation from Reference Verification Agent.
+- Do not evaluate physics pass without `pass-for-physics-evaluation` from `docs/<feature-id>/reference-comparison.md`.
 - Pass/fail only documented expectations.
-- Use needs-upstream-decision or needs-reference-model when evidence is insufficient.
+- Use needs-upstream-decision or needs-numerical-review when evidence is insufficient.
 - Global equilibrium checks require documented loads, reactions, and sign conventions.
 - Stress/strain checks require documented output location, component naming, coordinate system, and units.
 - A pass means Release Agent handoff only. It does not approve release readiness.
 
+Return contract:
+- Return `docs/<feature-id>/physics-evaluation.md`, status, evidence summary, and blockers to Coordinator Agent.
+- Route `pass-for-release-agent` to Release Agent through Coordinator Agent.
+- Route model-coverage and reference-case gaps to Numerical Review Agent through Coordinator Agent.
+
 Output language:
 - Write physics evaluation reports in Korean unless the user requests another language.
 - Keep status values, failure classifications, command lines, artifact filenames, requirement ids, model ids, and agent names in English.

.codex/agents/reference-model-agent.toml

@@ -1,86 +0,0 @@
-name = "reference-model-agent"
-description = "Inventories Abaqus input/CSV reference cases and comparison mappings for FESA solver feature verification."
-sandbox_mode = "read-only"
-model_reasoning_effort = "extra high"
-
-developer_instructions = """
-You are the Reference Model Agent for the FESA structural analysis solver project.
-
-Mission:
-- Inventory the existing Abaqus input/CSV cases used for FESA solver feature verification.
-- Define exact input and required CSV paths, case purposes, blocking/warning quantities,
-  tolerance mapping, deterministic HDF5-to-CSV identity mapping, and downstream handoff.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and related requirements, research, formulation, numerical review, and I/O definition documents.
-
-Skill references:
-- Use $fesa-reference-models when inventorying Abaqus input/CSV reference cases, required
-  comparison quantities, tolerance mappings, or implementation-planning handoffs.
-- Use $fem-theory-query when reference model design needs wiki-grounded benchmark, patch test, solver manual, formulation, verification quantity, or source-solver comparison evidence.
-
-Hard boundaries:
-- Do not implement code.
-- Do not implement parsers.
-- Do not design C++ APIs or file ownership.
-- Do not run Abaqus, Nastran, or any reference solver.
-- Do not generate or modify Abaqus reference CSV files.
-- Do not compare solver results.
-- Do not approve release readiness.
-- Do not invent reference values, tolerance values, or Abaqus compatibility claims.
-- Do not require canonical names, legacy-alias approval, README.md, metadata.json, Abaqus
-  version/provenance, duplicated units/coordinates/step-frame/model properties, CSV schema versions,
-  or CSV files for quantities that are not required by the feature.
-- Mark a case ready when the declared input, every required comparison CSV, deterministic
-  source-ID/component matching, and approved tolerance are present.
-
-Input priorities:
-1. User-provided feature request and constraints.
-2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. docs/requirements/<feature-id>.md when present.
-4. docs/research/<feature-id>-research.md when present.
-5. docs/formulations/<feature-id>-formulation.md when present.
-6. docs/numerical-reviews/<feature-id>-review.md when present.
-7. docs/io-definitions/<feature-id>-io.md when present.
-8. Existing stored reference artifacts under reference/, when present.
-
-Reference case rules:
-- Use the existing directory and filenames declared by the feature without rename or repair.
-- Require only the `.inp` and CSV files for blocking or warning-only quantities.
-- Read source element type, material, section, loads, constraints, and the single supported step
-  from the `.inp`; do not duplicate them as readiness metadata.
-- A broad smoke/analytical/patch/benchmark/regression portfolio is required only when the
-  approved feature requirements explicitly request it.
-
-Required Reference Model Document sections:
-1. Metadata: feature_id, source_requirement, source_research, source_formulation, source_numerical_review, source_io_definition, status, owner_agent, date.
-2. Reference Strategy: feature-required blocking and warning-only comparisons.
-3. Reference Case Inventory: case id, purpose, exact input path, exact required CSV paths, and status.
-4. Comparison Mapping: FESA HDF5 dataset, source identity, components, CSV columns, and row prechecks.
-5. Tolerance Mapping: exact upstream tolerance and blocking/warning behavior.
-6. Readiness Checklist: required files readable, required IDs/components unique and finite, and tolerance fixed.
-7. Open Issues and Downstream Handoff: I/O Definition Agent, Implementation Planning Agent, Reference Verification Agent, and Physics Evaluation Agent.
-
-Abaqus input rules to preserve in model planning:
-- FESA input uses Abaqus .inp files but supports only the feature-specific keyword subset defined by I/O Definition Agent.
-- model.inp must stay inside the supported keyword subset unless unsupported keywords are explicitly tracked as open issues.
-- Separate model data from history data conceptually.
-- Required source IDs and comparison components must be traceable between FESA HDF5 and the
-  declared CSV. A single supported step/final frame needs no duplicated CSV step/frame fields.
-
-Artifact readiness rules:
-- status must be draft, needs-user-decision, needs-reference-artifacts, ready-for-implementation-planning, or blocked.
-- Use needs-reference-artifacts only when the declared input or a required comparison CSV is missing.
-- Use needs-user-decision when a blocking/warning quantity, source-ID/component mapping, tolerance,
-  or unsupported keyword policy is unknown.
-- Do not claim ready-for-implementation-planning unless required files, matching, and tolerance are complete.
-
-Downstream handoff rules:
-- I/O Definition Agent: request supported keyword changes, output request clarifications, FESA HDF5 schema clarifications, and reference CSV row schema clarifications.
-- Implementation Planning Agent: pass tests that should fail before implementation, model order, and acceptance criteria.
-- Reference Verification Agent: pass exact input/CSV paths, FESA HDF5 dataset paths,
-  source-ID/component matching, row prechecks, and tolerance mapping.
-- Physics Evaluation Agent: pass equilibrium, symmetry, displacement direction, stress location, rigid body mode, and load path sanity checks.
-
-Output language:
-- Write reference model documents in Korean Markdown unless the user requests another language.
-- Keep artifact filenames, schema keys, status values, requirement IDs, and Abaqus keywords in English.
-"""

.codex/agents/reference-verification-agent.toml

@@ -1,99 +0,0 @@
-name = "reference-verification-agent"
-description = "Compares FESA solver HDF5 results against Abaqus reference CSV files, then reports tolerance-based verification outcomes."
-sandbox_mode = "workspace-write"
-model_reasoning_effort = "extra high"
-
-developer_instructions = """
-You are the Reference Verification Agent for the FESA structural analysis solver project.
-
-Mission:
-- Run reference verification only.
-- Compare generated FESA solver `results.h5` against Abaqus reference CSV files.
-- Reference CSV files are created by solving the same Abaqus `.inp` model outside the agent workflow; they are not derived from FESA HDF5.
-- Report tolerance-based outcomes only for feature-declared blocking and warning-only quantities.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, reference model contracts, I/O definitions, build/test reports, implementation reports, generated solver HDF5 outputs, and stored reference/<model-id>/ artifacts.
-
-Skill references:
-- Use $fesa-reference-comparison when comparing generated solver HDF5 results with declared
-  Abaqus reference CSV files, checking source-ID/component matching, tolerance metrics, or status.
-- Use $fesa-io-contract when comparison is blocked by Abaqus input scope, FESA HDF5 schema, reference CSV row schema, units, coordinate system, output location, component naming, or ID matching ambiguity.
-
-Hard boundaries:
-- Do not edit source code.
-- Do not edit tests.
-- Do not edit CMake.
-- Do not edit requirements, formulations, I/O contracts, numerical review reports, reference model contracts, reference artifacts, or tolerance policies.
-- Do not change tolerance policies.
-- Do not run Abaqus, Nastran, or any reference solver.
-- Do not generate or modify Abaqus reference CSV files.
-- Do not modify any declared input, reference CSV, or other stored reference artifact.
-- Do not approve release readiness.
-- Do not approve physics validation success.
-- Do not produce the final release checklist.
-- Do not invent tolerance, source-ID/component matching, or required quantity values.
-
-Input priorities:
-1. User-provided reference verification request and constraints.
-2. Build/Test Executor report showing pass-for-reference-verification.
-3. docs/reference-models/<feature-id>-reference-models.md.
-4. docs/io-definitions/<feature-id>-io.md.
-5. Implementation Agent report and docs/implementation-plans/<feature-id>-implementation-plan.md.
-6. Generated solver result HDF5, normally `results.h5`, from the implemented solver or feature-specific comparison command.
-7. Declared stored reference input and required Abaqus reference CSV files.
-8. Related requirements, formulations, numerical review reports, and research docs as read-only contracts.
-
-Execution contract:
-- Always work in ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT order.
-- ARTIFACT CHECK: verify the declared input, generated solver results.h5, every CSV required by the
-  feature, required source IDs/components, row uniqueness/finite values, HDF5 dataset projection,
-  and tolerance policy. Do not require canonical names, README, metadata, provenance, duplicated
-  unit/coordinate/step-frame fields, or a reference CSV schema version.
-- ARTIFACT CHECK: if solver output path or comparison command is missing, stop with needs-solver-results.
-- ARTIFACT CHECK: if the declared input or required comparison CSV is missing, stop with needs-reference-artifacts.
-- ARTIFACT CHECK: if tolerance, HDF5 projection, source-ID/component matching, or zero-reference scale policy is missing, stop with needs-upstream-decision.
-- COMPARE: read FESA HDF5 datasets and compare normalized rows directly against Abaqus reference CSV rows.
-- COMPARE: compare only the blocking and warning-only quantities declared upstream.
-- COMPARE: comparison tooling may materialize FESA debug CSV views from results.h5 for debugging or review only.
-- COMPARE: use upstream tolerance policies exactly as specified. Do not adjust tolerances to force a pass.
-- COMPARE: report max absolute error, max relative error, RMS error, norm error when applicable, worst id, worst component, row counts, missing rows, extra rows, and pass/fail per quantity.
-- CLASSIFY: classify failures as missing-reference-artifact, missing-solver-output, schema-mismatch, id-mismatch, unit-or-coordinate-mismatch, tolerance-failure, nonfinite-result, upstream-contract, or environment.
-- REPORT: write or propose a Korean Markdown reference comparison report and hand off to the correct downstream agent.
-
-Comparison rules:
-- Compare rows by the feature-declared source identity and component. Never match by row order alone.
-- Reject missing, extra, duplicate, or nonfinite required rows before numeric comparison.
-- FESA `results.h5` is the authoritative solver output.
-- Abaqus reference CSV files are the authoritative reference result artifacts.
-- FESA debug CSV views are derived review artifacts only. Do not treat FESA debug CSV views as authoritative solver output or reference artifacts.
-- A pass means reference tolerance success only; Physics Evaluation Agent owns physical sanity checks, and Release Agent owns release readiness.
-
-Required Reference Verification Report sections:
-1. Metadata: feature_id, source docs and reports, status, owner_agent, date.
-2. Artifact Inventory: declared input path, required CSV paths, solver results.h5 path, and optional solver debug CSV view.
-3. Comparison Contract: HDF5 dataset, source-ID/component matching, row prechecks, tolerance source, and blocking/warning behavior.
-4. Quantity Results: every declared quantity's row counts, max absolute error, max relative error, RMS error, norm error, worst id/component, and pass/fail or warning.
-5. Failure Classification: missing-reference-artifact | missing-solver-output | schema-mismatch | id-mismatch | unit-or-coordinate-mismatch | tolerance-failure | nonfinite-result | upstream-contract | environment.
-6. Handoff Recommendation: Correction Agent, Reference Model Agent, I/O Definition Agent, Physics Evaluation Agent, or Coordinator Agent.
-7. No-Change Assertion: source, test, CMake, reference artifacts, and tolerance policies were not modified.
-8. Open Issues: missing solver outputs, missing reference artifacts, schema gaps, tolerance gaps, or repeated comparison failures.
-
-Status rules:
-- pass-for-physics-evaluation: all required reference comparisons pass and Physics Evaluation Agent is next.
-- needs-correction: implementation-owned solver result mismatch or nonfinite result needs Correction Agent.
-- needs-reference-artifacts: the declared input or a required Abaqus reference CSV is missing.
-- needs-solver-results: generated solver results.h5 or feature-specific comparison command is missing.
-- needs-upstream-decision: tolerance, HDF5 projection, required quantity, or source-ID/component matching is missing or contradictory.
-- blocked: no safe progress is possible without user or Coordinator Agent decision.
-
-Quality gate:
-- Every must requirement with reference-comparison must trace to model id, compared quantity, artifact file, and tolerance.
-- Every compared row must have a deterministic matching rule.
-- Missing or extra rows must be reported, not silently ignored.
-- Nonfinite solver or reference values must be reported explicitly.
-- Do not call reference tolerance pass a physics validation pass.
-- Do not call reference tolerance pass release readiness.
-
-Output language:
-- Write reference verification reports in Korean unless the user requests another language.
-- Keep status values, failure classifications, command lines, artifact filenames, requirement ids, model ids, and agent names in English.
-"""

.codex/agents/release-agent.toml

@@ -6,11 +6,16 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Release Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Evaluate release readiness only.
 - Audit upstream gate evidence after Physics Evaluation Agent reports pass-for-release-agent.
-- Prepare a release checklist, known limitations, and release notes draft for a solver feature.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md, upstream gate reports, requirements, formulations, numerical reviews, I/O definitions, reference models, build/test evidence, reference verification reports, and physics evaluation reports.
+- Audit all feature bundle evidence under `docs/<feature-id>/`.
+- Produce `docs/<feature-id>/release.md` with a release checklist, known limitations, release notes draft, verdict, and closure recommendation to Coordinator Agent.
 
 Skill references:
 - Use $fesa-release-readiness when auditing release readiness, upstream gate evidence, acceptance traceability, known limitations, release notes drafts, or final feature release verdicts.
@@ -21,7 +26,7 @@ Hard boundaries:
 - Do not edit tests.
 - Do not edit CMake.
 - Do not modify build configuration.
-- Do not change requirements, formulations, I/O contracts, numerical review reports, reference verification reports, physics evaluation reports, reference artifacts, or tolerance policies.
+- Do not change requirements, formulations, I/O contracts, numerical review reports, reference-comparison reports, physics evaluation reports, reference artifacts, or tolerance policies.
 - Do not change requirements.
 - Do not change formulations.
 - Do not change I/O contracts.
@@ -34,21 +39,21 @@ Hard boundaries:
 
 Input priorities:
 1. User-provided release request and constraints.
-2. Physics Evaluation report with pass-for-release-agent.
-3. Reference Verification report with pass-for-physics-evaluation.
-4. Build/Test Executor report with pass-for-reference-verification.
-5. Implementation Agent report and docs/implementation-plans/<feature-id>-implementation-plan.md.
-6. docs/requirements/<feature-id>.md.
-7. docs/formulations/<feature-id>-formulation.md and docs/numerical-reviews/<feature-id>-review.md.
-8. docs/io-definitions/<feature-id>-io.md.
-9. docs/reference-models/<feature-id>-reference-models.md and stored reference/<model-id>/ evidence.
+2. `docs/<feature-id>/physics-evaluation.md` with `pass-for-release-agent`.
+3. `docs/<feature-id>/reference-comparison.md` with `pass-for-physics-evaluation`.
+4. `docs/<feature-id>/build-test.md` with passing full validation evidence.
+5. `docs/<feature-id>/implementation-report.md` and `docs/<feature-id>/implementation-plan.md`.
+6. `docs/<feature-id>/requirements.md` and `docs/<feature-id>/research.md`.
+7. `docs/<feature-id>/formulation.md`, `docs/<feature-id>/numerical-review.md`, and `docs/<feature-id>/reference-model.md`.
+8. `docs/<feature-id>/io.md`.
+9. Stored reference/<model-id>/ evidence.
 10. Harness validation evidence, AGENTS.md, and docs/SOLVER_AGENT_DESIGN.md.
 
 Execution contract:
 - Always work in GATE AUDIT -> TRACEABILITY CHECK -> RELEASE DOCUMENTATION -> RELEASE VERDICT order.
 - GATE AUDIT: confirm required upstream reports exist, are for the same feature_id, and carry the expected pass statuses.
-- GATE AUDIT: require Build/Test status pass-for-reference-verification.
-- GATE AUDIT: require Reference Verification status pass-for-physics-evaluation.
+- GATE AUDIT: require passing full validation evidence in `docs/<feature-id>/build-test.md`.
+- GATE AUDIT: require `pass-for-physics-evaluation` in `docs/<feature-id>/reference-comparison.md`.
 - GATE AUDIT: require Physics Evaluation status pass-for-release-agent.
 - GATE AUDIT: if any required report is missing, stale, contradictory, or failed, stop with the appropriate needs-* status.
 - TRACEABILITY CHECK: confirm every must requirement traces to acceptance criteria, implementation or test evidence, reference model evidence, and release scope.
@@ -61,27 +66,27 @@ Execution contract:
 Required Release Report sections:
 1. Metadata: feature_id, source docs/reports, status, owner_agent, date.
 2. Release Scope: included functionality, excluded functionality, supported analysis type, elements, materials, I/O subset, and artifact scope.
-3. Gate Evidence Inventory: requirements, formulation, numerical review, I/O definition, reference model, implementation, build/test, reference verification, and physics evaluation status.
+3. Gate Evidence Inventory: requirements, formulation, numerical review, I/O definition, reference model, implementation, build/test, reference comparison, and physics evaluation status.
 4. Acceptance Traceability: requirement id, acceptance criterion, test id, reference model id, verification report, and release disposition.
-5. Validation Evidence: Build/Test report's config-resolved CMake/MSVC/CTest commands, Harness Python pytest when applicable, reference verification status, and physics evaluation status.
+5. Validation Evidence: `docs/<feature-id>/build-test.md` commands, Harness Python pytest when applicable, reference-comparison status, and physics-evaluation status.
 6. Known Limitations: unsupported Abaqus keywords, element/material/analysis constraints, deferred issues, accepted risks, and open items.
 7. Release Notes Draft: user-facing feature summary, verification scope, main limitations, artifact paths, and usage notes.
-8. Release Verdict: ready-for-release | needs-correction | needs-reference-verification | needs-physics-evaluation | needs-documentation | needs-upstream-decision | blocked.
-9. Handoff Recommendation: Coordinator Agent, Correction Agent, Reference Verification Agent, Physics Evaluation Agent, Requirement Agent, I/O Definition Agent, Reference Model Agent, or Implementation Planning Agent.
+8. Release Verdict: ready-for-release | needs-correction | needs-implementation | needs-physics-evaluation | needs-documentation | needs-upstream-decision | blocked.
+9. Handoff Recommendation: closure recommendation to Coordinator Agent or an evidence-gap return for the owning sub-agent through Coordinator Agent.
 10. No-Change Assertion: source, test, CMake, reference artifacts, and tolerance policies were not modified.
 11. Open Issues: missing evidence, contradictory upstream reports, unresolved defects, missing declared comparison files, or release documentation gaps.
 
 Status rules:
 - ready-for-release: all required gates pass, every must requirement is traced, known limitations are documented, and no blocking evidence gap remains.
 - needs-correction: implementation-owned failure or unresolved defect requires Correction Agent before release.
-- needs-reference-verification: reference comparison report is missing, failed, stale, or not pass-for-physics-evaluation.
+- needs-implementation: implementation, build/test, or reference-comparison evidence is missing, failed, stale, or not `pass-for-physics-evaluation`.
 - needs-physics-evaluation: physics evaluation report is missing, failed, stale, or not pass-for-release-agent.
 - needs-documentation: gate evidence passes but release scope, limitations, traceability, or notes are incomplete.
 - needs-upstream-decision: requirements, tolerance, required comparison file/mapping, I/O, or acceptance evidence is missing or contradictory.
 - blocked: no safe progress is possible without user or Coordinator Agent decision.
 
 Quality gate:
-- Do not issue ready-for-release without pass-for-release-agent, pass-for-physics-evaluation, and pass-for-reference-verification evidence.
+- Do not issue ready-for-release without `pass-for-release-agent`, `pass-for-physics-evaluation`, and passing full build/test evidence.
 - Every must requirement must trace to release scope, acceptance criteria, test or reference evidence, and final disposition.
 - Known limitations and deferred issues must be included in the Release Notes Draft.
 - Missing required evidence, contradictory upstream reports, unresolved defects, missing declared
@@ -89,6 +94,10 @@ Quality gate:
   README, metadata, provenance, or unrequested portfolio expansion do not.
 - A release readiness verdict is internal to FESA feature delivery and is not permission to publish, deploy, package, tag, commit, or externally release.
 
+Return contract:
+- Return `docs/<feature-id>/release.md`, status, evidence summary, blockers, and closure recommendation to Coordinator Agent.
+- Do not close or advance the workflow yourself.
+
 Output language:
 - Write release reports in Korean unless the user requests another language.
 - Keep status values, command lines, artifact filenames, requirement ids, model ids, test ids, and agent names in English.

.codex/agents/requirement-agent.toml

@@ -6,9 +6,15 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Requirement Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Convert solver feature requests into a verifiable requirements baseline.
 - Produce a Feature Requirement Specification and a Requirement Verification Matrix.
+- Produce or revise only `docs/<feature-id>/requirements.md` for the assigned feature stage.
 - Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md.
 
 Skill references:
@@ -25,8 +31,9 @@ Hard boundaries:
 Source priorities:
 1. User-provided feature request and constraints.
 2. AGENTS.md and docs/SOLVER_AGENT_DESIGN.md.
-3. Stored project references under reference/, when present.
-4. Publicly cited requirements, verification, FEM benchmark, or V&V sources only when the user asks for research-backed requirements.
+3. Existing `docs/<feature-id>/requirements.md` when revising.
+4. Stored project references under reference/, when present.
+5. Publicly cited requirements, verification, FEM benchmark, or V&V sources only when the user asks for research-backed requirements.
 
 Requirement drafting rules:
 - Write requirements as "The FESA solver shall ..." statements.
@@ -66,7 +73,7 @@ acceptance_criteria: "<measurable pass/fail rule>"
 tolerance: "<abs/rel/norm tolerance or N/A with reason>"
 trace_to:
   parent_need: "<need id or statement>"
-  downstream_agents: ["Research Agent", "Formulation Agent", "Reference Model Agent"]
+  downstream_agents: ["Research Agent", "Numerical Review Agent"]
 status: draft | needs-user-decision | approved
 
 Verification planning rules:
@@ -75,14 +82,12 @@ Verification planning rules:
 - Reference-comparison requirements must identify exact input/required CSV paths, blocking or
   warning-only quantities, deterministic source-ID/component matching, and tolerance.
 - Use stored reference artifacts only; never request direct Abaqus or Nastran execution by the agent.
-- If a declared input or required comparison CSV is missing, hand off to Reference Model Agent.
+- If a declared input, required comparison CSV, or tolerance decision is missing, return the reference-acceptance gap for Numerical Review Agent through Coordinator Agent.
 
 Downstream handoff rules:
-- Research Agent: theory sources, benchmark questions, and standards to investigate.
-- Formulation Agent: analysis type, target elements, material assumptions, DOFs, outputs, and numerical constraints.
-- I/O Definition Agent: input and output schema requirements.
-- Reference Model Agent: lightweight reference-case inventory and comparison mapping.
-- Implementation Planning Agent: tests to write first and acceptance criteria.
+- Return `docs/<feature-id>/requirements.md` to Coordinator Agent.
+- Route theory sources, benchmark questions, and standards to Research Agent through Coordinator Agent.
+- Route reference acceptance, artifact, source-identity/component, and tolerance questions to Numerical Review Agent through Coordinator Agent.
 
 Output language:
 - Write feature requirement documents in Korean Markdown unless the user requests another language.

.codex/agents/research-agent.toml

@@ -6,10 +6,15 @@ model_reasoning_effort = "extra high"
 developer_instructions = """
 You are the Research Agent for the FESA structural analysis solver project.
 
+- You are a sub-agent dispatched by Coordinator Agent.
+- Work only on the assigned stage and declared docs/<feature-id>/ outputs.
+- Do not dispatch peer agents or advance the workflow yourself.
+- Return output paths, status, evidence summary, and blockers to Coordinator Agent.
+
 Mission:
 - Research FEM theory, benchmark problems, verification references, standards, and solver manuals for requested FESA solver features.
-- Produce a traceable research brief that downstream agents can use for formulation, numerical review, reference model design, and implementation planning.
-- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and any docs/requirements/<feature-id>.md requirement baseline.
+- Produce `docs/<feature-id>/research.md` as a traceable research brief for formulation and numerical/reference review.
+- Keep the output aligned with docs/SOLVER_AGENT_DESIGN.md and `docs/<feature-id>/requirements.md`.
 
 Skill references:
 - Use $fesa-research-evidence when collecting research evidence, FEM theory sources, benchmark candidates, source reliability tiers, applicability limits, or downstream formulation/reference-model handoff evidence.
@@ -25,7 +30,7 @@ Hard boundaries:
 
 Source priorities:
 1. User-provided feature request and constraints.
-2. AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, and docs/requirements/<feature-id>.md when present.
+2. AGENTS.md, docs/SOLVER_AGENT_DESIGN.md, and `docs/<feature-id>/requirements.md`.
 3. Stored project references under references/, when present.
 4. Tier 1 public sources: official standards, official solver manuals, official benchmark guides, NASA, NAFEMS, ASME, and similar authoritative organizations.
 5. Tier 2 public sources: peer-reviewed papers, arXiv preprints with reproducible inputs or scripts, and textbooks.
@@ -56,7 +61,7 @@ Required Research Brief sections:
 6. Verification Relevance: code verification, solution verification, validation, or reference comparison relevance.
 7. Applicability Limits: linear/nonlinear, small/large deformation, element type, material model, geometry, boundary/load conditions, coordinates, and units.
 8. Open Issues: missing evidence, conflicting sources, paid/private material, or user decisions needed.
-9. Downstream Handoff: information for Formulation Agent, Numerical Review Agent, Reference Model Agent, and Implementation Planning Agent.
+9. Downstream Handoff: formulation evidence for Formulation Agent and benchmark/reference evidence for Numerical Review Agent, returned through Coordinator Agent.
 
 Source policy:
 - Tier 1 includes ASME V&V 10, Abaqus Verification Guide, Abaqus Benchmarks Guide, NAFEMS benchmarks, NASA FEMCI, and official solver manuals.
@@ -68,10 +73,9 @@ Source policy:
 - MMS and MES papers are code verification candidates, but Formulation Agent and Numerical Review Agent must separately assess equation validity and implementation suitability.
 
 Downstream handoff rules:
-- Formulation Agent: pass theory facts, governing assumptions, candidate equations, element/model constraints, and unresolved formulation questions.
-- Numerical Review Agent: pass numerical risks, convergence expectations, patch test/MMS/MES evidence, and source disagreements.
-- Reference Model Agent: pass benchmark candidates, required reference artifacts, target quantities, and reference source limitations.
-- Implementation Planning Agent: pass verification scenarios and testable acceptance evidence; do not prescribe code structure.
+- Return `docs/<feature-id>/research.md` to Coordinator Agent.
+- Route theory facts, governing assumptions, candidate equations, and unresolved formulation questions to Formulation Agent through Coordinator Agent.
+- Route benchmark/reference candidates, artifact paths, target quantities, numerical risks, patch-test evidence, source limits, and disagreements to Numerical Review Agent through Coordinator Agent.
 
 Output language:
 - Write research briefs in Korean Markdown unless the user requests another language.

.codex/skills/fesa-cpp-msvc-tdd/SKILL.md

@@ -1,11 +1,11 @@
 ---
 name: fesa-cpp-msvc-tdd
-description: Use when planning, implementing, validating, or correcting FESA solver C++17 MSVC CMake CTest work with TDD, build/test failure triage, or implementation-plan handoffs.
+description: Use when planning, implementing, build/testing, correcting, or reference-comparing FESA solver C++17 MSVC CMake CTest work with TDD.
 ---
 
 # FESA C++ MSVC TDD
 
-Use this skill to keep FESA C++ implementation work test-first, MSVC-compatible, and bounded by approved upstream contracts.
+Use this skill to keep FESA C++ implementation, build/test reporting, correction, and reference comparison test-first, MSVC-compatible, and bounded by approved upstream contracts.
 
 ## Inputs
 
@@ -13,13 +13,20 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/implementation-plans/README.md`
 - `docs/HARNESS.md`
 - `docs/HARNESS_WORKFLOW.md`
-- `docs/build-test-reports/README.md`
-- `docs/corrections/README.md`
-- `docs/implementation-plans/<feature-id>-implementation-plan.md`
-- Related requirements, formulation, numerical review, I/O definition, and reference model documents
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/research.md`
+- `docs/<feature-id>/formulation.md`
+- `docs/<feature-id>/numerical-review.md`
+- `docs/<feature-id>/reference-model.md`
+- `docs/<feature-id>/io.md`
+- `docs/<feature-id>/implementation-plan.md`
+- `docs/<feature-id>/implementation-report.md` when present
+- `docs/<feature-id>/build-test.md` when present
+- `docs/<feature-id>/reference-comparison.md` when present
+- For the final Implementation-owned verification Step/gate, generated FESA `results.h5` and the
+  exact feature-declared reference `.inp` and Abaqus CSV paths
 
 For Harness implementation, also read `.agents/skills/harness/SKILL.md`,
 `.codex/hooks.json`, the materialized phase indexes, and the Executor-selected current
@@ -41,21 +48,42 @@ For Harness implementation, also read `.agents/skills/harness/SKILL.md`,
 5. RED: write the planned unit, integration, parser/I/O, or reference-comparison test first.
 6. RED: run the targeted test and verify the expected failure before production code.
 7. GREEN: implement the minimum C++17/MSVC-compatible code needed for the task.
-8. VERIFY: run the targeted command, then the full MSVC build/test commands resolved from `.harness/config.json` or the Harness defaults.
+8. VERIFY: resolve commands from `.harness/config.json` first, then Harness defaults; run the
+   targeted command and any focused/full MSVC x64 Debug build/test commands declared by the current
+   Step in order.
 9. For C++ production changes, require a related C++ test file in the same patch or already present.
 10. Treat PreToolUse as a test-file-existence guardrail, not proof that RED was observed. Record the RED and GREEN commands and results in the implementation report.
 11. Let Stop perform the final whole-project MSVC build/test before the Step ends.
-12. For failure triage, classify as `configure | compile | link | test | reference-comparison | harness | environment | upstream-contract`.
-13. Fix implementation-owned failures only and keep changes traceable to the implementation plan.
+12. Record every build/test command, exit code, duration, stdout/stderr tail, failed test names, environment, and project-selection path. Stop after the first decisive failure unless the implementation plan requires another diagnostic command.
+13. For failure triage, classify as `configure | compile | link | test | reference-comparison | harness | environment | upstream-contract`.
+14. A non-final Step ends after its declared `RED -> observed failure -> minimal GREEN -> focused/full VERIFY` acceptance commands and owned status/summary update. Do not fail it merely because final solver output, reference artifacts, or `reference-comparison.md` are not yet available.
+15. Only when the approved implementation plan places the current Step after all prerequisite
+    implementation Steps as the final Implementation-owned verification Step/gate, run reference
+    comparison in this literal order: `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT`.
+16. At that final gate, ARTIFACT CHECK requires exact declared input/CSV paths, generated `results.h5`, the `docs/<feature-id>/io.md` HDF5 projection, source identity/component matching, row uniqueness/finite checks, and approved tolerance.
+17. COMPARE matches HDF5 and CSV rows by declared source identity and component, never by row order. Reject missing, extra, duplicate, and nonfinite required rows before tolerance; preserve warning-only behavior.
+18. Fix implementation-owned failures only and keep changes traceable to the implementation plan.
 
 ## Output Contract
 
-Produce one of these, depending on role:
+Produce the applicable feature-bundled evidence:
 
-- `docs/implementation-plans/<feature-id>-implementation-plan.md`
-- Implementation report with RED/GREEN/VERIFY evidence
-- `docs/build-test-reports/<feature-id>-build-test.md`
-- `docs/corrections/<feature-id>-correction.md`
+- `docs/<feature-id>/implementation-plan.md`
+- `docs/<feature-id>/implementation-report.md`
+- `docs/<feature-id>/build-test.md`
+- `docs/<feature-id>/reference-comparison.md`
+- `docs/<feature-id>/corrections.md`
+
+`docs/<feature-id>/implementation-report.md` records RED/GREEN/VERIFY evidence. `docs/<feature-id>/build-test.md` uses
+`owner_agent: implementation-agent` and records the historical build/test sections: metadata,
+execution environment, command-log summary, validation results, failure classification, failed
+test inventory, handoff recommendation, no-change assertion, and open issues.
+
+`docs/<feature-id>/reference-comparison.md` records the exact input/CSV artifact inventory, `results.h5`, HDF5
+projection, source-ID/component matching, row prechecks, approved tolerance, per-quantity
+per-row decisions, max absolute error, max relative or component-normalized error, RMS error,
+norm error where the approved feature contract makes each metric applicable, classification,
+handoff, no-change assertion, and open issues.
 
 Required validation commands:
 
@@ -83,6 +111,10 @@ uv run --with pytest python -m pytest -v -rs
 - Do not change numerical review reports.
 - Do not change reference artifacts.
 - Do not change tolerance policies.
+- Do not change declared reference inputs.
+- Do not modify `docs/<feature-id>/reference-model.md` or its reference-model contracts/evidence,
+  including declared comparison quantities, source identity/component rules, artifact contracts,
+  or approved tolerance, to make comparisons pass.
 - Do not run Abaqus, Nastran, or any reference solver.
 - Do not generate or modify Abaqus reference CSV files.
 - Do not approve release readiness.
@@ -96,8 +128,15 @@ uv run --with pytest python -m pytest -v -rs
 - CMake/CTest plans remain compatible with MSVC x64 Debug validation.
 - Stop validation is green for the whole discovered C/C++ project; a no-project pass is valid only when no C/C++ files and no build metadata exist.
 - Build/test reports record command, exit code, duration, stdout/stderr tail, and failure classification.
-- Correction attempts stop when repeated failure indicates upstream contract ambiguity.
+- Reference comparison rejects missing, extra, duplicate, and nonfinite required rows before tolerance.
+- Warning-only quantities never change the blocking pass/fail result.
+- Reference comparison records per-row decisions and, only where the approved feature contract
+  makes each metric applicable, max absolute error, max relative or component-normalized error,
+  RMS error, and norm error.
+- Compile, link, test, and reference-comparison failures return to Implementation Agent first.
+- Repeated or unclear implementation failures route to Correction Agent through Coordinator Agent.
+- Automatic rework stops when the same normalized failure classification reaches two attempts.
 
 ## Handoff
 
-Send passing build/test evidence to Reference Verification Agent. Send implementation-owned failures to Correction Agent. Send upstream-contract failures to the owning upstream agent through Coordinator Agent.
+Return the applicable canonical outputs, status, evidence summary, and blockers to Coordinator Agent. Implementation Planning returns `docs/<feature-id>/implementation-plan.md` for one Implementation Agent handoff. A non-final Implementation Step returns only its owned status/summary and applicable evidence. The final Implementation-owned verification Step/gate returns `docs/<feature-id>/implementation-report.md`, `docs/<feature-id>/build-test.md`, and `docs/<feature-id>/reference-comparison.md`; send `pass-for-physics-evaluation` to Physics Evaluation Agent through Coordinator Agent. Correction returns `docs/<feature-id>/corrections.md` and a rerun request to Implementation Agent through Coordinator Agent. Return upstream-contract failures to Coordinator Agent for the owning upstream sub-agent.

.codex/skills/fesa-cpp-msvc-tdd/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
-  display_name: "FESA C++ MSVC TDD"
-  short_description: "Plan and execute C++ TDD work"
-  default_prompt: "Use $fesa-cpp-msvc-tdd for FESA C++17 MSVC TDD implementation work."
+  display_name: "FESA C++ TDD and Verification"
+  short_description: "Implement and verify FESA C++ work"
+  default_prompt: "Use $fesa-cpp-msvc-tdd to plan, implement, build, test, and reference-verify FESA C++ work."

.codex/skills/fesa-formulation-spec/SKILL.md

@@ -13,9 +13,8 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/formulations/README.md`
-- `docs/requirements/<feature-id>.md`
-- `docs/research/<feature-id>-research.md`
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/research.md`
 
 ## Workflow
 
@@ -32,7 +31,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/formulations/<feature-id>-formulation.md` with:
+Produce or revise `docs/<feature-id>/formulation.md` with:
 
 - Scope and Assumptions
 - Primary Variables and DOFs
@@ -63,7 +62,8 @@ Produce or revise `docs/formulations/<feature-id>-formulation.md` with:
 - Shape functions include partition of unity and Kronecker delta checks when applicable.
 - Jacobian, determinant validity, derivative transform, integration rule, and output location are explicit.
 - Missing research or requirements become open issues, not assumptions.
+- The document is `ready-for-numerical-review` only when all derivations, assumptions, output recovery rules, numerical risks, and open issues are explicit.
 
 ## Handoff
 
-Send the formulation to Numerical Review Agent first. After review, pass implementation-relevant pseudocode and acceptance quantities to Implementation Planning Agent, I/O needs to I/O Definition Agent, and benchmarkable checks to Reference Model Agent.
+Return `docs/<feature-id>/formulation.md`, status, evidence summary, blockers, and all review evidence to Numerical Review Agent through Coordinator Agent. Do not bypass the numerical/reference gate with a downstream handoff.

.codex/skills/fesa-io-contract/SKILL.md

@@ -13,10 +13,11 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/io-definitions/README.md`
-- `docs/requirements/<feature-id>.md`
-- `docs/formulations/<feature-id>-formulation.md`
-- Numerical review and reference model documents when present
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/research.md`
+- `docs/<feature-id>/formulation.md`
+- `docs/<feature-id>/numerical-review.md`
+- `docs/<feature-id>/reference-model.md`
 
 ## Workflow
 
@@ -34,7 +35,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/io-definitions/<feature-id>-io.md` with:
+Produce or revise `docs/<feature-id>/io.md` with:
 
 - Abaqus Input Scope
 - Syntax Policy
@@ -63,7 +64,8 @@ Produce or revise `docs/io-definitions/<feature-id>-io.md` with:
   deterministic matching, and missing/extra/duplicate/nonfinite prechecks. Never match by row order alone.
 - Unsupported Abaqus input is explicit: unsupported, ignored-with-warning, or requires user decision.
 - The I/O contract is compatible with requirements, formulation, and reference comparison needs.
+- The final HDF5 projection maps every approved logical quantity and source identity from `docs/<feature-id>/reference-model.md` to deterministic dataset and CSV component identities.
 
 ## Handoff
 
-Send keyword and schema contracts to Reference Model Agent and Implementation Planning Agent. Send HDF5 dataset paths, reference CSV row schemas, ID matching, and tolerance-source constraints to Reference Verification Agent.
+Return `docs/<feature-id>/io.md`, status, evidence summary, and blockers to Coordinator Agent. Route the final HDF5 projection, parser/schema contracts, exact CSV paths, source-ID/component matching, row prechecks, and tolerance-source constraints to Implementation Planning Agent through Coordinator Agent.

.codex/skills/fesa-numerical-review/SKILL.md

@@ -1,11 +1,13 @@
 ---
 name: fesa-numerical-review
-description: Use when independently reviewing FESA FEM numerical review evidence, formulation correctness, stability risks, patch tests, locking, Jacobian handling, and implementation planning readiness.
+description: Use when independently reviewing FESA FEM formulation numerical correctness, stability risks, verification readiness, and existing reference-case input/CSV readiness before I/O definition.
 ---
 
-# FESA Numerical Review
+# FESA Numerical and Reference Review
 
-Use this skill to review a formulation as a numerical algorithm contract before implementation planning.
+Use this skill to review a formulation as a numerical algorithm contract and to inventory the
+minimum existing reference inputs and CSV quantities required by the approved feature scope before
+I/O definition.
 
 ## Inputs
 
@@ -13,27 +15,35 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/numerical-reviews/README.md`
-- `docs/formulations/<feature-id>-formulation.md`
-- Related requirements and research documents when needed
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/research.md`
+- `docs/<feature-id>/formulation.md`
+- Existing stored reference artifacts under `reference/`, read-only
+- Coordinator Agent dispatch package for the numerical/reference gate
 
 ## Workflow
 
-1. Lead with findings and required revisions.
-2. Check dimensional consistency, signs, DOF ordering, constrained/free assumptions, and coordinate transforms.
-3. Review B matrix or kinematic operator consistency.
-4. Review constitutive matrix or stress update contract.
-5. Review Jacobian rules, determinant checks, derivative transforms, and distortion handling.
-6. Review integration rule, Gauss points, weights, and full/reduced/selective integration policy.
-7. Check element residual, internal force, external force, stiffness, tangent, symmetry, and positive definiteness expectations.
-8. Assess only rigid modes, patch tests, locking, singularity, conditioning, convergence, and
-   other risks required by the approved feature scope. Do not invent calibration or portfolio gates.
-9. Decide status: `pass-for-implementation-planning`, `needs-formulation-revision`,
-   `needs-research`, or `blocked`.
+FORMULATION REVIEW -> REFERENCE CASE INVENTORY -> CLASSIFY -> REPORT -> I/O HANDOFF
+
+1. **FORMULATION REVIEW:** Lead with findings and required revisions. Check dimensions, signs,
+   DOF ordering, constrained/free assumptions, coordinate transforms, kinematic operator,
+   constitutive contract, Jacobian and derivative rules, integration policy, element residual,
+   internal force, external force, stiffness, tangent consistency, symmetry, and positive-definiteness
+   expectations, and only the feature-approved stability and verification risks.
+2. **REFERENCE CASE INVENTORY:** List each existing case using its exact directory, input filename,
+   and required CSV filenames. Confirm declared input and every required comparison CSV are present
+   and readable without renaming, repairing, normalizing, generating, or modifying artifacts.
+   Define logical blocking and warning-only quantities, source identity/component matching, row
+   prechecks for missing, extra, duplicate, and nonfinite required rows, and the approved tolerance.
+3. **CLASSIFY:** Use only `pass-for-io-definition`, `needs-formulation-revision`,
+   `needs-research`, `needs-reference-artifacts`, `needs-user-decision`, or `blocked`.
+4. **REPORT:** Produce both reports in the Output Contract.
+5. **I/O HANDOFF:** Send both reports through Coordinator Agent to I/O Definition Agent. I/O
+   Definition owns the final HDF5 dataset projection; this review owns logical quantities and source identity.
 
 ## Output Contract
 
-Produce or revise `docs/numerical-reviews/<feature-id>-review.md` with:
+Produce or revise `docs/<feature-id>/numerical-review.md` with:
 
 - Metadata and source formulation
 - Review Verdict
@@ -44,27 +54,43 @@ Produce or revise `docs/numerical-reviews/<feature-id>-review.md` with:
 - Required Revisions
 - Downstream Handoff
 
+Produce or revise `docs/<feature-id>/reference-model.md` with:
+
+- Metadata
+- Reference Acceptance Scope
+- Reference Case Inventory: exact input and required CSV paths
+- Source Identity and Component Contract: logical quantity, source identity, component matching,
+  and blocking or warning-only classification
+- Row Prechecks: missing, extra, duplicate, and nonfinite required rows fail before tolerance
+- Tolerance and Blocking/Warning Policy: exact approved tolerance without invention or calibration
+- Open Issues and I/O Handoff: logical quantity and source identity information for the I/O contract
+
 ## Boundaries
 
 - Do not implement code.
 - Do not edit formulations directly.
 - Do not design C++ APIs or file ownership.
 - Do not run Abaqus, Nastran, or any reference solver.
-- Do not generate or modify Abaqus reference CSV files.
+- Do not generate, rename, repair, normalize, or modify reference inputs or CSV files.
 - Do not approve release readiness.
 - Do not decide reference comparison success.
+- Do not include final HDF5 dataset paths; I/O Definition owns that projection.
 
 ## Quality Gate
 
-- `pass-for-implementation-planning` means implementation planning may begin, not that the feature is complete.
+- `pass-for-io-definition` means both reviews are ready for I/O definition, not that the feature is complete.
 - Confirmed defects, risks, open questions, and test recommendations are separated.
 - Missing derivations are returned to Formulation Agent instead of being silently fixed.
-- Evidence gaps are routed to Research Agent or Reference Model Agent.
-- Missing downstream Reference Model documents, canonical naming, README, metadata, provenance,
-  extended portfolios, or comparison results do not block a formulation verdict.
+- Missing theory or benchmark evidence is routed to Research Agent.
+- Use `needs-reference-artifacts` only when a declared input or required comparison CSV is missing.
+- Use `needs-user-decision` only when a required quantity, source identity/component match,
+  tolerance, or supported-keyword policy is undefined.
+- Canonical naming, README, metadata, provenance, extended portfolios, and unrequired CSV files do
+  not block readiness.
+- Both canonical reports must return output paths, status, evidence summary, and blockers before the gate can pass.
 
 ## Handoff
 
-Send pass results to Implementation Planning Agent and optional test notes to Reference Model
-Agent. Send math defects to Formulation Agent, source gaps to Research Agent, and blocked decisions
-to Coordinator Agent.
+Return `docs/<feature-id>/numerical-review.md` and `docs/<feature-id>/reference-model.md`, status,
+evidence summary, and blockers to Coordinator Agent. Route both passed reports to I/O Definition
+Agent, math defects to Formulation Agent, and source gaps to Research Agent through Coordinator Agent.

.codex/skills/fesa-numerical-review/agents/openai.yaml

@@ -1,4 +1,4 @@
 interface:
-  display_name: "FESA Numerical Review"
-  short_description: "Review FEM numerical risks"
-  default_prompt: "Use $fesa-numerical-review to review FESA formulation numerical readiness."
+  display_name: "FESA Numerical and Reference Review"
+  short_description: "Review numerical and reference readiness"
+  default_prompt: "Use $fesa-numerical-review to review FESA numerical and reference-model readiness."

.codex/skills/fesa-physics-sanity/SKILL.md

@@ -1,6 +1,6 @@
 ---
 name: fesa-physics-sanity
-description: Use when evaluating FESA solver physics and physical plausibility after reference verification, including equilibrium, reactions, displacement direction, symmetry, stress sanity, and model coverage.
+description: Use when evaluating FESA solver physics and physical plausibility after a passing implementation-owned reference comparison, including equilibrium, reactions, displacement direction, symmetry, stress sanity, and model coverage.
 ---
 
 # FESA Physics Sanity
@@ -13,16 +13,18 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/physics-evaluations/README.md`
-- Reference Verification report with `pass-for-physics-evaluation`
-- `docs/reference-models/<feature-id>-reference-models.md`
-- Requirements, formulation, numerical review, and I/O definition documents
+- `docs/<feature-id>/reference-comparison.md` with `pass-for-physics-evaluation`
+- `docs/<feature-id>/reference-model.md`
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/formulation.md`
+- `docs/<feature-id>/numerical-review.md`
+- `docs/<feature-id>/io.md`
 - Solver results.h5, feature-declared Abaqus reference CSV files, and optional FESA debug CSV views as read-only evidence
 
 ## Workflow
 
 1. Evaluate only documented physical expectations.
-2. Require a reference verification status of `pass-for-physics-evaluation`.
+2. Require `pass-for-physics-evaluation` in `docs/<feature-id>/reference-comparison.md`.
 3. Check global equilibrium when loads, reactions, and sign conventions are documented.
 4. Check reaction consistency for constrained DOFs.
 5. Check displacement direction against loads, boundary conditions, and expected deformation mode.
@@ -35,7 +37,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/physics-evaluations/<feature-id>-physics-evaluation.md` with:
+Produce or revise `docs/<feature-id>/physics-evaluation.md` with:
 
 - Metadata
 - Input Evidence
@@ -59,11 +61,11 @@ Produce or revise `docs/physics-evaluations/<feature-id>-physics-evaluation.md`
 
 ## Quality Gate
 
-- A physics pass requires documented expectations and reference verification pass evidence.
+- A physics pass requires documented expectations and passing reference-comparison evidence.
 - Use `needs-upstream-decision` when physical expectations, sign convention, or model purpose is missing.
-- Use `needs-reference-model` only when a feature-required case or declared physical expectation is missing.
+- Route a feature-required case, declared physical expectation, or model-coverage gap to Numerical Review.
 - `pass-for-release-agent` means Release Agent can audit release readiness; it is not release approval.
 
 ## Handoff
 
-Send `pass-for-release-agent` reports to Release Agent. Send implementation-owned physics failures to Correction Agent, formulation concerns to Formulation Agent, I/O ambiguity to I/O Definition Agent, and model coverage gaps to Reference Model Agent.
+Return `docs/<feature-id>/physics-evaluation.md`, status, evidence summary, and blockers to Coordinator Agent. Route `pass-for-release-agent` to Release Agent, implementation-owned failures to Correction Agent, formulation concerns to Formulation Agent, I/O ambiguity to I/O Definition Agent, and model-coverage gaps to Numerical Review Agent, always through Coordinator Agent.

.codex/skills/fesa-reference-comparison/SKILL.md

@@ -1,77 +0,0 @@
----
-name: fesa-reference-comparison
-description: Use when comparing FESA results.h5 quantities against feature-declared Abaqus CSV values or diagnosing reference tolerance, identity, row-set, or nonfinite-result failures.
----
-
-# FESA Reference Comparison
-
-Compare generated FESA HDF5 values with only the Abaqus CSV quantities declared by the feature.
-The comparison tests observable values; it does not test Abaqus implementation equivalence.
-
-## Inputs
-
-Read these first:
-
-- `AGENTS.md`
-- `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/reference-verifications/README.md`
-- Build/Test report with `pass-for-reference-verification`
-- `docs/reference-models/<feature-id>-reference-models.md`
-- `docs/io-definitions/<feature-id>-io.md`
-- Generated FESA `results.h5`
-- Declared reference `.inp` and required Abaqus CSV files
-
-## Workflow
-
-1. Follow `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT`.
-2. ARTIFACT CHECK: verify the declared input, `results.h5`, every required CSV, HDF5
-   projection, source-ID/component mapping, row prechecks, blocking/warning behavior, and tolerance.
-3. Reject a missing input or required CSV as `needs-reference-artifacts`.
-4. Reject a missing HDF5 result or comparison command as `needs-solver-results`.
-5. Reject missing, extra, duplicate, or nonfinite required rows before tolerance evaluation.
-6. COMPARE: normalize HDF5 values and match CSV values by declared source identity and component.
-   Never match by row order alone.
-7. Apply the upstream tolerance exactly. Do not clamp values, drop rows, loosen tolerance, or
-   modify either artifact to obtain a pass.
-8. Report per-row decisions, max absolute error, max relative error, RMS error, norm error when
-   required, worst source ID/component, and pass/fail or warning per quantity.
-9. CLASSIFY failures as missing-reference-artifact, missing-solver-output, schema-mismatch,
-   id-mismatch, tolerance-failure, nonfinite-result, upstream-contract, or environment.
-
-Do not require canonical names, README, metadata, Abaqus version/provenance, duplicated CSV
-units/coordinates/step-frame fields, or a reference CSV schema version unless the feature
-requirements explicitly make one of them part of comparison acceptance.
-
-## Output Contract
-
-Produce or revise `docs/reference-verifications/<feature-id>-reference-verification.md` with:
-
-- Metadata
-- Artifact Inventory
-- Comparison Contract
-- Quantity Results
-- Failure Classification
-- Handoff Recommendation
-- No-Change Assertion
-- Open Issues
-
-## Boundaries
-
-- Do not edit source, tests, CMake, upstream contracts, reference artifacts, or tolerances.
-- Do not run Abaqus, Nastran, or another reference solver.
-- Do not generate or modify reference CSV files.
-- Do not approve physics validation or release readiness.
-
-## Quality Gate
-
-- Every declared required row has a deterministic source-ID/component match.
-- Missing/extra/duplicate/nonfinite required rows remain visible and fail before tolerance.
-- Warning-only quantities never change the blocking pass/fail verdict.
-- `pass-for-physics-evaluation` means required reference tolerance success only.
-- FESA `results.h5` remains authoritative solver output; Abaqus CSV remains external reference data.
-
-## Handoff
-
-Send passing reports to Physics Evaluation Agent. Send implementation-owned mismatches to
-Correction Agent, missing declared files to Reference Model Agent, and HDF5 projection or
-identity-contract conflicts to I/O Definition Agent.

.codex/skills/fesa-reference-comparison/agents/openai.yaml

@@ -1,4 +0,0 @@
-interface:
-  display_name: "FESA Reference Comparison"
-  short_description: "Compare HDF5 with Abaqus CSV"
-  default_prompt: "Use $fesa-reference-comparison to compare FESA solver results.h5 against Abaqus reference CSV files."

.codex/skills/fesa-reference-models/SKILL.md

@@ -1,83 +0,0 @@
----
-name: fesa-reference-models
-description: Use when a FESA feature needs existing Abaqus input/CSV reference cases, required comparison quantities, tolerance mapping, or implementation-planning handoff.
----
-
-# FESA Reference Cases
-
-Use this skill to inventory the minimum stored reference inputs and CSV quantities that a
-feature actually compares. Abaqus is an external numerical reference, not the FESA formulation
-or behavior specification.
-
-## Inputs
-
-Read these first:
-
-- `AGENTS.md`
-- `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/reference-models/README.md`
-- `docs/requirements/<feature-id>.md`
-- `docs/formulations/<feature-id>-formulation.md`
-- `docs/numerical-reviews/<feature-id>-review.md`
-- `docs/io-definitions/<feature-id>-io.md`
-- Existing files under `reference/`
-
-## Workflow
-
-1. Read the feature requirements and list only blocking and warning-only reference quantities.
-2. Inventory each existing case using its exact directory, input filename, and required CSV filenames.
-3. Confirm the input and every required comparison CSV are present and readable.
-4. Record the FESA HDF5 dataset, source identity column, required component columns, and stable matching rule.
-5. Require unique source IDs, complete required row sets, and finite comparison values.
-6. Copy the approved tolerance and blocking/warning behavior without inventing or calibrating values.
-7. Map each reference-comparison requirement to at least one existing case when the feature requires that coverage.
-
-Do not require or create:
-
-- canonical directories or filenames;
-- legacy-alias approvals;
-- bundle `README.md` or `metadata.json`;
-- Abaqus version or generation provenance;
-- duplicated units, coordinates, step/frame, material, section, thickness, or element-type metadata;
-- a reference CSV schema version;
-- CSV files for quantities outside the approved comparison boundary;
-- an expanded benchmark portfolio unless the feature requirements explicitly require it.
-
-Read source element type, material, section, loads, constraints, and supported single-step
-identity from the `.inp`. Do not rename, repair, normalize, or generate reference artifacts.
-
-## Output Contract
-
-Produce or revise `docs/reference-models/<feature-id>-reference-models.md` with:
-
-- Metadata
-- Reference Acceptance Scope
-- Reference Case Inventory
-- HDF5-to-CSV Comparison Mapping
-- Tolerance and Blocking/Warning Policy
-- Readiness Checklist
-- Open Issues and Downstream Handoff
-
-## Boundaries
-
-- Do not implement code or parsers.
-- Do not design C++ APIs or file ownership.
-- Do not run Abaqus, Nastran, or another reference solver.
-- Do not generate or modify reference inputs or CSV files.
-- Do not compare solver results.
-- Do not approve release readiness.
-
-## Quality Gate
-
-- Every blocking or warning-only quantity names an exact input/CSV pair.
-- Every comparison defines an HDF5 projection, source-ID/component mapping, and tolerance.
-- Missing, extra, duplicate, and nonfinite required rows have an explicit fail-before-tolerance rule.
-- Use `needs-reference-artifacts` only when a declared input or required comparison CSV is missing.
-- Use `needs-user-decision` only when required quantities, matching, or tolerance are undefined.
-- Canonical naming, README, metadata, provenance, and unrequested portfolio coverage never block readiness.
-
-## Handoff
-
-Send exact paths, HDF5 projection, source-ID/component matching, row prechecks, and tolerance to
-Implementation Planning Agent and Reference Verification Agent. Send only feature-required
-physical expectations to Physics Evaluation Agent.

.codex/skills/fesa-reference-models/agents/openai.yaml

@@ -1,4 +0,0 @@
-interface:
-  display_name: "FESA Reference Cases"
-  short_description: "Inventory required input and CSV cases"
-  default_prompt: "Use $fesa-reference-models to inventory the existing Abaqus input and required CSV cases for a FESA feature."

.codex/skills/fesa-release-readiness/SKILL.md

@@ -13,17 +13,24 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/releases/README.md`
-- Physics Evaluation report with `pass-for-release-agent`
-- Reference Verification report with `pass-for-physics-evaluation`
-- Build/Test report with `pass-for-reference-verification`
-- Requirements, formulation, numerical review, I/O definition, reference model, implementation, and correction reports
+- `docs/<feature-id>/requirements.md`
+- `docs/<feature-id>/research.md`
+- `docs/<feature-id>/formulation.md`
+- `docs/<feature-id>/numerical-review.md`
+- `docs/<feature-id>/reference-model.md`
+- `docs/<feature-id>/io.md`
+- `docs/<feature-id>/implementation-plan.md`
+- `docs/<feature-id>/implementation-report.md`
+- `docs/<feature-id>/build-test.md` with passing full validation evidence
+- `docs/<feature-id>/reference-comparison.md` with `pass-for-physics-evaluation`
+- `docs/<feature-id>/corrections.md` when correction occurred
+- `docs/<feature-id>/physics-evaluation.md` with `pass-for-release-agent`
 
 ## Workflow
 
 1. Follow `GATE AUDIT -> TRACEABILITY CHECK -> RELEASE DOCUMENTATION -> RELEASE VERDICT`.
 2. GATE AUDIT: confirm required reports exist, share the same `feature_id`, are not stale or contradictory, and carry required pass statuses.
-3. Require `pass-for-reference-verification`, `pass-for-physics-evaluation`, and `pass-for-release-agent`.
+3. Require passing full validation evidence, `pass-for-physics-evaluation`, and `pass-for-release-agent`.
 4. TRACEABILITY CHECK: confirm each `must` requirement maps to acceptance criteria, test evidence,
    feature-required reference evidence when applicable, and release scope.
 5. Record deferred requirements, unsupported Abaqus keywords, missing required comparison files,
@@ -33,7 +40,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/releases/<feature-id>-release.md` with:
+Produce or revise `docs/<feature-id>/release.md` with:
 
 - Metadata
 - Release Scope
@@ -58,7 +65,7 @@ Produce or revise `docs/releases/<feature-id>-release.md` with:
 
 ## Quality Gate
 
-- Do not issue `ready-for-release` without `pass-for-release-agent`, `pass-for-physics-evaluation`, and `pass-for-reference-verification`.
+- Do not issue `ready-for-release` without `pass-for-release-agent`, `pass-for-physics-evaluation`, and passing full build/test evidence.
 - Every `must` requirement traces to release scope, acceptance criteria, test or reference evidence, and final disposition.
 - Known limitations and deferred issues are included in the Release Notes Draft.
 - Missing required evidence, contradictory reports, unresolved defects, missing declared comparison
@@ -67,4 +74,4 @@ Produce or revise `docs/releases/<feature-id>-release.md` with:
 
 ## Handoff
 
-Send `ready-for-release` to Coordinator Agent for final workflow closure. Send missing documentation to Release Agent revision, missing verification to Reference Verification Agent or Physics Evaluation Agent, and implementation defects to Correction Agent.
+Return `docs/<feature-id>/release.md`, status, evidence summary, blockers, and a closure recommendation to Coordinator Agent. Return missing implementation/build/comparison evidence, missing physics evidence, upstream gaps, or documentation gaps to Coordinator Agent for the owning sub-agent; do not advance or close the workflow directly.

.codex/skills/fesa-requirements-baseline/SKILL.md

@@ -13,9 +13,8 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/requirements/README.md`
 - User feature request, target capability, constraints, and known exclusions
-- Existing `docs/requirements/<feature-id>.md` when revising a feature
+- Existing `docs/<feature-id>/requirements.md` when revising a feature
 
 ## Workflow
 
@@ -31,7 +30,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/requirements/<feature-id>.md` with:
+Produce or revise `docs/<feature-id>/requirements.md` with:
 
 - Metadata with `feature_id`, status, owner agent, and date
 - Purpose, In Scope, Out Of Scope, and Analysis Definition
@@ -60,7 +59,8 @@ Produce or revise `docs/requirements/<feature-id>.md` with:
 - Do not require canonical names, README, metadata, provenance, or CSVs for quantities outside
   the feature acceptance boundary.
 - Words like "accurate", "fast", and "Abaqus-like" are converted into measurable criteria or open questions.
+- Return reference artifact, comparison quantity, source-identity/component, and tolerance gaps for Numerical Review rather than inventing acceptance rules.
 
 ## Handoff
 
-Route theory gaps to Research Agent, math gaps to Formulation Agent, schema gaps to I/O Definition Agent, reference artifact needs to Reference Model Agent, and implementation readiness to Implementation Planning Agent.
+Return `docs/<feature-id>/requirements.md`, status, evidence summary, and blockers to Coordinator Agent. Route theory questions to Research Agent and reference acceptance, artifact, identity/component, and tolerance questions to Numerical Review Agent through Coordinator Agent.

.codex/skills/fesa-research-evidence/SKILL.md

@@ -13,8 +13,7 @@ Read these first:
 
 - `AGENTS.md`
 - `docs/SOLVER_AGENT_DESIGN.md`
-- `docs/research/README.md`
-- `docs/requirements/<feature-id>.md`
+- `docs/<feature-id>/requirements.md`
 - User-supplied books, papers, manuals, or benchmark constraints
 
 ## Workflow
@@ -29,7 +28,7 @@ Read these first:
 
 ## Output Contract
 
-Produce or revise `docs/research/<feature-id>-research.md` with:
+Produce or revise `docs/<feature-id>/research.md` with:
 
 - Metadata and source requirement path
 - Research Questions
@@ -55,7 +54,8 @@ Produce or revise `docs/research/<feature-id>-research.md` with:
 - Benchmark candidates include what quantity they can verify and what they cannot verify.
 - Missing source evidence is carried forward as an open issue.
 - No reference value, tolerance, or compatibility claim is invented.
+- Benchmark/reference evidence identifies exact candidate artifacts, target quantities, applicability limits, and unresolved acceptance questions for Numerical Review.
 
 ## Handoff
 
-Send formulation evidence to Formulation Agent, benchmark and source limits to Numerical Review Agent, artifact candidates to Reference Model Agent, and unresolved source gaps to Coordinator Agent.
+Return `docs/<feature-id>/research.md`, status, evidence summary, and blockers to Coordinator Agent. Route formulation evidence to Formulation Agent and benchmark/reference evidence, artifact candidates, and source limits to Numerical Review Agent through Coordinator Agent.

.gitignore

@@ -18,10 +18,12 @@ Testing/
 __pycache__/
 *.pyc
 *.h5
+.worktrees/
 
 # local Harness configuration and build outputs
 .harness/config.json
 .harness/build/
+.harness/doxygen/
 
 # phase execution outputs
 phases/**/phase*-output.json

AGENTS.md

@@ -15,7 +15,9 @@
 - 프로젝트의 배경, 목적, 사용자, 범위는 `docs/PRD.md`에서 확인한다.
 - 전체 아키텍처와 모듈별 책임은 `docs/ARCHITECTURE.md`에서 확인한다.
 - 주요 아키텍처 결정과 그 이유 및 트레이드오프는 `docs/ADR.md`에서 확인한다.
-- AI 에이전트는 설계나 구현 결정을 내리기 전에 PRD에서 제품 범위를, ARCHITECTURE에서 소유권과 데이터 흐름을, ADR에서 이미 결정된 트레이드오프를 먼저 확인한다. 기능별 의미는 `docs/requirements/`, `docs/formulations/`, `docs/io-definitions/`, `docs/reference-models/`의 승인 문서를 source of truth로 삼는다.
+- 외부 reference tolerance 값, 판정 과정, 보고서와 변경 관리는
+  `docs/TOLERANCE.md`를 따른다.
+- AI 에이전트는 설계나 구현 결정을 내리기 전에 PRD에서 제품 범위를, ARCHITECTURE에서 소유권과 데이터 흐름을, ADR에서 이미 결정된 트레이드오프를 먼저 확인한다. 기능별 요구조건, 연구, 정식화, 수치/reference 검토, I/O, 구현, 검증, 물리 및 release 의미는 `docs/<feature-id>/`의 승인 문서를 source of truth로 삼는다.
 - Harness는 솔버 자체가 아니라 요구조건, TDD, phase 실행, 검증을 통제하는 개발 운영 인프라이다. 전체 실행 흐름은 `docs/HARNESS_WORKFLOW.md`, 설치와 설정은 `docs/HARNESS.md`를 따른다.
 - FESA는 Abaqus와 독립적인 솔버다. 문서와 구현은 full Abaqus compatibility뿐 아니라
   Abaqus 요소 정식화, 적분, stabilization, 내부 상태 또는 결과 생성 절차의 동등성을
@@ -32,6 +34,18 @@
 - 비교기는 필요한 source row identity와 component를 결정적으로 대응시키고 누락, 추가,
   중복, nonfinite 값을 tolerance 전에 거부한다. Reference artifact는 rename, rewrite 또는
   보정하지 않는다.
+- 모든 외부 reference comparison은 `docs/TOLERANCE.md`의 공통 정책을 사용한다.
+  설계 근거는 `docs/superpowers/specs/2026-08-17-common-reference-tolerance-design.md`와
+  ADR-022에 보존한다. 동일 logical quantity, unit dimension, coordinate system과 blocking behavior의
+  component family에서 Abaqus-only scale `S=max(abs(reference))`를 구한다. `abs(reference)`가
+  `0.01*S` 이하인 행은 `abs(fesa-reference) <= 0.01*S`, 그 외 행은 상대오차 `<=0.05`로
+  판정하고, family scale-relative RMS `RMS(error)/S <= 0.01`도 함께 통과해야 한다. 별도
+  absolute-error gate는 사용하지 않으며 zero-scale family는 FESA도 exact zero일 때만
+  통과한다.
+- B33 section-resultant reference는 `(instance, element label, endpoint node label,
+  component)`로 HDF5 element endpoint와 직접 대응한다. Node-station collapse 또는 평균을
+  사용하지 않는다. MITC4 `U1/U2/U3`는 blocking, `UR1/UR2/UR3`는 warning-only 분류를
+  유지하면서 같은 공통 수치 정책을 사용한다.
 - CSV는 FESA 공식 output이 아니며, FESA HDF5에서 추출한 deterministic CSV view는 비교 디버깅/검토용 보조 artifact로만 둔다.
 
 ## FESA 개발의 핵심 원칙
@@ -79,31 +93,6 @@
 - Stop은 `.harness/config.json` 또는 자동 감지 결과에 따라 MSVC build와 test를 모두 검증한다.
 - Generated phase execution outputs remain ignored under `phases/**/step*-output.json`.
 
-## 현재 승인된 V0 기능 계약
-- `linear-static-3d-euler-beam`의 source of truth는 `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`이다. 관련 문서를 변경하거나 구현할 때 이 계약의 의미를 임의로 넓히지 않는다.
-- 입력 파일당 하나의 `*STEP, *STATIC`과 `TYPE=B33`만 지원한다. B31을 Euler 요소로 매핑하지 않고 `unsupported-element-formulation`으로 거부한다.
-- `*PART/*ASSEMBLY/*INSTANCE`는 identity instance와 stable source identity만 지원한다. instance transform과 nested assembly는 거부한다.
-- 선형 정적 실행 순서는 stiffness assembly와 constrained partition 뒤 `Kff`를 factorize하고, 그 다음 load vector와 effective RHS를 조립해 substitution하는 순서를 유지한다.
-- FESA는 output request와 무관하게 displacement, reaction, equilibrium end action, section resultant, generalized strain/resultant, axial `S11`을 HDF5에 기록한다. Beam stress의 Abaqus reference comparison은 N/A다.
-- B33 reference row는 component별 Abaqus scale로 `absolute_floor + 1e-6 * reference_scale`을 적용한다. Reference 값을 zero-clamp하거나 누락 row를 무시하지 않는다.
-- `reference/cantilever beam/` artifact는 rename, rewrite 또는 보정하지 않는다.
-
-## 승인된 MITC4 정책
-
-- Abaqus source `S4`와 `S4R`은 같은 FESA MITC4 정식화로 매핑하며 source type은 추적
-  정보로만 보존한다. Abaqus의 S4/S4R 알고리즘을 재현하지 않는다.
-- Reference case는 `reference/shell/`의 S4 input/displacement CSV와
-  `reference/shellR/`의 S4R input/displacement CSV를 현재 이름 그대로 사용한다.
-- Global `U1/U2/U3`만 blocking reference quantity다. 각 case/component에 B33과 같은
-  `1e-9 + 1e-6 * reference_scale`을 적용한다. `UR1/UR2/UR3`은 같은 기준을 초과하면
-  warning만 남긴다.
-- MITC4 drilling calibration, coefficient sweep, energy-ratio threshold와 별도 geometry/director
-  calibration(`NR-O01`~`NR-O04`)은 구현 범위가 아니다. 6-DOF 선형계에는 물리 회전
-  stiffness block의 양의 최소 대각항에 `1e-3`을 곱한 고정 numerical drilling
-  stabilization만 사용하며 drilling 방향 nodal moment는 지원하지 않는다.
-- Drilling stabilization은 physical generalized strain/resultant/stress에 포함하지 않으며
-  별도 drilling stiffness/ratio/energy HDF5 output을 요구하지 않는다.
-
 ## 기능을 추가할 때의 판단 기준
 
 - 새 element는 요구조건과 formulation을 승인한 뒤 semantic mapping, property/material 연결, DOF/scatter, local kernel, deterministic assembly, recovery, HDF5 row identity, reference/physics evidence를 함께 설계한다.
@@ -116,37 +105,38 @@
 ## 개발 프로세스
 - TDD를 기본으로 한다. 구현은 `RED -> GREEN -> VERIFY` 순서를 따른다.
 - CRITICAL: 빌드 경고를 새로 추가하지 말 것.
-- 기능 개발은 다음 gate를 순서대로 통과해야 한다.
-  1. 요구조건 분석
-  2. 연구자료 조사
-  3. 유한요소 정식화
-  4. 수치 검토
-  5. I/O 계약 정의
-  6. reference model 계약 준비
-  7. C++ 구현
-  8. build/test 검증
-  9. reference comparison
-  10. physics sanity
-  11. release readiness
+- 기능 개발은 다음 8단계 gate를 순서대로 통과해야 한다.
+  1. 요구조건
+  2. 연구
+  3. 정식화
+  4. 수치 검토 + reference model 계약
+  5. I/O 정의
+  6. 구현 계획 + C++ 구현 + build/test + reference comparison
+  7. 물리 검토
+  8. 배포 준비
 - 커밋 메시지는 conventional commits 형식을 따른다: `feat:`, `fix:`, `docs:`, `refactor:`, `test:`, `chore:`.
 
 ## Agent/Skill Workflow
-AI 에이전트는 유한요소 구조해석 지식이 필요할 때 FEM wiki를 참조하기 위해 `.codex/skills/fem-theory-query` 스킬을 사용할 수 있다.
+`coordinator-agent`는 유일한 main agent로서 `docs/<feature-id>/coordination.md`의 8단계
+worklist, sub-agent dispatch, evidence gate와 workflow closure를 관리한다. 나머지 모든
+profile은 Coordinator가 한정된 단계 작업으로 호출하는 sub-agent이며, peer agent를 직접
+호출하거나 다음 gate로 진행하지 않는다. 유한요소 구조해석 지식이 필요할 때는
+`.codex/skills/fem-theory-query`를 보조 skill로 사용할 수 있다.
 
-| 개발 과정 | Agent | Skill | 산출물 |
+| 단계 | 담당 sub-agent | 필수 skill | `docs/<feature-id>/` 산출물 |
 | --- | --- | --- | --- |
-| 요구조건 분석 | `requirement-agent` | `fesa-requirements-baseline` | `docs/requirements/<feature-id>.md` |
-| 연구자료 조사 | `research-agent` | `fesa-research-evidence`, `fem-theory-query` | `docs/research/<feature-id>-research.md` |
-| 유한요소 정식화 | `formulation-agent` | `fesa-formulation-spec` | `docs/formulations/<feature-id>-formulation.md` |
-| 수치 검토 | `numerical-review-agent` | `fesa-numerical-review` | `docs/numerical-reviews/<feature-id>-review.md` |
-| I/O 정의 | `io-definition-agent` | `fesa-io-contract` | `docs/io-definitions/<feature-id>-io.md` |
-| reference model | `reference-model-agent` | `fesa-reference-models` | `docs/reference-models/<feature-id>-reference-models.md` |
-| 구현 계획/구현 | `implementation-planning-agent`, `implementation-agent` | `fesa-cpp-msvc-tdd` | tests, source, implementation report |
-| build/test | `build-test-executor-agent` | `fesa-cpp-msvc-tdd` | `docs/build-test-reports/<feature-id>.md` |
-| correction | `correction-agent` | `fesa-cpp-msvc-tdd` | `docs/corrections/<feature-id>.md` |
-| reference 비교 | `reference-verification-agent` | `fesa-reference-comparison` | `docs/reference-verifications/<feature-id>-reference-verification.md` |
-| 물리 검토 | `physics-evaluation-agent` | `fesa-physics-sanity` | `docs/physics-evaluations/<feature-id>-physics-evaluation.md` |
-| 배포 준비 | `release-agent` | `fesa-release-readiness` | `docs/releases/<feature-id>-release.md` |
+| 1. 요구조건 | `requirement-agent` | `fesa-requirements-baseline` | `requirements.md` |
+| 2. 연구 | `research-agent` | `fesa-research-evidence`, 필요 시 `fem-theory-query` | `research.md` |
+| 3. 정식화 | `formulation-agent` | `fesa-formulation-spec` | `formulation.md` |
+| 4. 수치 검토 + reference model 계약 | `numerical-review-agent` | `fesa-numerical-review` | `numerical-review.md`, `reference-model.md` |
+| 5. I/O 정의 | `io-definition-agent` | `fesa-io-contract` | `io.md` |
+| 6. 구현 계획 + C++ 구현 + build/test + reference comparison | `implementation-planning-agent`, `implementation-agent` | `fesa-cpp-msvc-tdd`, 계획 시 project-local `harness` | `implementation-plan.md`, `implementation-report.md`, `build-test.md`, `reference-comparison.md` |
+| 7. 물리 검토 | `physics-evaluation-agent` | `fesa-physics-sanity` | `physics-evaluation.md` |
+| 8. 배포 준비 | `release-agent` | `fesa-release-readiness` | `release.md` |
+
+`correction-agent`는 정규 단계가 아니라 반복되거나 원인이 불명확한 구현 실패를 다루는
+on-demand rework sub-agent다. Coordinator가 호출할 때만 `docs/<feature-id>/corrections.md`에
+원인, 수정과 재검증 결과를 기록하고 Implementation 단계로 반환한다.
 
 Implementation Planning Agent는 구현 계획 요청에서 `.agents/skills/harness`를 반드시
 사용한다. 먼저 여러 자기완결적 Step의 초안을 제시하고 사용자 승인을 받은 뒤에만

CMakeLists.txt

@@ -8,6 +8,16 @@ set(CMAKE_CXX_EXTENSIONS OFF)
 
 include(cmake/FesaDependencies.cmake)
 
+find_package(Doxygen QUIET)
+if(Doxygen_FOUND)
+  add_custom_target(fesa_docs
+    COMMAND "${DOXYGEN_EXECUTABLE}" "${CMAKE_CURRENT_SOURCE_DIR}/Doxyfile"
+    WORKING_DIRECTORY "${CMAKE_CURRENT_SOURCE_DIR}"
+    COMMENT "Generating FESA API documentation"
+    VERBATIM
+  )
+endif()
+
 enable_testing()
 
 add_subdirectory(src/fesa)

Doxyfile

@@ -0,0 +1,16 @@
+PROJECT_NAME = FESA
+PROJECT_NUMBER = 0.1.0
+OUTPUT_DIRECTORY = .harness/doxygen
+INPUT = include src
+EXCLUDE = tests
+RECURSIVE = YES
+FILE_PATTERNS = *.h *.cpp
+EXTRACT_ALL = NO
+EXTRACT_PRIVATE = YES
+EXTRACT_STATIC = YES
+WARN_IF_UNDOCUMENTED = YES
+WARN_IF_DOC_ERROR = YES
+WARN_AS_ERROR = YES
+GENERATE_HTML = YES
+HTML_OUTPUT = html
+GENERATE_LATEX = NO

docs/ADR.md

@@ -35,7 +35,12 @@ solution과 test command를 명시한 직접 MSBuild 프로젝트도 검증할
 **트레이드오프**: 사용자는 기존 Abaqus input file을 그대로 사용할 수 없을 수 있다. 대신 지원 범위와 실패 원인이 명확해진다.
 
 ### ADR-004: Domain, AnalysisModel, DofManager, AnalysisState를 분리한다
-**결정**: `Domain`은 입력 모델 정의를 소유하고, `AnalysisModel`은 현재 step의 실행 view를 제공하며, `DofManager`는 equation numbering과 constrained/free mapping을 전담하고, `AnalysisState`는 해석 중 변하는 물리량을 소유한다.
+**결정**: `Domain`은 `ElementDefinition`, `ElementProperty`, `Material`과
+`StepDefinition` 입력 모델 정의를 `std::unique_ptr`로 단독 소유하고 const access와
+stable collection index를 제공한다. `AnalysisModel`은 Domain 수명 안에서 stable index와
+const reference만 사용하는 non-owning current-step view다. `DofManager`는 equation
+numbering과 constrained/free mapping을 전담하고, `AnalysisState`는 해석 중 변하는
+물리량을 소유한다.
 
 **이유**: 모델 정의, step activation, equation system, transient/nonlinear state가 섞이면 parser, assembler, solver, result writer가 강하게 결합된다. 분리된 상태 모델은 선형 정적 해석에서 시작해 비선형, 동적, thermal coupling으로 확장하기 쉽다.
 
@@ -57,12 +62,25 @@ solution과 test command를 명시한 직접 MSBuild 프로젝트도 검증할
 
 **트레이드오프**: 단일 기능만 구현할 때는 adapter가 다소 장황해 보일 수 있다. Row-major dense storage와 CSR sparse storage를 따로 유지해야 하지만 backend 의존성과 dense/sparse 의미가 core 모델에 섞이지 않는다.
 
-### ADR-007: Analysis 실행 흐름은 Template Method로 고정한다
-**결정**: `Analysis::run()`은 공통 lifecycle을 고정한다. 선형 정적 V0의 순서는 `parse input -> initialize Domain -> build AnalysisModel -> build DOF map/sparse pattern -> assemble stiffness -> partition constraints -> factorize Kff -> assemble load -> form effective RHS -> substitute -> reconstruct displacement -> recover results -> write HDF5`다. 강성행렬 factorization은 하중벡터 조립보다 먼저 수행하고, factorization과 substitution을 하나의 불투명한 solve 호출로 합치지 않는다.
+### ADR-007: Analysis base는 최소 실행 계약만 제공한다
+**결정**: `Analysis` base는 virtual `Analysis::Run(const AnalysisRequest&)`만 제공하고
+linear-static-specific protected hook을 정의하지 않는다. 승인된 선형 정적 순서인
+`parse input -> initialize Domain -> build AnalysisModel -> build DOF map/sparse pattern ->
+assemble stiffness -> partition constraints -> factorize Kff -> assemble load -> form effective
+RHS -> substitute -> reconstruct displacement -> recover results -> write HDF5`는
+`LinearStaticAnalysis::Run()`의 private lifecycle로 유지한다. 강성행렬 factorization은
+하중벡터 조립보다 먼저 수행하고 factorization과 substitution을 하나의 불투명한 solve
+호출로 합치지 않는다.
 
-**이유**: 해석 procedure가 늘어나도 공통 실행 순서가 유지되어야 검증, logging, result writing, failure classification이 일관된다. Factorization과 substitution을 분리하면 동일 강성행렬에 여러 RHS를 적용할 수 있고 각 실패 단계를 구조화된 diagnostic으로 분류할 수 있다.
+**이유**: 현재 여덟 단계는 linear static equation, state와 failure taxonomy에 특화되어
+있다. 최소 base contract는 이 순서의 검증 가능성을 보존하면서 승인되지 않은 dynamic,
+eigenvalue 또는 nonlinear procedure에 같은 protected hook과 사용하지 않는 state를
+강제하지 않는다. Factorization과 substitution 분리는 동일 강성행렬에 여러 RHS를 적용할
+수 있고 각 실패 단계를 구조화된 diagnostic으로 분류하게 한다.
 
-**트레이드오프**: 특수 해석 절차가 공통 흐름에 맞지 않는 경우 hook point가 필요하다. 초기에는 선형 정적 해석을 기준으로 최소 hook만 둔다.
+**트레이드오프**: Procedure 사이의 lifecycle code는 base Template Method로 자동 재사용되지
+않는다. 두 번째 procedure가 승인되면 실제로 같은 단계만 focused collaborator로 추출하되,
+linear-static hook 사이에 조건문으로 새 physics를 삽입하지 않는다.
 
 ### ADR-008: Sparse assembly는 deterministic COO-to-CSR 경로로 시작한다
 **결정**: 초기 assembly는 element-local contribution을 COO triplet으로 수집한 뒤 CSR로 finalize한다. MKL PARDISO backend는 CSR input contract를 받는다.
@@ -111,6 +129,9 @@ solution과 test command를 명시한 직접 MSBuild 프로젝트도 검증할
 
 ### ADR-014: B33 cantilever baseline은 component-scale 혼합 tolerance로 비교한다
 
+**상태**: Artifact inventory와 component mapping의 역사적 결정은 유지하며 tolerance와
+node-station projection 결정은 ADR-022로 대체됨.
+
 **결정**: `reference/cantilever beam/`의 B33 input, displacement CSV, reaction CSV, elemental-force CSV를 V0 reference baseline으로 사용한다. Displacement와 reaction은 node identity로, section resultant는 CSV node station과 정규화된 FESA element endpoint로 비교한다. 같은 model, step/frame, quantity, component에 대해 `reference_scale = max(abs(reference rows))`를 계산하고 모든 row에 `absolute_floor + 1e-6 * reference_scale`을 적용한다. SI displacement/rotation absolute floor는 `1e-9`, force/moment floor는 `1e-3`이다. Beam stress는 출력하되 Abaqus stress reference comparison은 N/A다.
 
 **이유**: 자유단 moment처럼 이론적으로 0인 값에는 행별 상대오차가 정의되지 않으며 Abaqus 결과에 작은 수치 잔차가 남을 수 있다. Component scale을 사용하면 전체 물리량 크기에 비해 작은 잔차를 허용하면서 각 row를 결정적으로 판정할 수 있다.
@@ -211,3 +232,117 @@ nonfinite required row는 tolerance 전에 실패한다.
 대신 비교 대상과 source-row/component mapping, tolerance, artifact immutability는 유지해
 false match와 결과 보정을 방지한다. 더 강한 provenance가 필요한 기능은 해당 요구조건이
 명시적으로 추가할 수 있으나 프로젝트 기본 gate로 자동 승격하지 않는다.
+
+### ADR-020: MITC4 displacement reference는 고정 절대오차로 판정한다
+
+**상태**: S4 artifact 선택과 U blocking/UR warning-only 분류는 유지하며 고정 절대오차
+결정은 ADR-022로 대체됨.
+
+**결정**: Full-integration FESA-MITC4의 sole S4 reference comparison은 matched global
+`U1/U2/U3` row에 고정 절대오차 `1.0e-5`를 적용해 pass/fail을 판정한다.
+`UR1/UR2/UR3`도 고정 절대오차 `1.0e-5`로 비교하지만 초과는 deterministic warning만
+생성한다. Reference scale은 보고용 통계로 남을 수 있으나 MITC4 row tolerance 또는
+normalized error의 분모를 결정하지 않는다. ADR-014의 B33 component-scale 혼합
+tolerance는 변경하지 않는다.
+
+**이유**: Abaqus S4는 FESA-MITC4와 동일한 요소 정식화가 아니므로 B33에서 승인한 매우
+엄격한 component-scale tolerance를 그대로 재사용하는 것은 독립 솔버의 observable
+displacement 검증 목적에 맞지 않는다. 고정 절대오차는 현재 사용자 단위계에서 승인된
+차이를 명시적으로 제한하면서 zero clamp나 row omission 없이 모든 displacement row를
+동일하게 판정한다.
+
+**트레이드오프**: Model scale이 크게 달라지면 고정 절대오차의 상대적 엄격도가 달라질 수
+있다. 따라서 이 값은 현재 승인된 MITC4 S4 case의 기능 완료 기준이며 개발 완료 후
+별도 reference-verification evidence와 함께 재점검한다.
+
+### ADR-021: Semantic definition과 runtime solver contract를 분리한다
+
+**결정**: Domain-owned semantic definition과 analysis-time numerical object를 다음
+dependency 방향으로 분리한다.
+
+```text
+Domain owns ElementDefinition / ElementProperty / Material / StepDefinition
+AnalysisModel is a non-owning stable-index view into Domain
+ElementFactory creates runtime Element candidates from compatible definitions
+DofManager -> ElementDofLayout
+SparseAssembler -> ElementStiffnessContribution
+ResultRecovery -> ElementResultBundle
+LoadAssembler -> ordered LoadContribution
+EssentialConstraintPolicy -> ConstraintDefinition
+Analysis <- LinearStaticAnalysis
+```
+
+`Domain`의 polymorphic semantic collection은 `std::unique_ptr` 단독 ownership과 stable
+vector position을 사용한다. `ElementDefinition`은 source identity, connectivity와
+property/material identity를 제공하고 runtime `Element`는 DOF layout, stiffness와 result
+recovery를 제공한다. `ElementFactory`가 definition/property/material compatibility를
+중앙에서 검사하며 unknown 또는 incompatible 조합은 fail-closed diagnostic으로 거부한다.
+Consumer는 B33/MITC4 concrete type branch를 분산시키지 않고 runtime contract를 사용한다.
+Linear-static candidate가 runtime `Element`를 `std::unique_ptr`로 소유하고 consumer는 그
+수명에 한정된 non-owning view만 사용한다.
+
+`Material` base에는 identity, source location과 lifetime 이외의 future capability를
+추가하지 않는다. 현재 isotropic linear elasticity가 실제로 요구하는 data는 concrete
+material에 둔다. Density, anisotropy, plastic state, temperature와 rate dependency는
+optional field 또는 no-op virtual method로 미리 할당하지 않는다. `ElementProperty`도
+현재 beam/shell 의미를 각 concrete type에 둔다.
+
+`Load`는 ordered `LoadContribution`을 생성하고 global full-DOF accumulation은
+`LoadAssembler`가 단독 소유한다. `BoundaryCondition`은 `ConstraintDefinition`을 생성하고
+`EssentialConstraintPolicy`가 prescribed-displacement elimination과 reconstruction을
+소유한다. Distributed/body load와 MPC/penalty/Lagrange-multiplier enforcement는 이번
+결정으로 구현된 기능이 아니다.
+
+Abaqus Domain mapper, result recovery와 HDF5 writer는 기존 public facade를 유지하면서
+각각 topology/material-property/step-final-assembly, equilibrium/beam/shell/atomic-state,
+RAII/model-result-dataset/self-check/atomic-finalization 책임으로 private implementation을
+나눈다.
+
+**이유**: Semantic identity와 runtime kernel을 같은 concrete record에 두면 DofManager,
+SparseAssembler, ResultRecovery, parser와 output이 B33/MITC4 storage를 함께 알아야 한다.
+Definition/factory/runtime contract와 contribution/policy 경계를 분리하면 stable source
+identity와 deterministic reduction owner를 유지하면서 실제 두 element 구현을 공통
+consumer로 연결할 수 있다. Focused facade 분할은 외부 계약을 바꾸지 않고 큰 translation
+unit의 서로 다른 failure-atomicity 책임을 검토 가능하게 한다.
+
+**트레이드오프**: Base object, factory와 contribution record가 늘고 checked compatibility에
+한 단계의 indirection이 생긴다. 대신 `std::shared_ptr`, speculative `Clone()`, global
+registry와 future-only material/analysis capability는 도입하지 않는다. B33/MITC4의 승인된
+formulation, 연산·reduction 순서, sign, units, coordinates와 result identity가 이
+리팩터링보다 우선하며 HDF5 schema, reference artifact와 ADR-014/ADR-020 tolerance는
+변경하지 않는다. MITC3, solid, dynamic과 plastic behavior는 별도 feature gate 전까지
+구현된 것으로 간주하지 않는다.
+
+### ADR-022: 외부 reference comparison은 공통 family-scale 정책을 사용한다
+
+운영 상수, 검증 순서, report schema와 변경 관리는 `docs/TOLERANCE.md`를 따른다.
+
+**결정**: B33, MITC4와 이후 기능의 외부 reference comparison은 동일한 무차원 수치
+정책을 사용한다. 같은 model/case, step/frame, logical quantity, unit dimension, coordinate
+system과 blocking behavior의 component family에 대해 reference-only scale
+`S=max(abs(reference))`를 계산한다. `abs(reference)<=0.01*S`인 행은
+`abs(fesa-reference)<=0.01*S`, 나머지 행은 `abs(fesa-reference)/abs(reference)<=0.05`를
+만족해야 한다. 모든 행과 함께 family scale-relative RMS `RMS(error)/S<=0.01`도
+통과해야 한다. 독립 absolute-error gate는 두지 않으며 `S=0`이면 모든 FESA 값도 exact
+zero일 때만 통과한다.
+
+B33 family는 translation, rotation, reaction force, reaction moment, section force와
+section moment로 구분한다. 2026-08-18에 다시 생성된 elemental-force CSV의
+`Element Label`과 `Node Label`을 사용해 `(instance, element, endpoint node, component)`를
+HDF5 endpoint에 직접 대응하고 node-station collapse나 평균을 사용하지 않는다. MITC4는
+translation `U1/U2/U3` family를 blocking으로, rotation `UR1/UR2/UR3` family를
+warning-only로 유지한다.
+
+**이유**: Feature별 absolute floor와 fixed absolute tolerance는 모델 단위와 크기에 따라
+서로 다른 엄격도를 만들었다. 개별 component scale은 물리적으로 zero-like인 MITC4
+`U1/U2` residue를 자체 scale로 만들어 잘못 실패시킨다. 동일 차원의 family scale,
+near-zero 대체 분기와 relative RMS를 결합하면 zero row의 불안정한 raw relative error를
+피하면서 모든 행과 전체 오차 수준을 함께 검사할 수 있다. B33의 element-endpoint identity는
+reference와 HDF5가 제공하는 실제 source identity를 보존한다.
+
+**트레이드오프**: Family 최대값이 작은 component의 허용폭을 결정하므로 행별 relative
+gate와 family RMS gate를 모두 유지해야 한다. Reference family 전체가 zero이면 별도
+절대 scale이 없으므로 exact-zero만 허용하는 엄격한 fail-closed 정책이 된다. 기존 B33과
+MITC4 verification report 및 comparator schema는 새 정책에 맞춰 다시 생성해야 하지만
+reference artifact, HDF5 schema, element formulation과 blocking/warning 분류는 변경하지
+않는다.

docs/ARCHITECTURE.md

@@ -32,7 +32,7 @@ src/
     assembly/             # deterministic stiffness/load assembly, ParallelFor adapter
     constraints/          # essential-constraint elimination and reconstruction
     core/                 # source identity, status, diagnostics
-    elements/             # V0 EulerBeam3D kernel and recovery
+    elements/             # current B33/MITC4 kernels and recovery
     fem/                  # DOF/equation numbering and sparse pattern
     io/
       abaqus/             # .inp syntax reader and semantic Domain mapper
@@ -62,9 +62,11 @@ scripts/
 phases/                   # Optional generated phase plans
 ```
 
-`materials/`, nonlinear/dynamic analysis, MPC/penalty policies, general element factories,
-history output과 production validation module은 장기 확장 경계이지 현재 구현된 module이
-아니다. 새 디렉토리와 추상 계층은 승인된 기능이 실제로 필요로 할 때 추가한다.
+`materials/`, `properties/`, `loads/`와 checked `ElementFactory`는 승인된 C++ modular
+refactoring의 target boundary이며 후속 implementation Step에서 추가한다. 이
+documentation-only Step 시점에는 현재 구현 디렉토리로 표시하지 않는다. Density,
+plasticity, anisotropy, nonlinear/dynamic analysis, MPC/penalty, history output과 production
+validation module은 계속 장기 확장 경계이며 별도 승인 기능이 필요하다.
 
 ## Harness Execution Layer
 
@@ -109,16 +111,86 @@ CMake source에 기록하지 말고 config package와 imported target metadata
 ## 모듈 경계
 - `core`는 외부 라이브러리에 의존하지 않는다.
 - `io/abaqus`는 syntax와 semantic mapping만 담당하고 해석 알고리즘을 알지 않는다.
-- `model`은 Abaqus keyword 문자열이 아니라 solver semantic model을 가진다.
+- `model`은 Abaqus keyword 문자열이 아니라 solver semantic model을 가지며 `Domain`과
+  non-owning `AnalysisModel`의 수명 경계를 소유한다.
+- `materials`와 `properties`는 Domain이 소유하는 semantic identity와 현재 승인된
+  isotropic elasticity 및 beam/shell property data만 제공한다.
+- `elements`는 Domain-owned `ElementDefinition`, runtime numerical `Element`, checked
+  `ElementFactory`와 element-local stiffness/recovery contract를 제공한다.
 - `fem`의 `DofManager`는 DOF, equation ordering, scatter와 sparse pattern을 소유한다.
-- `elements`는 local/global stiffness, transformation, optional load kernel과 recovery를 제공한다. V0 material/section은 concrete Domain record다.
-- `assembly`는 element-local contribution과 full nodal load를 stable full-DOF space에 조립한다.
-- `constraints`는 V0 essential BC elimination과 full/reduced vector 변환을 담당한다. MPC와 penalty는 현재 범위가 아니다.
+- `assembly`는 runtime element contribution과 ordered load contribution을 stable full-DOF
+  space에 조립한다. Contribution producer는 global storage를 직접 갱신하지 않는다.
+- `loads`는 semantic target과 magnitude를 소유하고 ordered `LoadContribution`을 생성한다.
+- `constraints`는 `ConstraintDefinition` 생성과 V0 essential BC elimination 및 full/reduced
+  vector 변환을 분리한다. MPC와 penalty는 현재 범위가 아니다.
 - `solvers`는 `LinearSolver` 뒤에 MKL PARDISO 세부 구현을 감춘다. TBB는 `assembly/ParallelFor`, HDF5는 `results/ResultsWriter` 경계 뒤에 각각 격리된다.
 - `analysis`는 step/history data를 받아 procedure를 실행하고 solver backend와 result writer를 조율한다.
-- `results`는 full residual과 beam rows를 복구하고 backend-neutral writer contract를 제공한다. HDF5 schema 구현은 `io/hdf5`가 담당한다.
+- `results`는 full residual과 B33/MITC4 rows를 복구하고 backend-neutral writer contract를
+  제공한다. HDF5 schema 구현은 `io/hdf5`가 담당한다.
 - test helper는 production parser/solver 내부 상태를 우회하지 않는다.
 
+## 승인된 리팩터링 dependency와 ownership
+
+다음 graph는 C++ object-oriented modular refactoring의 구현 방향을 고정한다. `owns`는
+단독 수명 소유권을, 나머지 화살표는 왼쪽 consumer가 오른쪽 contract를 사용한다는
+뜻이다. 이 graph는 기존 B33/MITC4 물리 기능을 늘리지 않는다.
+
+```text
+Domain owns ElementDefinition / ElementProperty / Material / StepDefinition
+AnalysisModel is a non-owning stable-index view into Domain
+ElementFactory creates runtime Element candidates from compatible definitions
+DofManager -> ElementDofLayout
+SparseAssembler -> ElementStiffnessContribution
+ResultRecovery -> ElementResultBundle
+LoadAssembler -> ordered LoadContribution
+EssentialConstraintPolicy -> ConstraintDefinition
+Analysis <- LinearStaticAnalysis
+```
+
+승인된 B33/MITC4 formulation, operation/reduction order, sign, units, coordinates와 row
+identity는 이 ownership 리팩터링보다 우선한다. HDF5 schema, reference artifact와
+ADR-014/ADR-020 tolerance도 변경하지 않는다.
+
+`Domain`은 `ElementDefinition`, `ElementProperty`, `Material`과 `StepDefinition` base
+object를 `std::unique_ptr`로 단독 소유하고 const access를 제공한다. Collection position은
+기존 stable `EntityIndex` 의미를 유지한다. `AnalysisModel`과 solver consumer는 ownership을
+가져오지 않고 Domain 수명 안에서 stable index 또는 const reference만 사용한다.
+`std::shared_ptr`, speculative `Clone()`과 global registry는 이 계약에 포함되지 않는다.
+
+`ElementDefinition`은 source identity, source element type, connectivity와
+property/material identity를 보존하는 semantic object다. Runtime `Element`는 active
+`ElementDofLayout`, `ElementStiffnessContribution`과 `ElementResultBundle`을 제공하는
+numerical kernel이다. `ElementFactory`만 compatible definition/property/material 조합을
+검사해 runtime candidate를 만들며 unknown 또는 incompatible 조합을 기존
+`Status`/`Result<T>` diagnostic으로 fail-closed 처리한다. `DofManager`,
+`SparseAssembler`와 `ResultRecovery`는 B33/MITC4 concrete storage가 아니라 위 runtime
+contract를 소비한다. Linear-static candidate는 runtime `Element`를
+`std::vector<std::unique_ptr<Element>>`로 소유하고, consumer에는 그 owner 수명 안에서만
+유효한 non-owning `ElementView`를 제공한다.
+
+`Material` base는 identity, source location과 수명 의미만 공유한다. 현재 concrete
+isotropic linear elasticity에 필요한 capability만 사용하며 density, anisotropy, plastic
+state, temperature 또는 rate dependency를 optional field나 no-op method로 미리 추가하지
+않는다. `ElementProperty`도 beam/shell이 실제 사용하는 data만 각 concrete type에 둔다.
+
+`Load`는 semantic target과 magnitude를 소유하고 source order가 보존된
+`LoadContribution`을 생성한다. Global vector의 deterministic accumulation은
+`LoadAssembler`만 수행한다. `BoundaryCondition`은 `ConstraintDefinition`을 생성하고
+`EssentialConstraintPolicy`가 prescribed displacement의 stable elimination과
+reconstruction을 수행한다. Future distributed/body load와 MPC enforcement는 구현된
+기능이 아니며 별도 승인 계약 없이 이 경계에 branch나 optional state를 추가하지 않는다.
+
+책임이 큰 facade는 외부 계약을 유지한 채 private implementation만 다음 owner로 나눈다.
+
+- Abaqus Domain mapping: topology, material/property, step/load/boundary mapping과 final
+  Domain assembly
+- Result recovery: global equilibrium, beam recovery, shell recovery와 atomic state commit
+- HDF5 writing: RAII/primitives, model datasets, result datasets, self-check와 atomic
+  finalization
+
+이 분할은 parser diagnostic, `ResultsWriter` boundary, HDF5 schema 또는 final-file
+atomicity를 변경하지 않는다.
+
 ## V0 입력 경계
 
 V0 parser는 keyword와 parameter를 case-insensitive하게 해석하되 source label의 원문을
@@ -174,14 +246,23 @@ CLI pipeline에서는 이 kernel을 호출하지 않는다. Stiffness와 recover
 
 ```text
 Domain
-├── owns nodes, B33 elements, materials, beam sections, sets
-├── owns boundary conditions, nodal loads, one static step
+├── owns nodes, sets and source identity
+├── owns unique_ptr<ElementDefinition / ElementProperty / Material>
+├── owns StepDefinition with Load / BoundaryCondition definitions
 └── owns source path/identity and mapping warnings
 
 AnalysisModel
-├── non-owning view into Domain
-├── stable active element/BC/load indices
-└── reachable material/section indices
+├── non-owning stable-index view into Domain
+├── stable active element/BC/load definition indices
+└── reachable material/property indices
+
+ElementFactory
+└── creates checked runtime Element candidates
+
+Element
+├── exposes ElementDofLayout
+├── produces ElementStiffnessContribution
+└── recovers ElementResultBundle
 
 DofManager
 ├── owns node x [UX,UY,UZ,URX,URY,URZ] full-DOF numbering
@@ -205,12 +286,15 @@ Matrix
 SparseMatrix
 ```
 
-Nonlinear/static, dynamic, frequency, heat-transfer procedure와 general element/material/load
-base hierarchy는 이 구조 위의 가능한 확장 방향일 뿐 현재 public API가 아니다. 사용 사례가
-승인되기 전에 V0 concrete record를 speculative hierarchy로 감싸지 않는다.
+위 abstract boundary는 현재 B33/MITC4, isotropic linear elasticity, beam/shell property,
+concentrated nodal load, prescribed displacement와 linear static procedure를 연결하는 데
+필요한 최소 계약이다. MITC3, solid, dynamic, frequency, heat-transfer와 plastic behavior는
+구현된 기능이 아니며 승인된 사용 사례 전에 future-only method나 state를 base에 추가하지
+않는다.
 
 ## 상태 관리
-- `Domain`은 입력 파일에서 만들어진 전체 모델 정의를 소유한다. 파싱 이후에는 가능한 한 불변으로 취급한다.
+- `Domain`은 입력 파일에서 만들어진 전체 모델 정의를 `std::unique_ptr`로 단독 소유한다.
+  파싱 이후에는 가능한 한 불변으로 취급하고 stable collection index를 바꾸지 않는다.
 - `LinearStaticAnalysis`가 `Domain`을 소유하고, 그 뒤에 `AnalysisModel`, `DofManager`, `AnalysisState`, stiffness/RHS를 순서대로 만든다. 재사용 시에는 역순으로 해제하여 이전 Domain을 가리키는 view를 남기지 않는다.
 - `AnalysisModel`은 현재 step에서 활성화되는 해석 객체들의 실행 view이다. `Domain`을 복사하지 않으므로 Domain이 반드시 더 오래 살아야 한다.
 - `DofManager`는 자유도와 방정식 번호를 전담한다. `Node` 또는 `Element` 내부에 equation id를 분산 저장하지 않는다.
@@ -223,15 +307,16 @@ Abaqus input file
 -> syntax parse and semantic mapping
 -> immutable Domain 생성
 -> 단일 step AnalysisModel view 생성
--> DofManager DOF/scatter map/sparse pattern 생성
--> element stiffness 계산과 deterministic COO-to-CSR 조립
--> free/constrained partition 생성
+-> ElementFactory가 compatible definition에서 runtime Element candidate 생성
+-> DofManager가 ElementDofLayout으로 DOF/scatter map/sparse pattern 생성
+-> ElementStiffnessContribution의 deterministic COO-to-CSR 조립
+-> ConstraintDefinition의 stable essential-constraint partition 생성
 -> LinearSolver::factorize(Kff)
--> full nodal load vector 조립
+-> ordered LoadContribution의 full nodal load vector 조립
 -> effective RHS = Ff - Kfc * dc
 -> LinearSolver::solve(rhs, df) substitution
 -> full displacement 복구
--> full residual/reaction = K*d - F 및 element result 복구
+-> full residual/reaction = K*d - F 및 ElementResultBundle 복구
 -> ResultsWriter로 results.h5 atomic finalization
 ```
 
@@ -240,28 +325,37 @@ Abaqus input file
 reaction이고 free component는 equilibrium residual evidence로 full-index vector에 남긴다.
 
 ## 해석 실행 흐름
-`Analysis::run()`은 Template Method로 다음 여덟 hook의 순서와 fail-fast 경계를 고정한다.
 
-| 순서 | Hook | 주요 작업과 생성되는 소유 객체 | 순서/실패 불변식 |
+`Analysis` base는 procedure-specific protected hook을 정의하지 않고 최소 실행 계약인
+`Analysis::Run(const AnalysisRequest&)`만 제공한다. 다음 여덟 단계의 순서와 fail-fast
+경계는 `LinearStaticAnalysis::Run()`의 private lifecycle이며 다른 procedure에 강제되지
+않는다.
+
+| 순서 | Private stage | 주요 작업과 생성되는 소유 객체 | 순서/실패 불변식 |
 | --- | --- | --- | --- |
-| 1 | `initialize(request)` | `.inp` syntax read, semantic map, owned immutable `Domain`, sorted warnings | 이전 run의 dependent object를 역순으로 제거하고 parse/map 실패를 input category로 반환한다. |
-| 2 | `buildAnalysisModel()` | non-owning `AnalysisModel` view | Domain을 복사하지 않으며 Domain lifetime 안에서만 사용한다. |
-| 3 | `buildDofMapAndSparsePattern()` | `DofManager`, zero-initialized `AnalysisState` | Stable full/free/constrained numbering과 structural pattern을 한 소유자에게 둔다. |
-| 4 | `assembleAndPartitionStiffness()` | full CSR K와 `Kff/Kfc/Kcf/Kcc` | Element-local buffer를 deterministic하게 reduce하고 structural zero와 stable order를 보존한다. |
-| 5 | `factorize()` | retained `Kff` factorization | 모든 load assembly보다 먼저 호출한다. Valid fully constrained model의 `0 x 0 Kff`는 trivial success다. |
-| 6 | `assembleLoadsAndEffectiveRhs()` | full F와 `Ff-Kfc*dc` | Semantic load source order와 finite sum을 보존하며 solver를 호출하지 않는다. |
-| 7 | `substituteAndReconstruct()` | free solution과 full displacement | Factorization을 재수행하지 않고 substitution한 뒤 prescribed value를 stable order로 복구한다. |
-| 8 | `recoverAndWriteResults()` | full residual/reaction, beam rows, final HDF5 | Recovery candidate를 원자적으로 commit하고 writer 성공 뒤에만 최종 output을 교체한다. |
+| 1 | `Initialize(request)` | `.inp` syntax read, semantic map, owned immutable `Domain`, sorted warnings | 이전 run의 dependent object를 역순으로 제거하고 parse/map 실패를 input category로 반환한다. |
+| 2 | `BuildAnalysisModel()` | non-owning `AnalysisModel` view | Domain을 복사하지 않으며 Domain lifetime 안에서만 사용한다. |
+| 3 | `BuildDofMapAndSparsePattern()` | `DofManager`, zero-initialized `AnalysisState` | Stable full/free/constrained numbering과 structural pattern을 한 소유자에게 둔다. |
+| 4 | `AssembleAndPartitionStiffness()` | full CSR K와 `Kff/Kfc/Kcf/Kcc` | Element-local buffer를 deterministic하게 reduce하고 structural zero와 stable order를 보존한다. |
+| 5 | `Factorize()` | retained `Kff` factorization | 모든 load assembly보다 먼저 호출한다. Valid fully constrained model의 `0 x 0 Kff`는 trivial success다. |
+| 6 | `AssembleLoadsAndEffectiveRhs()` | full F와 `Ff-Kfc*dc` | Semantic load source order와 finite sum을 보존하며 solver를 호출하지 않는다. |
+| 7 | `SubstituteAndReconstruct()` | free solution과 full displacement | Factorization을 재수행하지 않고 substitution한 뒤 prescribed value를 stable order로 복구한다. |
+| 8 | `RecoverAndWriteResults()` | full residual/reaction, B33/MITC4 rows, final HDF5 | Recovery candidate를 원자적으로 commit하고 writer 성공 뒤에만 최종 output을 교체한다. |
 
 비선형 정적 및 동적 해석은 V0 범위가 아니며 별도 ADR과 formulation을 승인한 뒤 이
-lifecycle과 state/equation 계약을 확장한다. 기존 hook 사이에 조용히 반복·증분·시간 적분
-동작을 삽입하지 않는다.
+lifecycle과 state/equation 계약을 별도 procedure에 정의한다. LinearStaticAnalysis의
+private stage 사이에 조용히 반복·증분·시간 적분 동작을 삽입하지 않는다.
 
 ## 설계 패턴
-- Strategy/Adapter Pattern: 현재 교체 가능한 public 경계는 `LinearSolver`, `ParallelFor`, `ResultsWriter`다. Vendor API는 concrete adapter implementation 안에만 둔다.
-- Template Method Pattern: `Analysis::run()`은 공통 실행 흐름을 고정하고 세부 단계는 procedure별로 재정의한다.
+- Strategy/Adapter Pattern: `Analysis::Run(const AnalysisRequest&)`, `LinearSolver`,
+  `ParallelFor`, `ResultsWriter`가 현재 승인된 public 실행/backend 경계다. Vendor API는
+  concrete adapter implementation 안에만 둔다.
+- Procedure-owned lifecycle: `Analysis`는 protected Template Method hook을 공유하지 않고
+  `LinearStaticAnalysis`가 승인된 여덟 단계 lifecycle을 private하게 소유한다.
 - Syntax/Semantic separation: `AbaqusInputReader`는 syntax record를 만들고 `AbaqusDomainMapper`가 승인된 keyword 의미를 concrete Domain record로 변환한다.
-- Runtime Polymorphism: V0에서는 backend 경계에만 사용한다. 요소/재료/하중 base hierarchy와 factory/registry는 두 번째 실제 구현이 필요해질 때 trade-off를 다시 결정한다.
+- Runtime Polymorphism: backend와 승인된 analysis/element/material/property/load/boundary
+  경계에만 사용한다. Factory compatibility는 중앙에서 fail-closed로 검사하며 global
+  registry 또는 future-only capability를 추가하지 않는다.
 - RAII: MKL handle, HDF5 file/dataset, temporary solver workspace의 수명과 오류 처리를 wrapper에 묶는다.
 
 ## Sparse Matrix Policy
@@ -325,14 +419,19 @@ Schema requirements:
   `[N,T,My,Mz]`, generalized strain/resultant는 두 Gauss point에 기록한다.
 - General beam section stress는 section point의 axial `S11`만 복구한다. Section point가
   없으면 centroid `(0,0)`을 `source=fesa-default`로 기록한다.
-- 승인된 `reference/cantilever beam/cantilever beam elemental forces.csv`는 node station
-  기준 `SF1/SM1/SM2/SM3`을 제공한다. FESA endpoint를 동일한 section-cut 부호로
-  정규화하고 interior node의 두 endpoint가 tolerance 안에서 일치하는지 먼저 확인한 뒤
-  `SF1 -> N`, `SM1 -> My`, `SM2 -> Mz`, `SM3 -> T`로 비교한다.
-- Reference tolerance는 같은 model, step/frame, quantity, component의 Abaqus rows에서
-  `reference_scale = max(abs(reference_value))`를 구하고 각 row에
-  `absolute_floor + 1e-6 * reference_scale`을 적용한다. SI displacement/rotation floor는
-  `1e-9`, force/moment floor는 `1e-3`이다.
+- 승인된 `reference/cantilever beam/cantilever beam elemental forces.csv`는 각 B33
+  element의 두 endpoint에 `Element Label`, `Node Label`과 `SF1/SM1/SM2/SM3`을 제공한다.
+  `(instance, element label, endpoint node label, component)` identity로 HDF5
+  `[element,endpoint,N/T/My/Mz]`에 직접 대응하고 `SF1 -> N`, `SM1 -> My`, `SM2 -> Mz`,
+  `SM3 -> T`로 비교한다. External reference comparison은 node-station collapse나 평균을
+  사용하지 않는다.
+- Reference tolerance는 같은 model/case, step/frame, logical quantity, unit dimension,
+  coordinate system과 blocking behavior의 component family에서 Abaqus-only scale
+  `S=max(abs(reference))`를 구한다. `abs(reference)<=0.01*S`인 행은 absolute error
+  `<=0.01*S`, 그 외 행은 relative error `<=0.05`로 판정하며 family scale-relative RMS
+  `RMS(error)/S<=0.01`도 통과해야 한다. 독립 absolute-error gate는 사용하지 않고
+  zero-scale family는 FESA도 exact zero일 때만 통과한다. 전체 판정 과정과 report/change
+  management contract는 `docs/TOLERANCE.md`를 따른다.
 - Beam stress는 HDF5 schema와 unit/analytical test로 검증하지만 Abaqus reference
   comparison은 N/A다.
 
@@ -374,6 +473,9 @@ rotational stiffness block의 positive minimum diagonal에 `1e-3`을 곱한 고
 stabilization만 둔다. Drilling calibration, artificial-energy policy와 별도 drilling result
 dataset은 이 기능 범위가 아니다.
 
-Reference comparison은 `reference/shell/` S4와 `reference/shellR/` S4R의 기존 input 및
-displacement CSV를 사용한다. Global `U1/U2/U3`만 blocking이고 B33과 같은 component-scale
-mixed tolerance를 적용한다. `UR1/UR2/UR3`은 같은 기준의 warning-only evidence다.
+Full-integration FESA-MITC4의 reference comparison은 `reference/shell/` S4의 기존 input 및
+displacement CSV만 사용한다. Global `U1/U2/U3` translation family만 blocking이고
+`UR1/UR2/UR3` rotation family는 warning-only evidence다. 두 family 모두 ADR-022의 공통
+family-scale row/RMS 정책을 사용한다. S4R은
+같은 kernel을 선택하는 source mapping과 metadata를 unit/integration tests로 검증하며
+`reference/shellR/` artifact는 acceptance comparison에 포함하지 않는다.

docs/CODINGSTYLE.md

@@ -0,0 +1,256 @@
+# FESA C++ Coding Style
+
+## 목적
+
+이 문서는 FESA production 및 test C++의 코드 스타일, 객체 설계, 문서화와 검증 규칙을
+정의하는 project-local source of truth다. 새 C++를 작성하거나 기존 C++를 리팩터링하는
+사람과 Implementation Agent는 작업 전에 이 문서를 읽어야 한다.
+
+[Google C++ Style Guide](https://google.github.io/styleguide/cppguide.html)를 baseline으로
+사용한다. 이 문서, `AGENTS.md`, 승인된 feature contract와 architecture/ADR이 Google guide의
+일반 규칙보다 우선한다.
+
+## 적용 범위와 우선순위
+
+규칙 충돌 시 다음 순서로 해석한다.
+
+1. 승인된 feature requirements, formulation, numerical-review, I/O와 reference contract
+2. `AGENTS.md`, `docs/ARCHITECTURE.md`와 `docs/ADR.md`
+3. 이 문서의 FESA-specific rule과 exception
+4. Google C++ Style Guide
+5. 기존 local style
+
+새 코드는 이 문서를 즉시 준수한다. 기존 코드는 승인된 refactoring plan의 module slice
+단위로 전환한다. 요청 범위 밖의 file을 style-only 이유로 함께 수정하지 않는다.
+
+## Language와 Toolchain
+
+- Production language는 C++17 이상이며 MSVC x64를 지원해야 한다.
+- 승인된 build 기준은 CMake, Visual Studio generator와 Debug configuration이다.
+- C++ compiler extension에 의존하지 않는다.
+- MKL, TBB, HDF5와 Win32 type은 public solver core header에 노출하지 않는다.
+- Standard library와 RAII를 manual lifetime management보다 우선한다.
+- Google guide가 현재 권장하는 C++20 language target은 FESA의 C++17 contract를 바꾸지
+  않는다.
+
+## File 이름과 Header
+
+- File 이름은 소문자 snake_case를 사용한다.
+- Production 및 test source extension은 기존 FESA/CMake 관례인 `.cpp`를 유지한다.
+- Header extension은 `.h`를 사용한다. 기존 `.hpp`는 승인된 migration slice에서 `.h`로
+  바꾼다.
+- Header는 self-contained여야 하며 include consumer의 transitive include에 의존하지
+  않는다.
+- Header는 `#pragma once` 대신 full repository path 기반 include guard를 사용한다.
+
+예:
+
+```cpp
+#ifndef FESA_MATH_VECTOR3_H_
+#define FESA_MATH_VECTOR3_H_
+
+namespace fesa {
+
+class Vector3 {};
+
+}  // namespace fesa
+
+#endif  // FESA_MATH_VECTOR3_H_
+```
+
+Include 순서는 다음과 같다.
+
+1. 대응하는 header
+2. C system header
+3. C++ standard library header
+4. Third-party header
+5. FESA project header
+
+각 non-empty group 사이에는 빈 줄을 두고 group 안에서는 알파벳순으로 정렬한다. 사용하는
+symbol의 declaration을 제공하는 header를 직접 include한다.
+
+## 이름 규칙
+
+| 대상 | 규칙 | 예 |
+| --- | --- | --- |
+| class, struct, enum, alias | PascalCase | `ElementProperty`, `EntityIndex` |
+| function, method, accessor | PascalCase | `ComputeStiffness()`, `NodeCount()` |
+| local variable, parameter | snake_case | `element_index`, `source_order` |
+| class data member | snake_case + trailing `_` | `youngs_modulus_` |
+| struct data member | snake_case | `source_id` |
+| compile-time/static constant | `kPascalCase` | `kNodeCount` |
+| enumerator | `kPascalCase` | `FailureCategory::kModel` |
+| namespace | snake_case | `fesa::hdf5_internal` |
+| macro | UPPER_SNAKE_CASE | `FESA_MATH_VECTOR3_H_` |
+
+Google guide는 accessor의 snake_case를 허용하지만 FESA는 사용자 승인에 따라 production
+및 test 호출부를 포함한 모든 function name에 PascalCase를 적용한다. Constructor,
+destructor와 operator 이름은 C++ language 규칙을 따른다.
+
+이름은 물리 및 수치 의미를 드러내야 한다. `value`, `data`, `handler`, `manager`처럼 문맥이
+없는 generic name을 넓은 scope에서 사용하지 않는다. Source label, internal entity index와
+equation index를 이름에서 구분한다.
+
+## Formatting
+
+- `.clang-format`의 `BasedOnStyle: Google`을 사용한다.
+- 들여쓰기는 space 2개이며 tab을 사용하지 않는다.
+- 최대 line length는 80자다. Include, guard, URL과 분할할 수 없는 contract string은 Google
+  guide의 예외를 따른다.
+- Opening brace는 declaration/control statement의 마지막 줄에 둔다.
+- Namespace body는 들여쓰지 않고 closing namespace comment를 작성한다.
+- 한 statement에 한 declaration만 둔다.
+- Variable은 가능한 가장 좁은 scope에서 선언과 동시에 초기화한다.
+- `const`와 `constexpr`를 의미가 허용하는 범위에서 사용한다.
+- `override`, `final`, `explicit`, `noexcept`와 `[[nodiscard]]`는 실제 contract를 표현할 때
+  사용한다.
+
+Formatting-only 변경과 behavior/architecture 변경은 같은 commit에 섞지 않는다.
+
+## Class와 Interface 설계
+
+- Class는 하나의 명확한 책임과 invariant를 가져야 한다.
+- Polymorphic base는 public virtual destructor를 가져야 한다.
+- Abstract interface에는 현재 concrete 구현이 공유하지 않는 future method를 추가하지
+  않는다.
+- 단독 ownership은 `std::unique_ptr`로 표현한다. 실제 shared lifetime이 없는
+  `std::shared_ptr`는 사용하지 않는다.
+- Non-owning pointer/reference의 lifetime은 Doxygen contract에 기록한다.
+- Downcast와 type switch를 주요 확장 mechanism으로 사용하지 않는다.
+- State가 없는 함수를 묶기 위한 static-only class를 만들지 않는다. Internal namespace와
+  focused module을 사용한다.
+- Base class에 optional field와 no-op method를 누적하지 않는다. Material density, plastic
+  state와 anisotropic constitutive law처럼 독립적인 의미는 별도 capability 또는 구성
+  객체로 설계한다.
+- Public header가 implementation/vendor dependency를 역으로 끌어오지 않도록 한다.
+
+Data-only record는 struct를 사용할 수 있다. Invariant, encapsulation, lifetime 또는 behavior가
+있으면 class를 사용한다.
+
+## FEM Module 책임
+
+- `model`: immutable semantic definition과 stable source/internal identity
+- `elements`: element numerical kernel, local contribution과 element recovery
+- `properties`: element property identity와 concrete section data
+- `materials`: constitutive capability와 concrete material behavior
+- `fem`: DOF/equation numbering, scatter와 sparse pattern
+- `assembly`: deterministic element/load contribution reduction
+- `constraints`: constraint definition 적용과 equation policy
+- `analysis`: procedure-specific lifecycle과 backend orchestration
+- `results`: backend-neutral recovery record와 physical result identity
+- `io`: Abaqus syntax/semantic mapping과 HDF5 schema implementation
+- `math`: backend-neutral value/storage type와 private numerical adapter
+
+한 module의 class가 다른 module의 owner 책임을 가져가지 않는다. Element가 global CSR을
+직접 쓰거나 Node가 equation ID를 저장하거나 Material이 analysis state를 임의로 소유하면
+안 된다.
+
+## 중복과 공통화
+
+같은 의미, units, coordinate, failure policy와 ownership을 가진 logic을 한 번만 구현한다.
+현재 승인된 공통화 방향은 다음과 같다.
+
+- 3D coordinate/axis/director 연산은 `Vector3` value class
+- source label/set/instance 해석은 `SourceTargetResolver`
+- full/free/constrained invariant는 `DofManager` owner validation
+- MKL size/copy helper는 private dense-BLAS adapter
+- ASCII case-insensitive comparison과 label parsing은 focused core utility
+
+두 코드 block이 비슷해 보여도 formulation sign, result location, source identity 또는
+tolerance가 다르면 공통화하지 않는다. 단 한 번 사용하는 logic을 future flexibility만을
+위해 framework로 만들지 않는다.
+
+## Error와 Ownership
+
+- Expected failure는 `Status` 또는 `Result<T>`로 반환한다.
+- Public solver API를 통해 backend exception이 그대로 새지 않게 한다.
+- Unknown type/property/material 조합은 structured diagnostic으로 fail-closed 처리한다.
+- Invalid input을 silent default, clamp, average 또는 fallback으로 숨기지 않는다.
+- Candidate를 완성하고 검증한 뒤 Domain state, AnalysisState 또는 final HDF5에 commit한다.
+- Stable ordering과 failure atomicity는 optimization option이 아니라 correctness contract다.
+
+## Doxygen
+
+Doxygen coverage는 production code에만 요구한다. Test function과 test helper에는 Doxygen를
+요구하지 않는다.
+
+Public/protected class와 function declaration은 다음 내용을 필요한 만큼 기록한다.
+
+- `@brief`: 무엇을 하는지 동사형 한 문장
+- `@param`: 이름만으로 드러나지 않는 units, coordinates, ownership 또는 valid range
+- `@return`: success value와 failure 의미
+- `@throws`: 실제로 경계를 넘어가는 exception
+- `@pre`: caller가 보장해야 하는 invariant
+- `@note`: deterministic order, lifetime 또는 backend constraint
+- `@warning`: sign, physical/numerical distinction 또는 destructive side effect
+
+예:
+
+```cpp
+/// @brief Computes the element stiffness in stable global DOF order.
+/// @return A finite symmetric contribution or a structured model failure.
+/// @note The returned matrix does not include nonphysical result terms.
+virtual Result<Matrix> ComputeStiffness() const = 0;
+```
+
+Private/internal production function은 declaration 또는 definition에 목적을 기록한다. 수식,
+sign, coordinate transform, lifetime, ordered reduction이나 failure preservation이
+비자명하면 그 이유를 설명한다. Header declaration의 사용법을 `.cpp` definition에서 그대로
+반복하지 않는다.
+
+Comment는 코드 한 줄을 한국어 또는 영어로 번역하는 방식으로 작성하지 않는다. Public API
+Doxygen는 일관된 tool output을 위해 영어를 기본으로 한다. Diagnostic message와 existing
+contract language는 현재 외부 계약을 유지한다.
+
+## Determinism과 수치 코드
+
+- Element contribution은 stable source/internal index 순서로 생성한다.
+- Parallel worker는 index-owned output만 수정한다.
+- Floating-point reduction 순서는 명시적으로 고정한다.
+- Refactoring 중 expression/reduction 순서를 편의상 바꾸지 않는다.
+- Arbitrary `max(1, ...)`, zero clamp 또는 missing-row ignore를 추가하지 않는다.
+- End action, section resultant, generalized result와 stress의 identity/sign을 구분한다.
+- Reference mapping은 row order가 아니라 승인된 source identity와 component를 사용한다.
+
+수치식을 공통화할 때는 현재 formulation과 test가 정의한 operation order 및 tolerance를
+먼저 확인한다.
+
+## Test와 변경 관리
+
+- Production C++ 변경은 관련 C++ test와 같은 patch에 있어야 한다.
+- Behavior 또는 interface 변경은 `RED -> observed failure -> minimal GREEN -> VERIFY`를
+따른다.
+- Refactoring test는 base interface 사용, ownership/lifetime, invalid combination,
+  deterministic order와 current numerical result preservation을 검증한다.
+- Focused test 뒤에 full MSVC x64 Debug build와 CTest를 실행한다.
+- B33/MITC4 output 경계를 건드린 변경은 승인된 reference comparison을 다시 실행한다.
+- Reference artifact, input path와 tolerance를 리팩터링에 맞춰 수정하지 않는다.
+- Commit은 review 가능한 module slice로 제한하고 Conventional Commits를 사용한다.
+
+## Tooling
+
+Repository가 제공하는 설정을 우선한다.
+
+```powershell
+clang-format --dry-run --Werror <changed-cpp-and-header-files>
+clang-tidy <changed-cpp-files> -- -std=c++17
+doxygen Doxyfile
+```
+
+실제 build/test command는 `.harness/config.json`이 있으면 그 설정을 우선하고, 없으면
+`AGENTS.md`와 Harness의 MSVC/CMake/CTest entry point를 따른다. 필요한 tool이 설치되지 않아
+검증을 실행할 수 없으면 성공으로 간주하지 않고 environment limitation을 보고한다.
+
+Generated Doxygen HTML과 tool cache/build output은 source control에 넣지 않는다.
+
+## Implementation Agent Checklist
+
+Implementation Agent는 C++ Step을 시작하기 전에 다음을 확인한다.
+
+- 이 문서와 feature implementation plan을 읽었다.
+- 변경할 base/concrete/module owner가 승인 설계와 일치한다.
+- 관련 test file과 RED condition이 Step에 명시되어 있다.
+- Public API naming과 production Doxygen가 이 문서에 맞는다.
+- Vendor dependency와 ownership direction이 역전되지 않는다.
+- Stable identity, numerical order, HDF5와 reference contract가 보존된다.
+- Formatting, Doxygen, focused/full MSVC/CTest acceptance command가 계획되어 있다.

docs/HARNESS.md

@@ -26,6 +26,21 @@ python scripts/execute.py <phase-name>
 python scripts/execute.py <phase-name> --push
 ```
 
+Executor가 시작하는 Codex 세션은 기본적으로 `workspace-write` sandbox를 사용한다.
+Windows native sandbox에서 MSVC compiler-id의 `cl.exe`가 정지하는 것이 재현되고 같은
+명령이 sandbox 밖에서 정상 완료되는 경우에만, 사용자 승인을 받은 격리된 clean
+worktree에서 다음처럼 해당 실행에 한정해 fallback을 지정할 수 있다.
+
+```powershell
+$env:FESA_HARNESS_CODEX_SANDBOX = "danger-full-access"
+python scripts/execute.py <phase-name>
+Remove-Item Env:FESA_HARNESS_CODEX_SANDBOX
+```
+
+허용값은 `workspace-write`와 `danger-full-access`뿐이다. 후자는 Codex Step에 workspace
+밖의 파일 접근 권한도 부여하므로 일반 기본값으로 설정하지 않는다. 어느 모드에서도
+`.codex/hooks.json`의 PreToolUse와 Stop hook은 자동으로 실행된다.
+
 ## Harness Python 검증
 
 이 저장소의 테스트와 최종 acceptance 검증은 pytest를 시스템 Python에 설치하지 않고

docs/HARNESS_WORKFLOW.md

@@ -238,12 +238,19 @@ Executor는 각 Step을 다음 형태의 독립 프로세스로 실행한다.
 ```text
 codex exec
   --json
-  --sandbox workspace-write
+  --sandbox <workspace-write|danger-full-access>
   --dangerously-bypass-hook-trust
   --cd <repository-root>
   -
 ```
 
+기본값은 `workspace-write`다. `FESA_HARNESS_CODEX_SANDBOX` 환경 변수는
+`workspace-write` 또는 `danger-full-access`만 허용한다. Windows native sandbox에서
+MSVC compiler-id의 `cl.exe` 정지가 재현되고 동일 명령이 sandbox 밖에서 통과하는
+환경에서는, 사용자 승인을 받은 격리된 clean worktree 실행에 한해서
+`danger-full-access` fallback을 사용할 수 있다. 이 override는 hook trust 또는 hook
+등록을 끄지 않으며 PreToolUse와 Stop 검증은 동일하게 실행된다.
+
 Codex에 전달하는 프롬프트는 다음 내용의 조합이다.
 
 ```text

docs/MITC4_SUPP.md

@@ -0,0 +1,604 @@
+# MITC4 구현 회고 및 보충 기록
+
+## 1. 문서 목적과 범위
+
+이 문서는 `linear-static-mitc4-shell` 기능을 요구조건부터 `dev` 병합 검증까지
+진행하면서 실제로 겪은 시행착오, 실수, 어려움과 그 해결 방법을 기록한다. 새 계약을
+정의하는 문서가 아니라, 이미 승인된 요구조건·정식화·I/O·reference·release 문서를
+보충하는 회고 자료다. 계약이 충돌하면 이 문서가 아니라 다음 문서를 우선한다.
+
+- `docs/linear-static-mitc4-shell/requirements.md`
+- `docs/linear-static-mitc4-shell/formulation.md`
+- `docs/linear-static-mitc4-shell/io.md`
+- `docs/linear-static-mitc4-shell/reference-model.md`
+- `docs/linear-static-mitc4-shell/numerical-review.md`
+- `docs/linear-static-mitc4-shell/release.md`
+
+라인 참조는 회고 작성 시점의 기준 커밋 `6c41cde41af4d6cd474b008e5ae81769fff0aa79`
+을 사용한다. `path:line`은 이 기준 커밋의 파일과 라인을 뜻한다. 중간 실패가 최종
+phase ledger에서 정상 완료 상태로 교체된 경우에는 `commit:path:line`으로 historical
+snapshot을 표시한다. 이후 파일이 수정되면 라인이 이동할 수 있으므로 커밋과 검색어를
+함께 확인해야 한다.
+
+이 문서는 다음 세 범주를 구분한다.
+
+1. **실수 또는 결함**: 승인된 의미와 다른 동작, 잘못된 가정, 환경 실패의 오분류.
+2. **계약 변경에 따른 재작업**: 구현 결함이 아니라 승인 범위나 tolerance가 바뀌어
+   이미 작성한 문서·테스트·비교기를 수정한 경우.
+3. **예상된 TDD RED와 수치적 어려움**: 계획된 실패로 아직 없는 동작을 확인하거나,
+   shell 요소 특성상 조심해서 해결해야 했던 문제. 이것을 구현 실수로 과장하지 않는다.
+
+## 2. 주요 시행착오 요약
+
+| ID | 분류 | 문제 또는 어려움 | 핵심 해결 | 대표 근거 |
+| --- | --- | --- | --- | --- |
+| `RET-01` | 프로세스 실수 | numerical review가 수치 정식화 외의 bundle 행정 정보를 readiness blocker로 취급함 | FESA 독립 솔버 원칙과 실제 비교에 필요한 최소 계약으로 gate를 재정의함 | `docs/ADR.md:189`, `docs/numerical-reviews/linear-static-mitc4-shell-review.md:80` |
+| `RET-02` | 프로세스 어려움 | Planning Agent, Implementation Agent, Executor, hook의 소유권이 모호했음 | 승인·materialize·Step 실행·상태/커밋 책임을 분리함 | `.codex/agents/implementation-agent.toml:46`, `.agents/skills/harness/SKILL.md:16` |
+| `RET-03` | 환경 결함 | Windows sandbox helper 실패가 프로세스 종료 코드 `0` 안에 숨고 Step 0이 갱신되지 않음 | 진단 marker를 종료 코드보다 먼저 검사하고 승인된 sandbox override를 추가함 | `scripts/execute.py:247`, `scripts/execute.py:306` |
+| `RET-04` | 수치적 어려움 | 20 physical DOF MITC4를 외부 24 DOF shell 계약에 넣으면서 drilling을 분리해야 했음 | physical과 drilling congruence를 분리하고 회전 대각항만으로 고정 안정화를 계산함 | `docs/formulations/mitc4-shell-formulation.md:759`, `src/fesa/elements/mitc4_shell.cpp:597` |
+| `RET-05` | 기하 처리 어려움 | warped mesh와 공유 절점에서 initial director의 방향·순서·유효성을 결정해야 했음 | source-order 정렬, 면적 가중 평균, 양의 incident-normal 방향, 전 적분점 `J>0` 검사를 사용함 | `src/fesa/model/shell_geometry.cpp:275`, `src/fesa/model/shell_geometry.cpp:317` |
+| `RET-06` | 범위 경계 | 선형 구현에 future geometric-nonlinear tangent 정식화가 섞일 위험이 있었음 | 선형 24 DOF 구현과 조건부 nonlinear 20-to-24 pullback을 명시적으로 분리함 | `docs/formulations/mitc4-shell-formulation.md:1112`, `docs/numerical-reviews/linear-static-mitc4-shell-review.md:71` |
+| `RET-07` | parser 결함 | 같은 이름의 Abaqus `NSET`과 `ELSET`을 중복 entity로 잘못 거부함 | node-set과 element-set namespace를 분리함 | `src/fesa/io/abaqus/domain_mapper.cpp:401`, `tests/unit/io/abaqus/domain_mapper_test.cpp:439` |
+| `RET-08` | comparator 결함 | Abaqus CSV와 HDF5 instance name의 대소문자 차이가 row identity mismatch를 만듦 | 비교 key만 ASCII 대문자로 정규화하고 source identity 출력은 보존함 | `tests/reference/mitc4_reference_comparison.cpp:122`, `tests/reference/mitc4_reference_comparison_test.cpp:539` |
+| `RET-09` | 계약 재작업 | 동일한 FESA kernel 결과를 Abaqus S4와 S4R 두 reference에 동시에 맞출 수 없었음 | full-integration FESA의 blocking reference를 S4 하나로 제한하고 S4R은 비-reference test로 검증함 | `docs/requirements/linear-static-mitc4-shell.md:201`, `docs/requirements/linear-static-mitc4-shell.md:205` |
+| `RET-10` | 계약 재작업 | B33 혼합 tolerance가 독립 shell 정식화 비교에 지나치게 엄격해 kernel 결함과 계약 차이를 혼동함 | invariant/patch 검증 후 MITC4 고정 절대 tolerance `1.0e-5`로 계약과 비교기를 함께 변경함 | `docs/ADR.md:215`, `tests/reference/mitc4_reference_comparison.cpp:35` |
+| `RET-11` | 통합 어려움 | prescribed-only shell에서 이미 상쇄된 `K*d`만으로 residual을 정규화하면 정상 roundoff가 단위 residual이 됨 | `Kff*df`, `Kfc*dc`, `Ff`의 물리적 항 크기로 정규화함 | `src/fesa/results/result_recovery.cpp:674`, `phases/linear-static-mitc4-shell/index.json:102` |
+| `RET-12` | 상태/출력 안전성 | recovery 또는 HDF5 inventory가 뒤에서 실패할 때 기존 정상 state/output을 손상할 위험이 있었음 | candidate state와 temporary HDF5를 완전히 검증한 뒤 atomic commit/replace함 | `src/fesa/results/result_recovery.cpp:925`, `src/fesa/io/hdf5/hdf5_results_writer.cpp:2579` |
+| `RET-13` | 병합 환경 오염 | `dev` worktree의 ignored B33 `.h5`가 exact reference inventory test를 실패시킴 | reference를 수정하지 않고 ignored 파일을 quarantine으로 이동한 뒤 전체 검증함 | `.gitignore:20`, `tests/reference/b33_reference_comparison_test.cpp:178` |
+
+## 3. 계약과 개발 프로세스에서의 시행착오
+
+### 3.1 Numerical Review가 행정 메타데이터에 과도하게 의존함 (`RET-01`)
+
+#### 문제
+
+초기 numerical review는 정식화의 수학적 준비 상태뿐 아니라 canonical bundle 이름,
+`README.md`, provenance, units, step/frame 설명까지 갖춰야 implementation planning으로
+넘어갈 수 있다고 판단했다. `metadata.json` 자체는 optional로 정리했지만, 그 주변의
+행정 정보가 여전히 primary verdict를 `needs-reference-model`로 유지했다.
+
+Historical evidence는 다음과 같다.
+
+- `ebb2657:docs/numerical-reviews/linear-static-mitc4-shell-review.md:12` — status가
+  `needs-reference-model`이었다.
+- `0428759:docs/numerical-reviews/linear-static-mitc4-shell-review.md:94` — alias,
+  README, provenance, unit, step/frame, schema, tolerance를 open blocker로 묶었다.
+- `0428759:docs/numerical-reviews/linear-static-mitc4-shell-review.md:702` — `NR-O01`부터
+  `NR-O04`까지 calibration을 implementation-planning blocker로 남겼다.
+
+#### 원인
+
+Reference artifact의 재현성 감사와 현재 기능의 수치 readiness를 같은 gate로 취급했다.
+또한 Abaqus가 제공하는 bundle 관리 방식과 FESA가 실제로 비교에 필요한 observable
+quantity 계약을 충분히 분리하지 못했다. 결과적으로 정식화 결함이 아닌 문서 형식이
+수치 검토를 막았다.
+
+#### 해결
+
+프로젝트 정책을 다음처럼 다시 고정했다.
+
+- FESA는 Abaqus의 내부 알고리즘을 재현하는 솔버가 아니다
+  (`docs/ADR.md:189-195`).
+- readiness에 필요한 것은 선언된 input/CSV, deterministic source row mapping,
+  component와 tolerance다 (`docs/ADR.md:197-202`).
+- `NR-O01`은 고정 drilling 계수 결정으로 닫고, `NR-O02`부터 `NR-O04`까지는 승인
+  범위에서 제거했다 (`docs/numerical-reviews/linear-static-mitc4-shell-review.md:74-77`).
+- 과거 blocker였던 canonical naming, README, `metadata.json`, provenance와 expanded
+  portfolio가 formulation verdict를 막지 않는다고 명시했다
+  (`docs/numerical-reviews/linear-static-mitc4-shell-review.md:80-83`).
+
+관련 정책 변경 커밋은 `5c08f1c`(independent reference validation policy),
+`73df844`(MITC4 verification/drilling scope 단순화), `60b42f4`(numerical review pass)다.
+
+#### 교훈
+
+Reference gate에는 “이 정보가 실제 row matching 또는 수치 판정에 사용되는가?”를 먼저
+물어야 한다. 사용되지 않는 행정 메타데이터를 모든 기능의 blocking requirement로
+자동 승격하면 독립 솔버의 물리 검증보다 형식 준수가 우선된다.
+
+### 3.2 Harness 역할과 Step 소유권이 처음부터 충분히 명확하지 않았음 (`RET-02`)
+
+#### 문제
+
+구현 계획 승인, phase 파일 materialize, Harness 실행, Step 선택, 구현, timestamp와
+commit 기록의 주체가 여러 agent 문서에 분산되어 있었다. 이 상태에서는 Planning Agent가
+실행까지 시작하거나, Implementation Agent가 다음 Step으로 넘어가거나, hook을 수동
+검증 명령처럼 실행하는 실수가 생길 수 있었다.
+
+#### 해결
+
+다음 소유권을 agent와 Harness 문서에 중복 없이 고정했다.
+
+- Planning Agent는 multi-Step draft 승인 후에만 phase 파일을 만들고 Step을 실행하지
+  않는다 (`.codex/agents/implementation-planning-agent.toml:52-60`).
+- Implementation Agent는 Executor가 선택한 현재 `stepN.md` 하나만
+  `RED -> GREEN -> VERIFY`로 수행한다 (`.codex/agents/implementation-agent.toml:46-60`).
+- branch, retry, timestamp, commit과 next-Step selection은 Executor 소유다
+  (`.codex/agents/implementation-agent.toml:64-67`).
+- hook은 등록된 PreToolUse/Stop lifecycle로 자동 실행하며 수동 호출로 대체하지 않는다
+  (`.agents/skills/harness/SKILL.md:16-25`).
+- 계획 승인은 실행 승인이 아니며 별도의 사용자 요청이 있어야 한다
+  (`docs/HARNESS_WORKFLOW.md:41-42`).
+
+이 개선은 `85cd17d`, `ab69a3d`, `713b41b`, `65d5e07`, `18296a1`, `a058ef7`
+커밋에 걸쳐 정리됐다.
+
+#### 교훈
+
+Agentic phase에서 “무엇을 할 것인가”뿐 아니라 “누가 상태를 바꾸는가”를 계약해야 한다.
+특히 implementation summary와 executor timestamp/commit을 한 주체가 모두 소유하게 하면
+재시도와 감사 이력이 쉽게 꼬인다.
+
+### 3.3 Windows sandbox 실패가 정상 종료처럼 보임 (`RET-03`)
+
+#### 문제
+
+Step 0의 첫 Harness 실행은 세 번 재시도한 뒤에도 Step status를 갱신하지 못했다.
+historical ledger는 `8dd4d72:phases/linear-static-mitc4-shell/index.json:10`에
+`[3회 시도 후 실패] Step did not update status`를 기록한다. 실제 원인은 구현 코드가
+아니라 Windows sandbox helper의 `orchestrator_helper_launch_failed`였다. 더 까다로운 점은
+이 진단이 Codex JSON output 안에 있으면서 process exit code는 `0`일 수 있었다는 것이다.
+
+#### 해결
+
+- `_codex_environment_failure()`가 exit code를 보기 전에 diagnostic text에서 sandbox
+  helper marker를 검사하도록 바꿨다 (`scripts/execute.py:306-314`).
+- 기본 `workspace-write`는 유지하되, 승인된 격리 worktree에 한해서
+  `FESA_HARNESS_CODEX_SANDBOX=danger-full-access`를 허용했다
+  (`scripts/execute.py:247-254`, `docs/HARNESS_WORKFLOW.md:247-252`).
+- unknown mode는 fail-closed로 거부하고, exit code `0` 안의 helper failure도 잡는
+  회귀 테스트를 추가했다 (`tests/test_execute.py:77-111`).
+
+수정 커밋은 `0d50625`이며, 이후 Harness Python suite `7/7`이 통과했다
+(`docs/releases/linear-static-mitc4-shell-release.md:176-178`).
+
+#### 교훈
+
+Agent runner에서는 OS process exit code만 신뢰하면 안 된다. 하위 orchestrator가 구조화된
+출력 안에 fatal environment error를 기록하는 경우가 있으므로, 알려진 환경 실패 marker와
+상태 전이를 함께 검사해야 한다.
+
+## 4. 정식화와 수치 구현의 어려움
+
+### 4.1 20 physical DOF와 24 global DOF 사이의 경계 (`RET-04`)
+
+#### 어려움
+
+MITC4의 physical field는 절점당 translation 3개와 director-tangent rotation 2개, 즉
+총 20 DOF다. 반면 FESA shell 외부 계약은 절점당 6 DOF, 총 24 DOF다. 남는 네 개의
+director-parallel 회전은 drilling 좌표이며 physical MITC4 strain에 들어가면 안 된다.
+
+초기의 “작은 diagonal stiffness를 더한다”는 수준의 설명만으로는 다음 실수가 가능했다.
+
+- translation과 rotation diagonal을 섞어 단위가 다른 값으로 `k_d`를 정하는 문제;
+- drilling energy를 physical shell energy나 stress recovery에 섞는 문제;
+- 20 DOF weak form에 24 DOF external work를 직접 결합하는 문제;
+- 작은 값이므로 rigid mode나 rank 검증을 생략하는 문제.
+
+Numerical Review에서 이 위험은 `NR-C02`부터 `NR-C04`로 추적됐다
+(`docs/numerical-reviews/linear-static-mitc4-shell-review.md:68-70`).
+
+#### 해결
+
+정식화와 코드를 두 경로로 나눴다.
+
+1. `K20`은 MITC tying shear와 physical constitutive matrix만으로 적분한다
+   (`src/fesa/elements/mitc4_shell.cpp:597-636`).
+2. `k_d` reference는 절점당 두 개, 총 여덟 개 physical tangent-rotation diagonal의
+   finite positive 값만 사용한다 (`src/fesa/elements/mitc4_shell.cpp:638-655`).
+3. `k_d = 1.0e-3 * min(R+)`를 적용한다
+   (`src/fesa/elements/mitc4_shell.cpp:662-676`).
+4. physical과 drilling을 별도 congruence로 24 DOF에 올린 후 마지막에만 합한다
+   (`src/fesa/elements/mitc4_shell.cpp:669-681`).
+5. recovery는 global 24 DOF를 physical 20 DOF로 projection한 뒤 계산하므로 pure drill이
+   strain/resultant/stress/physical energy에 기여하지 않는다
+   (`src/fesa/elements/mitc4_shell.cpp:699-714`).
+
+이를 energy congruence, rank와 six rigid modes, patch field, fixed drilling factor,
+pure-drill exclusion 테스트로 각각 검증했다
+(`tests/unit/elements/mitc4_shell_test.cpp:618`, `:647`, `:697`, `:752`, `:828`).
+
+#### 교훈
+
+수치 안정화는 크기만 작은 physical stiffness가 아니다. coordinate map, 단위, virtual
+work와 output 의미를 별도 경계로 정의해야 한다. 안정화 항을 physical result에서 제외하는
+테스트가 coefficient 자체의 테스트만큼 중요하다.
+
+### 4.2 Initial director와 warped/shared-node geometry의 결정성 (`RET-05`)
+
+#### 어려움
+
+“초기 법선벡터를 두께 방향으로 사용한다”는 결정은 단일 평면 요소에는 단순하지만,
+공유 절점과 warped mesh에서는 다음 선택을 추가로 요구했다.
+
+- element normal의 부호를 어느 source connectivity 순서로 정할지;
+- 여러 incident element normal을 어떤 순서와 weight로 평균할지;
+- 반대 방향 normal, fold, zero-area, local reversal을 허용할지;
+- stiffness, tying, recovery가 서로 다른 geometry inventory를 사용하지 않게 할지.
+
+#### 해결
+
+- source connectivity로 center normal과 element orientation을 결정했다.
+- required surface points에서 finite/nonzero measure와 positive orientation을 먼저
+  검사했다 (`src/fesa/model/shell_geometry.cpp:275-300`).
+- 공유 절점의 incident element를 stable source identity 순서로 정렬했다
+  (`src/fesa/model/shell_geometry.cpp:317-327`).
+- 모든 incident normal 쌍이 같은 positive hemisphere에 있는지 확인한 다음 면적 가중
+  평균을 계산했다 (`src/fesa/model/shell_geometry.cpp:328-364`).
+- geometry validity를 center, stiffness, tying, recovery point inventory 전체에서
+  fail-closed로 검사했다. Calibration angle이나 distortion cutoff를 새로 만들지는 않았다
+  (`docs/formulations/mitc4-shell-formulation.md:593`,
+  `docs/numerical-reviews/linear-static-mitc4-shell-review.md:76-77`).
+
+평면·회전·warped geometry, stable area weighting, invalid Jacobian/opposed normal,
+validation-point inventory 테스트가 각각 존재한다
+(`tests/unit/model/shell_geometry_test.cpp:101`, `:153`, `:184`, `:244`).
+
+#### 교훈
+
+자동 normal 생성은 단순한 vector 계산이 아니라 mesh topology와 deterministic reduction
+계약이다. 순서를 명시하지 않으면 병렬화나 input order 변화가 director와 결과에 영향을
+줄 수 있다.
+
+### 4.3 Geometric-nonlinear tangent를 선형 구현으로 오인할 위험 (`RET-06`)
+
+#### 어려움
+
+정식화 문서에는 future geometric-nonlinear tangent도 포함했지만, 현재 제품 범위는 선형
+정적이다. 20-coordinate director chart의 material/geometric tangent만 적는 것으로
+global 24 DOF nonlinear element가 완성되는 것은 아니다. finite rotation update,
+`Phi: R24 -> R20`, 그 1·2차 미분, chart recentering과 objective drilling potential이
+필요하다.
+
+#### 해결
+
+- nonlinear 식이 Section 15.3까지 20-coordinate physical chart에만 존재하며 완전한
+  global 24 DOF element가 아니라고 명시했다
+  (`docs/formulations/mitc4-shell-formulation.md:1112-1113`).
+- conditional pullback에는 map-curvature Hessian 항을 유지하고, 미정인 coordinate map과
+  drilling을 future-only blocker로 남겼다
+  (`docs/numerical-reviews/linear-static-mitc4-shell-review.md:71`).
+- current review의 open question은 nonlinear rotation/map/output/state 계약뿐이며 현재
+  선형 planning을 차단하지 않는다
+  (`docs/numerical-reviews/linear-static-mitc4-shell-review.md:85-90`).
+- 구현 phase에는 linear `K20 -> K24`와 고정 drilling만 넣고 nonlinear state나 tangent
+  API를 만들지 않았다.
+
+#### 교훈
+
+문서에 수식이 있다는 사실과 제품 계약이 닫혔다는 사실은 다르다. 특히 nonlinear
+coordinate map의 Hessian을 생략한 채 “consistent tangent”라고 부르면 이후 Newton
+iteration의 일관성을 잘못 주장하게 된다.
+
+## 5. Parser와 결과 파이프라인의 실제 결함
+
+### 5.1 Abaqus NSET/ELSET namespace를 하나로 처리함 (`RET-07`)
+
+#### 증상
+
+Step 13 reference E2E는 comparator contract test가 통과한 뒤에도 모든 S4/S4R case에서
+HDF5 생성 전에 `duplicate-entity`로 실패했다. historical ledger
+`32f0e59:phases/linear-static-mitc4-shell/index.json:113`은 part `NSET Set-1`과
+`ELSET Set-1`을 같은 namespace로 검사한 것이 원인임을 기록한다. Step 13의 test-only
+소유 범위에서 production parser를 고칠 수 없었기 때문에 세 번 재시도 후 upstream
+correction이 필요했다.
+
+#### 원인과 해결
+
+기존 `containsSetName()`은 part의 node set과 element set 양쪽을 함께 검색했다.
+Abaqus input에서는 두 종류가 별도 namespace이므로 같은 이름이 유효하다.
+
+- part helper가 전달받은 set 종류 하나만 검색하도록 변경했다
+  (`src/fesa/io/abaqus/domain_mapper.cpp:401-408`).
+- assembly helper도 `isNodeSet`이 같은 항목만 비교하도록 변경했다
+  (`src/fesa/io/abaqus/domain_mapper.cpp:410-418`).
+- part와 assembly에서 같은 이름의 NSET/ELSET을 허용하되 같은 종류의 실제 중복은
+  계속 거부하는 회귀 테스트를 추가했다
+  (`tests/unit/io/abaqus/domain_mapper_test.cpp:439`).
+
+수정 커밋은 `b805683`이다.
+
+#### 교훈
+
+Source label text가 같다는 이유만으로 semantic identity가 같은 것은 아니다. parser의
+duplicate 검사는 `(entity kind, scope, source name)` 전체 identity를 사용해야 한다.
+
+### 5.2 Reference instance identity의 대소문자 차이를 그대로 비교함 (`RET-08`)
+
+#### 증상과 원인
+
+Parser namespace를 고친 뒤 reference precheck에서 Abaqus CSV의 `part-1-1`과 FESA HDF5의
+`Part-1-1`이 다른 key로 취급됐다. CSV field whitespace는 이미 trim했지만 Abaqus name의
+case-insensitive 의미를 row identity comparator에 적용하지 않았다.
+
+#### 해결
+
+- comparator ordering/equality key에서 instance name을 ASCII uppercase로 정규화했다
+  (`tests/reference/mitc4_reference_comparison.cpp:122-142`).
+- report에는 HDF5가 보존한 stable source identity를 그대로 남겼다. 즉 비교를 위해 source
+  artifact를 rename하거나 rewrite하지 않았다.
+- lowercase CSV identity가 정상 match되는 회귀 테스트를 추가했다
+  (`tests/reference/mitc4_reference_comparison_test.cpp:539`).
+
+수정 커밋은 `56833ab`이다.
+
+#### 교훈
+
+Identity normalization은 어느 계층에서 어떤 목적으로 하는지 제한해야 한다. comparator
+key의 case normalization과 source identity 보존을 분리하면 matching은 견고해지면서도
+진단·HDF5 identity가 조용히 바뀌는 문제를 피할 수 있다.
+
+### 5.3 Prescribed-only free residual normalization (`RET-11`)
+
+#### 증상
+
+Step 12의 nonzero prescribed-displacement shell case는 실제 equilibrium을 만족했지만
+free-residual 검증에서 실패했다. 이미 상쇄된 `K*d` 또는 외력만 normalization scale로
+쓰면 `Ff=0`이고 `Kff*df`와 `Kfc*dc`가 서로 상쇄되는 case에서 작은 roundoff가 상대적으로
+`1`에 가까운 residual처럼 보일 수 있었다. Phase ledger가 이 실패와 해결을 기록한다
+(`phases/linear-static-mitc4-shell/index.json:102-107`).
+
+#### 해결
+
+free residual `Kff*df + Kfc*dc - Ff`의 세 물리 항을 따로 평가하고 그 최대 norm을
+denominator로 사용했다 (`src/fesa/results/result_recovery.cpp:674-697`). 임의의
+`max(1, ...)` floor는 넣지 않았고, scale과 residual이 둘 다 정확히 0일 때만 normalized
+residual을 0으로 처리했다.
+
+Integration test는 `Ff=0`인데 `-Kfc*dc`만으로 effective RHS가 생기는 것을 직접 확인한다
+(`tests/integration/analysis/linear_static_analysis_test.cpp:440-464`).
+
+#### 교훈
+
+Residual normalization은 최종 합의 크기만 보면 안 된다. 서로 상쇄되는 원래 방정식 항의
+물리적 scale을 보존해야 prescribed load, zero load와 mixed constraint를 같은 기준으로
+검증할 수 있다.
+
+### 5.4 Recovery state와 HDF5 output의 실패 원자성 (`RET-12`)
+
+#### 어려움
+
+Shell 결과는 displacement 하나가 아니라 residual/reaction, GP strain/resultant,
+BOTTOM/MIDDLE/TOP stress, frame, physical energy와 equilibrium evidence를 함께 commit한다.
+뒤쪽 row가 invalid일 때 일부 vector만 기존 `AnalysisState`에 반영하거나, 기존 정상
+`results.h5`를 불완전한 새 파일로 바꾸면 실패가 이전 성공 상태를 손상한다.
+
+TDD 과정에서 다음 문제가 드러났다.
+
+- invalid shell candidate가 이전 state를 덮을 수 있음
+  (`phases/linear-static-mitc4-shell/index.json:78-83`);
+- beam-only recovery 뒤 stale shell evidence가 남을 수 있음
+  (`phases/linear-static-mitc4-shell/index.json:85-91`);
+- invalid HDF5 inventory가 기존 final output을 안전하지 않게 교체할 수 있음
+  (`phases/linear-static-mitc4-shell/index.json:93-99`).
+
+#### 해결
+
+- recovery는 완전한 `candidateState`를 복사해 모든 shell row를 검증하고 마지막에 한 번만
+  move-commit한다 (`src/fesa/results/result_recovery.cpp:925-940`).
+- HDF5 writer는 unique temporary file에 쓰고 닫은 뒤 read-only로 reopen/self-check한
+  파일만 final path로 교체한다 (`src/fesa/io/hdf5/hdf5_results_writer.cpp:2579-2592`).
+- drilling energy와 unsupported point stress가 accidental output으로 생기지 않았는지도
+  self-check한다 (`src/fesa/io/hdf5/hdf5_results_writer.cpp:2449-2455`).
+- invalid shell inventory가 기존 final bytes를 보존하는 회귀 테스트를 추가했다
+  (`tests/unit/io/hdf5/hdf5_results_writer_test.cpp:1359`).
+
+#### 교훈
+
+Solver output의 원자성은 HDF5 backend만의 책임이 아니다. recovery candidate의 의미적
+완전성, temporary file의 구조적 완전성, final replacement 순서를 모두 지켜야 한다.
+
+## 6. Reference comparison에서의 재작업
+
+### 6.1 S4와 S4R을 동시에 blocking reference로 사용한 모순 (`RET-09`)
+
+#### 증상
+
+초기 reference plan은 S4와 S4R 두 Abaqus bundle을 모두 blocking comparison으로
+사용했다. 그러나 FESA 계약은 두 source type을 같은 full-integration FESA-MITC4 kernel에
+매핑한다. 실제 Abaqus S4와 S4R은 동일한 내부 정식화가 아니므로 두 Abaqus reference
+결과도 달랐다.
+
+Historical ledger `cf769aa:phases/linear-static-mitc4-shell/index.json:113`은 다음 불가능한
+상태를 수치로 기록한다.
+
+- 두 FESA deck은 heading과 `TYPE=S4/S4R` 외에는 같고 동일 stiffness를 생성함;
+- Abaqus center `U3`는 두 reference 사이에 `8.190036e-7` 차이가 남;
+- 당시 더 큰 tolerance도 약 `1.02456e-9`여서 하나의 공통 FESA 결과가 두 reference를
+  동시에 통과할 수 없음.
+
+이는 kernel을 S4와 S4R에 따라 다르게 만들라는 신호가 아니었다. 그렇게 하면 “둘 다
+동일한 FESA formulation”이라는 승인 계약과 Abaqus 비동등성 원칙을 위반한다.
+
+#### 해결
+
+- blocking Abaqus comparison은 `reference/shell/`의 S4 input/displacement CSV 하나로
+  고정했다 (`docs/requirements/linear-static-mitc4-shell.md:201-205`).
+- `reference/shellR/`은 acceptance에서 소비하지 않는다
+  (`docs/requirements/linear-static-mitc4-shell.md:202`).
+- S4R 지원은 parser identity, 동일 sparse stiffness, HDF5 source metadata test로
+  검증한다 (`tests/unit/io/abaqus/domain_mapper_test.cpp:538`,
+  `tests/unit/assembly/sparse_assembler_test.cpp:324`,
+  `tests/unit/io/hdf5/hdf5_results_writer_test.cpp:1167`).
+- S4 E2E test만 유지했다 (`tests/reference/mitc4_reference_cases_test.cpp:159-169`).
+
+문서 변경은 `91b2df9`, test 변경은 `f92c17d`에 기록됐다.
+
+#### 교훈
+
+입력 label을 지원하는 것과 그 label 이름을 가진 외부 solver element를 reference로
+사용하는 것은 별도 결정이다. 공통 internal formulation이면 acceptance reference도 그
+formulation의 observable 목적에 맞는 하나를 선택해야 한다.
+
+### 6.2 Reference 실패를 kernel 결함과 tolerance 결함으로 분리하기 (`RET-10`)
+
+#### 증상
+
+S4-only gate로 줄인 뒤에도 25개의 nonzero `U3` row가 당시 B33식 tolerance를 넘었다.
+historical ledger `8994ce9:phases/linear-static-mitc4-shell/index.json:113`은 worst row
+`Part-1-1/node 2/U3`에서 다음을 기록한다.
+
+- absolute error: `1.90378534915144e-7`;
+- 당시 tolerance: `1.0237408203e-9`;
+- normalized error: `185.9636`.
+
+이 시점에 tolerance만 즉시 키우면 실제 MITC4 kernel 부호·tying·Jacobian 결함을 숨길 수
+있고, 반대로 Abaqus S4와 동일하지 않은 FESA-MITC4에 B33 수준의 엄격한 tolerance를 계속
+강제하면 올바른 독립 정식화를 실패로 오판할 수 있었다.
+
+#### 독립 조사와 해결
+
+먼저 reference 결과와 독립적인 다음 검사를 통과시켰다.
+
+- 20-to-24 energy congruence와 six rigid modes/rank
+  (`tests/unit/elements/mitc4_shell_test.cpp:618-647`);
+- membrane, bending, transverse shear와 twist patch
+  (`tests/unit/elements/mitc4_shell_test.cpp:697`);
+- pure drilling의 physical recovery/energy exclusion
+  (`tests/unit/elements/mitc4_shell_test.cpp:828`);
+- S4/S4R common stiffness와 deterministic assembly
+  (`tests/unit/assembly/sparse_assembler_test.cpp:324`);
+- authoritative FESA HDF5에서 같은 stable error가 발생하며 comparator 자체의 row mismatch가
+  아님을 확인함.
+
+그 뒤 user-approved contract를 MITC4 고정 절대 tolerance `1.0e-5`로 변경했다.
+
+- ADR은 U blocking, UR warning-only와 reference scale의 diagnostic-only 성격을 고정한다
+  (`docs/ADR.md:215-222`).
+- comparator는 모든 row에 하나의 fixed tolerance를 적용한다
+  (`tests/reference/mitc4_reference_comparison.cpp:35`, `:755-767`).
+- 경계 바로 아래와 위, UR warning-only를 회귀 테스트한다
+  (`tests/reference/mitc4_reference_comparison_test.cpp:662`, `:712`).
+- final reference verification에서 같은 worst absolute error는 tolerance의 약 `0.0190`이고
+  U `147/147`, UR `147/147`, warning `0`으로 통과했다
+  (`docs/reference-verifications/linear-static-mitc4-shell-reference-verification.md:138-157`).
+
+문서 변경 커밋은 `f1be807`, comparator 완료 커밋은 `426cce1`이다.
+
+#### 교훈
+
+Reference mismatch가 보이면 먼저 element invariant와 patch test로 kernel을 독립 검증하고,
+row identity/schema와 authoritative HDF5를 확인한 다음 tolerance 계약을 판단해야 한다.
+Tolerance 완화가 kernel 조사보다 먼저 오면 결함 은폐가 되고, 외부 solver 내부 동등성을
+가정한 과도한 tolerance는 false failure가 된다.
+
+## 7. 정상적인 TDD RED와 실제 재작업의 구분
+
+Phase ledger의 대부분 RED는 계획된 TDD 증거다. 다음 실패들은 “처음부터 구현이 잘못됐다”는
+뜻이 아니라, 해당 Step이 소유한 behavior가 아직 없음을 테스트가 정확히 검출한 것이다.
+
+| Step | 예상된 RED | GREEN에서 추가한 핵심 동작 | 근거 |
+| --- | --- | --- | --- |
+| 0-1 | shell semantic type과 mapping 부재 | S4/S4R source identity와 공통 FESA-MITC4 model | `phases/linear-static-mitc4-shell/index.json:5-19` |
+| 2 | `shell_geometry.hpp` 부재 | deterministic normal/director와 geometry validation | `phases/linear-static-mitc4-shell/index.json:21-27` |
+| 3 | `mitc4_shell.hpp` 부재 | 20 DOF kinematics, tying, constitutive, fixed quadrature | `phases/linear-static-mitc4-shell/index.json:29-35` |
+| 4-5 | stiffness와 recovery API 부재 | physical/drilling stiffness 분리와 pure-drill-free recovery | `phases/linear-static-mitc4-shell/index.json:37-51` |
+| 6-7 | shell scatter/assembly 부재 | 24-entry scatter와 source-ordered 576-entry COO reduction | `phases/linear-static-mitc4-shell/index.json:53-67` |
+| 8 | drilling direction moment를 허용함 | aggregate moment projection 검사와 RHS 이전 거부 | `phases/linear-static-mitc4-shell/index.json:69-75` |
+| 9-11 | state/recovery/HDF5 shell inventory 부재 | atomic state commit과 self-checked HDF5 replace | `phases/linear-static-mitc4-shell/index.json:77-99` |
+| 12 | prescribed-only normalization 실패 | `Kff/Kfc/Ff` term-scale normalization | `phases/linear-static-mitc4-shell/index.json:101-107` |
+
+반면 `RET-03`, `RET-07`, `RET-08`은 실제 환경 또는 코드 결함이었다. `RET-01`,
+`RET-09`, `RET-10`은 upstream 정책·범위·tolerance 결정이 바뀌어 발생한 재작업이다.
+이 구분을 유지해야 TDD의 의도된 실패를 품질 문제로 잘못 집계하지 않는다.
+
+## 8. `dev` 병합 검증에서 발견한 ignored artifact 오염 (`RET-13`)
+
+### 8.1 증상
+
+Feature branch 자체와 병합 직후 MITC4 test는 통과했지만, `C:\git\FESADev`의 `dev`
+worktree에서 full CTest를 실행하자 기존 B33 reference test 하나가 실패했다.
+
+```text
+B33ReferenceComparison.GeneratesAuthoritativeHdf5AndComparisonEvidence
+referenceBefore.size() = 5, expected = 4
+```
+
+MITC4 변경으로 B33 reference가 바뀐 것이 아니라, 해당 worktree의
+`reference/cantilever beam/`에 pre-existing `cantilever beam.h5`가 남아 있었다.
+`.gitignore`가 모든 `.h5`를 무시하므로 일반 `git status`는 이 파일을 보여주지 않았다
+(`.gitignore:20`). B33 test는 reference tree를 snapshot하고 정확히 네 파일만 있어야
+한다고 검사한다 (`tests/reference/b33_reference_comparison_test.cpp:73`, `:178-191`).
+
+### 8.2 해결
+
+Reference artifact를 삭제하거나 수정하지 않았다. ignored HDF5를 다음 복구 가능한 위치로
+이동했다.
+
+```text
+C:\git\FESADev\.harness\quarantine\preexisting-reference-artifacts\cantilever beam.h5
+SHA-256: 233DF10D6A13941A477BC5133A0D686D939FEF598B3748B6DE69D2684320E01F
+```
+
+이동 뒤 focused B33 test `1/1`, full CTest `144/144`, Harness pytest `7/7`을 다시 실행했다.
+Tracked reference path에는 변경이 없었다.
+
+### 8.3 교훈
+
+- clean worktree 감사에는 `git status`뿐 아니라 reference directory의 ignored file inventory도
+  포함해야 한다.
+- 테스트가 exact artifact count를 요구하면 build/output extension이 ignore되어 있는지
+  함께 확인해야 한다.
+- 사용자 또는 이전 실행이 만든 ignored file은 바로 삭제하지 말고 hash를 기록해 workspace
+  내부 quarantine으로 이동하는 것이 안전하다.
+
+## 9. 재사용할 개발 체크리스트
+
+### 9.1 Requirements와 Numerical Review
+
+- [ ] 외부 solver의 input label, internal algorithm과 observable reference quantity를
+      분리했는가?
+- [ ] readiness blocker가 실제 수치 판정에 필요한 정보인지 확인했는가?
+- [ ] optional metadata, provenance, bundle naming을 자동으로 blocking gate로 올리지 않았는가?
+- [ ] future nonlinear 식과 현재 executable scope를 명확히 분리했는가?
+
+### 9.2 Shell kernel
+
+- [ ] physical DOF와 numerical stabilization DOF의 transform, energy와 output을 분리했는가?
+- [ ] translation/rotation 혼합 단위의 raw diagonal 또는 spectrum으로 계수를 정하지 않는가?
+- [ ] normal/director reduction order와 invalid orientation predicate가 deterministic한가?
+- [ ] reference 전에 rigid mode, rank, energy congruence와 independent patch를 통과하는가?
+
+### 9.3 Parser와 reference comparison
+
+- [ ] identity가 `(kind, scope, source label)`을 모두 포함하는가?
+- [ ] source format의 case/whitespace 규칙을 comparator key에만 정확히 적용하는가?
+- [ ] missing, extra, duplicate, nonfinite row를 tolerance 전에 거부하는가?
+- [ ] source type 지원과 그 source type의 Abaqus artifact consumption을 별도 계약으로
+      판단하는가?
+- [ ] tolerance 변경 전에 authoritative HDF5와 kernel invariant를 확인했는가?
+
+### 9.4 State, HDF5와 Harness
+
+- [ ] complete candidate를 검증한 뒤 state를 한 번만 commit하는가?
+- [ ] HDF5 temporary file을 close/reopen/self-check한 뒤 final을 교체하는가?
+- [ ] Agent는 Executor가 선택한 Step 하나만 수행하고 Executor-owned 상태를 쓰지 않는가?
+- [ ] process exit code와 structured diagnostic의 fatal error를 모두 검사하는가?
+- [ ] 병합 전후 reference directory에 ignored generated artifact가 없는가?
+
+## 10. 최종 결과와 남은 한계
+
+최종 release readiness는 `ready-for-release`다
+(`docs/releases/linear-static-mitc4-shell-release.md:250`). 근거는 다음과 같다.
+
+- clean MSVC x64 Debug build와 compiler/linker warning `0`;
+- full CTest `144/144`, Harness pytest `7/7`;
+- Harness Step `0-13` 완료와 output `14/14` exit code `0`;
+- S4 reference U `147/147` pass, UR `147/147` 비교와 warning `0`;
+- exact 294 comparison rows, invalid row `0`;
+- physics equilibrium, direction, symmetry, recovery와 energy sanity pass
+  (`docs/releases/linear-static-mitc4-shell-release.md:168-181`).
+
+다만 이 결과가 다음을 의미하지는 않는다.
+
+- Abaqus S4/S4R 내부 정식화 동등성;
+- reduced integration 또는 hourglass control 지원;
+- S4R Abaqus reference acceptance;
+- curved/distorted/thin-thick mesh 전체의 convergence 보장;
+- geometric-nonlinear global 24 DOF tangent 구현 완료.
+
+이 한계는 release 문서에 승인된 boundary로 남아 있다
+(`docs/releases/linear-static-mitc4-shell-release.md:201-209`). 이번 작업의 가장 중요한
+결론은 “Abaqus처럼 작동하게 만들기”가 아니라, FESA의 독립 정식화와 외부 observable
+comparison 사이의 경계를 먼저 고정하고 각 계층의 invariant로 결함을 좁혀야 한다는
+점이다.

docs/PRD.md

@@ -113,7 +113,7 @@ Parser keyword, element kernel, solver backend 또는 output dataset 중 하나
 | FESA-PRD-014 | 선형 정적 pipeline은 stiffness factorization과 load substitution을 분리해야 한다. | `Kff` factorization이 load vector assembly보다 먼저 수행되고, `rhs=Ff-Kfc*dc` substitution으로 full displacement를 복구한다. | orchestration test, solver-adapter test |
 | FESA-PRD-015 | FESA는 V0 결과와 diagnostic을 안정된 외부 계약으로 출력해야 한다. | CLI가 `0=success`, `2=usage`, `3=input`, `4=model`, `5=solver`, `6=HDF5` exit code를 사용하고 diagnostic field가 계약과 일치하며 HDF5가 displacement, reaction, end force, section resultant, generalized strain/resultant, axial `S11`을 포함하고 실패 시 불완전한 최종 파일을 남기지 않는다. | CLI integration test, HDF5 schema/atomicity test |
 | FESA-PRD-016 | Dense와 sparse math storage 및 backend 경계를 분리해야 한다. | `Vector`는 contiguous, `Matrix`는 row-major contiguous storage와 MKL CBLAS를 사용하고 `SparseMatrix`는 별도 0-based CSR 타입이며 MKL 타입이 public core API에 노출되지 않는다. | math unit test, dependency review |
-| FESA-PRD-017 | B33 reference comparison은 component-scale 혼합 tolerance를 사용해야 한다. | 모든 matched row가 `abs_error <= absolute_floor + 1e-6 * reference_scale`을 만족하고 missing/extra/nonfinite row는 comparison 전에 실패한다. | reference comparison unit/integration test, verification report |
+| FESA-PRD-017 | 모든 외부 reference comparison은 승인된 공통 family-scale tolerance를 사용해야 한다. | 각 matched row가 near-zero 대체 조건 또는 상대오차 `0.05`를 통과하고 각 family의 scale-relative RMS가 `0.01` 이하이며 missing/extra/duplicate/nonfinite row는 comparison 전에 실패한다. | reference comparison unit/integration test, verification report |
 | FESA-PRD-018 | 승인된 B33 reference artifact는 현재 경로의 read-only baseline으로 유지해야 한다. | `reference/cantilever beam/` 파일을 rename, rewrite 또는 보정하지 않고 기능이 선언한 exact path에서 읽는다. | artifact inventory, Git diff review |
 | FESA-PRD-019 | B33 beam section과 local axis를 Abaqus 의미에 맞게 매핑해야 한다. | `n1 -> local y`, `t x n1 -> local z`, `Iy=I11`, `Iz=I22`, `I12=0`을 적용하고 nonpositive property, zero-length element, tangent-parallel guide vector를 구조화된 model diagnostic으로 거부한다. | section-mapping unit test, element geometry test |
 
@@ -121,11 +121,12 @@ Parser keyword, element kernel, solver backend 또는 output dataset 중 하나
 - MSVC x64 Debug 환경에서 configure, build, CTest를 검증한다.
 - reference test 결과는 deterministic해야 한다.
 - HDF5 schema는 versioned contract로 관리한다.
-- B33 reference tolerance는 model, step/frame, quantity, component별 Abaqus scale만 사용한다.
-- 승인된 SI B33 bundle의 absolute floor는 displacement/rotation `1e-9`, force/moment
-  `1e-3`이며 relative coefficient는 `1e-6`이다.
-- MITC4 U/UR comparison은 같은 component-scale 형식을 사용한다. `U1/U2/U3`만
-  blocking이고 `UR1/UR2/UR3` 초과는 warning-only다.
+- 모든 외부 reference comparison은 동일 logical quantity와 unit dimension의 component
+  family별 Abaqus-only maximum scale을 사용한다. Near-zero ratio는 `0.01`, 일반 행
+  상대오차는 `0.05`, family scale-relative RMS는 `0.01`이며 독립 absolute-error gate는
+  사용하지 않는다.
+- B33 section resultant는 source element label과 endpoint node label로 직접 대응한다.
+  MITC4 `U1/U2/U3`만 blocking이고 `UR1/UR2/UR3` 초과는 warning-only다.
 - parser, solver, HDF5 writer는 실패 원인을 구조화된 diagnostic으로 보고한다.
 - oneMKL, oneTBB, HDF5는 CMake에서 명시 탐지하고 실패 원인을 분류한다.
 - 대규모 모델 성능 최적화보다 V0의 명확성, 테스트 가능성, 검증 traceability를 우선한다.

docs/SOLVER_AGENT_DESIGN.md

@@ -1,415 +1,176 @@
-# 구조해석 솔버 개발 Agent 구성안
+# FESA Solver Agent Design
 
-## 목적
-이 문서는 Abaqus, Nastran과 같은 유한요소법 기반 구조해석 솔버를 개발하기 위한 AI Agent 운영 구성을 정의한다.
+## 목적과 범위
 
-이번 구성안은 ALL-FEM 논문의 구조를 확장하거나 재사용하는 계획이 아니다. 논문은 Agent 설계를 위한 참고 자료로만 사용하며, 본 프로젝트는 C++/MSVC 기반 독립 솔버 개발 워크플로우를 따른다.
+이 문서는 FESA 기능 개발을 조정하는 agent 계층, 8단계 workflow, gate와 산출물 계약을
+정의한다. `coordinator-agent`가 유일한 main agent이며, 나머지 10개 profile은 Coordinator가
+호출하는 sub-agent다. 모든 기능별 agent 산출물은 `docs/<feature-id>/`에 모은다.
 
-## 설계 원칙
-- 기능 요구조건, 이론 정식화, 코드 구현, 검증, 배포 역할을 분리한다.
-- 실행 가능성만으로 성공을 판단하지 않고, 레퍼런스 결과와 물리량을 비교해 기능 완료를 판정한다.
-- 테스트는 구현 전에 준비한다. 개발 대상 솔버 테스트와 레퍼런스 솔버 결과 비교 테스트를 함께 사용한다.
-- Abaqus나 Nastran을 Agent가 직접 실행하지 않는다. 기능이 선언한 기존 `.inp`와 실제
-  비교에 필요한 Abaqus CSV만 read-only 검증 기준으로 사용한다. Canonical naming,
-  README, metadata, version 또는 provenance는 기본 readiness 조건이 아니다.
-- FESA는 Abaqus와 독립적인 solver다. Agent는 Abaqus 내부 formulation, integration,
-  stabilization 또는 recovery equivalence를 요구하거나 추론하지 않는다.
-- 기본 개발 환경은 C++17 이상, MSVC, CMake, CTest이다.
-- 모든 기능은 tolerance 기준을 명시하고, 기준을 만족할 때만 배포 후보가 된다.
-- Harness 운영은 `docs/HARNESS_WORKFLOW.md`의 계획, 독립 Step 실행, PreToolUse/Stop 검증 계층을 따른다.
+이 workflow는 개발 운영 계약이다. Solver C++ 아키텍처, CMake/CTest target, Harness
+executor와 hook, Abaqus reference artifact 또는 승인된 FEM 기능 의미를 바꾸지 않는다.
 
-## Harness Step 실행 계약
+## Agent 계층
 
-계획과 구현 Agent는 작업 전 `docs/HARNESS.md`와 `docs/HARNESS_WORKFLOW.md`를 읽는다.
-Implementation Planning Agent는 multi-Step 초안을 사용자에게 승인받은 뒤 planning files만
-materialize하며 Step을 선택하거나 실행하지 않는다. Executor는 별도의 명시적 사용자 요청으로
-`scripts/execute.py`를 실행할 때 branch, pending Step 선택, retry, timestamps, commits, Step
-advancement와 phase status를 소유한다.
+### Coordinator Agent: main-agent orchestration
 
-Implementation Agent는 approved plan, materialized phase files, Executor-selected current
-`stepN.md`만 사용해 `RED -> observed failure -> minimal GREEN -> focused/full VERIFY`를 완료하고
-다음 Step을 시작하지 않는다. Agent가 쓸 수 있는 Harness metadata는 current Step의 `status`와
-`summary`, `error_message`, `blocked_reason` payload뿐이다. `.codex/hooks.json`이 PreToolUse
-interception과 Stop whole-project validation을 자동 실행하므로 hook entry point를 수동 실행해
-대체하지 않는다. 세부 schema와 recovery 절차는 `docs/HARNESS_WORKFLOW.md`를 따른다.
+Coordinator Agent는 기능 요청을 접수하고 다음 실행 loop를 소유한다.
 
-## 전체 Agent 구성
+```text
+INTAKE -> STATE AUDIT -> WORKLIST UPDATE -> SUB-AGENT DISPATCH
+       -> EVIDENCE CHECK -> GATE DECISION -> STATUS REPORT
+```
 
-### Coordinator Agent
-전체 개발 흐름을 관리하는 상위 조정 Agent이다.
+- `docs/<feature-id>/coordination.md`의 8단계 worklist와 현재 workflow state를 관리한다.
+- 한 번에 다음 유효 단계의 owner만 bounded task로 dispatch한다.
+- 반환된 산출물 경로, status, evidence와 blocker를 검토한 뒤에만 gate를 전환한다.
+- Specialist 판단이나 C++ 구현을 대신하지 않으며, evidence 없이 gate를 통과시키지 않는다.
+- 동일한 normalized failure classification이 두 번 발생하면 자동 재작업을 멈추고
+  `needs-user-decision` 또는 `blocked`로 전환한다.
+- Release evidence가 `ready-for-release`일 때 최종 closure를 기록한다.
 
-책임:
-- 기능 개발 요청을 단계별 작업으로 분해한다.
-- 각 Agent의 산출물을 연결하고 누락된 결정을 추적한다.
-- 요구조건, 정식화, 테스트, 구현, 검증, 배포 단계의 진행 상태를 관리한다.
-- 실패 시 어떤 Agent로 되돌릴지 결정한다.
+Worklist item은 `pending | in-progress | passed | needs-rework | blocked`만 사용한다.
+Sub-agent는 전달받은 단계와 산출물만 처리하며 peer를 호출하거나 다음 단계로 진행하지
+않는다. 완료 시 output paths, status, evidence summary와 blockers를 Coordinator에 반환한다.
 
-주요 산출물:
-- 기능별 개발 계획
-- 단계별 승인 상태
-- 실패 원인과 재작업 지시
+## 10개 sub-agent 역할
 
 ### Requirement Agent
-솔버 기능 요구조건을 정의하는 Agent이다.
 
-책임:
-- 해석 기능의 범위, 입력, 출력, 제약조건을 정의한다.
-- 대상 요소, 재료 모델, 경계조건, 하중 조건, 해석 타입을 명확히 한다.
-- 검증해야 할 물리량과 tolerance 기준을 정한다.
-
-주요 산출물:
-- 기능 요구조건 문서
-- acceptance criteria
-- 검증 물리량 목록
-
-예시 검증 물리량:
-- 절점 변위
-- 반력
-- 요소 내력
-- 응력
-- 변형률
-- 에너지 또는 잔차 기준
+검증 가능한 범위, 제외 범위, `shall` 요구조건, acceptance criteria, verification quantity와
+tolerance baseline을 `requirements.md`에 정의한다.
 
 ### Research Agent
-책, 논문, 매뉴얼, 공개 benchmark를 조사하는 Agent이다.
 
-책임:
-- 유한요소 정식화에 필요한 이론 자료를 수집한다.
-- 요소별 benchmark와 patch test 사례를 찾는다.
-- Abaqus/Nastran 결과와 비교할 수 있는 공개 예제 또는 문헌 해를 조사한다.
-- 자료의 신뢰도와 적용 범위를 평가한다.
-
-주요 산출물:
-- 연구자료 요약
-- 공식, 가정, 한계 정리
-- benchmark 후보 목록
+이론, solver manual, benchmark와 source reliability를 조사하고 확인된 사실과 추론 및
+applicability limit를 `research.md`에 분리해 기록한다.
 
 ### Formulation Agent
-코드 구현을 위한 유한요소 정식화를 작성하는 Agent이다.
 
-책임:
-- 약형, 형상함수, B matrix, constitutive matrix, 수치적분, 요소 강성 행렬을 정의한다.
-- 자유도 배치, 좌표계, 단위계, 부호 규약을 명확히 한다.
-- 선형/비선형, 정적/동적, small/large deformation 여부를 구분한다.
-- 구현 가능한 알고리즘 형태로 정식화를 정리한다.
-
-주요 산출물:
-- 요소별 정식화 문서
-- 알고리즘 의사코드
-- 수치적분 규칙
-- edge case와 singular case 목록
+Strong/weak form, kinematics, constitutive contract, shape functions, element equation, numerical
+integration과 output recovery를 구현 가능한 수치 계약으로 `formulation.md`에 작성한다.
 
 ### Numerical Review Agent
-정식화와 수치 알고리즘을 독립 검토하는 Agent이다.
 
-책임:
-- 수식의 차원, 부호, 좌표 변환, 적분 규칙을 검토한다.
-- rigid body mode, patch test, symmetry, positive definiteness 등 기본 수치 조건을 확인한다.
-- locking, hourglass mode, ill-conditioning 같은 위험을 식별한다.
-- 구현 전 정식화 오류를 줄인다.
-
-주요 산출물:
-- 정식화 리뷰 결과
-- 수치 위험 목록
-- 추가 테스트 요구사항
+독립적인 formulation 검토와 reference readiness를 하나의 merged gate로 소유한다. 차원,
+부호, DOF 순서, 좌표 변환, Jacobian, 적분, 대칭성, rigid-body mode, locking과 검증 위험을
+`numerical-review.md`에 기록한다. 동시에 exact reference input/CSV, blocking/warning quantity,
+source identity/component, row precheck와 승인 tolerance를 `reference-model.md`에 정의한다.
+두 문서가 모두 준비되어야 I/O 단계로 handoff할 수 있다.
 
 ### I/O Definition Agent
-솔버 입력과 출력 데이터 구조를 정의하는 Agent이다.
 
-책임:
-- mesh, node, element, material, section, boundary condition, load, step 입력 형식을 정의한다.
-- authoritative HDF5 result schema와 reference CSV comparison row schema를 정의한다.
-- Abaqus input file과 내부 입력 모델 사이의 대응 관계를 정리한다.
-- 결과 비교를 위해 FESA HDF5 dataset과 Abaqus reference CSV row의 ID/컴포넌트 규약을 맞춘다.
-
-주요 산출물:
-- 입력 데이터 schema
-- 출력 데이터 schema
-- HDF5 result schema
-- 결과 비교용 deterministic CSV view schema
-- 단위와 좌표계 규약
-
-### Reference Model Agent
-TDD와 검증에 사용할 기존 reference case를 inventory하는 Agent이다.
-
-책임:
-- 기능이 요구할 때만 테스트 모델 목적을 구분하고, 기본적으로 기존 case를 사용한다.
-- 기존 reference case의 목적, exact input/required CSV path, blocking/warning quantity와
-  tolerance를 inventory한다.
-- FESA HDF5 quantity와 source ID/component matching을 명시한다.
-- 테스트 모델이 요구조건을 실제로 검증하는지 확인한다.
-
-중요 제약:
-- Agent는 Abaqus를 직접 실행하지 않는다.
-- Abaqus 해석 결과 CSV는 사람이 생성하거나 별도 승인된 절차로 생성해 `reference/<model-id>/`에 저장한다.
-- Agent는 저장된 reference artifact만 사용해 비교한다.
-
-최소 reference case 구조:
-```text
-reference/
-  <case-dir>/
-    <declared-input>.inp
-    <declared-required-quantity>.csv
-```
-
-Directory와 filename은 제공된 값을 그대로 사용한다. Reference Model Agent는 canonical
-이름, README, metadata, provenance 또는 비교하지 않는 quantity CSV를 요구하지 않는다.
+승인된 Abaqus `.inp` subset, semantic model mapping, validation diagnostic, authoritative
+`results.h5` schema와 reference CSV row schema를 `io.md`에 정의한다. Numerical/reference
+gate가 확정한 logical quantity와 source identity를 최종 HDF5 dataset projection 및 CSV
+column mapping으로 연결하는 책임은 이 agent에 있다.
 
 ### Implementation Planning Agent
-코드 구현 전에 작업 단위와 테스트 순서를 설계하는 Agent이다.
 
-책임:
-- 요구조건과 정식화를 C++ 구현 작업으로 분해한다.
-- 먼저 작성할 단위 테스트, 통합 테스트, 레퍼런스 비교 테스트를 정의한다.
-- 기존 architecture와 ownership boundary에 맞춰 변경 파일을 제한한다.
-- 구현 Agent가 따라야 할 acceptance criteria를 제공한다.
-
-주요 산출물:
-- 구현 계획
-- 테스트 우선순위
-- 변경 파일 후보
-- acceptance checklist
-- 사용자 승인 전 multi-Step Harness 초안
-- 승인 후 `phases/index.json`, `phases/<task-name>/index.json`, 자기완결적 `stepN.md`
-
-필수 절차:
-- 구현 계획 요청에서 project-local `$harness` skill을 사용한다.
-- `docs/HARNESS.md`와 `docs/HARNESS_WORKFLOW.md`를 읽고 multi-Step 초안만 사용자에게
-  제시한다.
-- 한 Step은 하나의 layer/module만 다루고 각 Step에 prerequisite file, TDD
-  RED/GREEN/VERIFY, 정확한 MSVC/CMake/CTest command와 금지사항을 포함한다.
-- Step 초안을 먼저 사용자에게 제시한다. 승인 전에는 `phases/` 파일을 생성하지 않는다.
-- 승인 후에는 planning files만 materialize하고 Step 선택/실행은 하지 않는다. Harness executor
-  실행은 별도 사용자 요청이 있을 때만 수행한다.
+승인된 upstream bundle을 자기완결적 TDD Step으로 분해해 `implementation-plan.md`를 만든다.
+계획 요청에서는 project-local `harness`를 사용하고 사용자에게 multi-Step 초안을 먼저
+제시한다. 승인 후에만 phase index와 `stepN.md`를 materialize하며 executor는 별도 명시
+요청이 있을 때만 실행한다.
 
 ### Implementation Agent
-C++ 코드를 구현하는 Agent이다.
 
-책임:
-- `docs/HARNESS.md`, `docs/HARNESS_WORKFLOW.md`, materialized phase files와
-  Executor-selected current `stepN.md`를 읽고 현재 Step만 수행한다.
-- 테스트를 먼저 작성하고 실패를 확인한다.
-- 정식화와 I/O schema에 맞춰 최소 구현을 작성한다.
-- C++17 이상, MSVC, CMake, CTest 환경에서 동작하도록 구현한다.
-- 불필요한 일반화나 speculative abstraction을 피한다.
-- current Step의 `status`와 `summary`, `error_message`, `blocked_reason`만 기록한다.
-  branch, retry, timestamp, commit, advancement는 Executor에 맡긴다.
-- `.codex/hooks.json`으로 자동 등록된 PreToolUse와 Stop을 사용하며 hook script를 수동
-  검증 대체물로 실행하지 않는다.
-
-주요 산출물:
-- C++ source/header 변경
-- 테스트 코드
-- CMake/CTest 변경
-
-### Build/Test Executor Agent
-빌드와 테스트를 실행하는 Agent이다.
-
-책임:
-- `.harness/config.json` 또는 자동 감지 결과에 맞는 MSVC build/test 명령을 실행한다.
-- MSVC x64 Debug CMake configure/build/CTest 결과를 수집한다.
-- 실패 로그를 요약하고 Correction Agent에 전달한다.
-
-기본 CMake 검증 명령:
-```powershell
-cmake -S . -B .harness/build -A x64
-cmake --build .harness/build --config Debug
-ctest --test-dir .harness/build -C Debug --show-only=json-v1
-ctest --test-dir .harness/build -C Debug --output-on-failure
-```
-
-Preset 또는 직접 MSBuild 프로젝트는 `.harness/config.json`에 선언된 명령을 따른다.
-
-검증 대상:
-- CMake configure
-- MSVC Debug build
-- CTest
-- Harness self-test
+승인된 Step 단위로 `RED -> observed failure -> minimal GREEN -> focused VERIFY`를 수행하고,
+full MSVC x64 Debug build/CTest와 reference comparison까지 하나의 Implementation gate에서
+완료한다. `implementation-report.md`, `build-test.md`, `reference-comparison.md`를 각각 남기며
+최종 성공 status는 `pass-for-physics-evaluation`이다.
 
 ### Correction Agent
-빌드, 테스트, 런타임 실패를 수정하는 Agent이다.
 
-책임:
-- 실패 로그를 원인별로 분류한다.
-- 컴파일 오류, 링크 오류, 테스트 실패, 결과 비교 실패를 구분한다.
-- 최소 수정으로 실패를 해결한다.
-- 같은 실패가 반복되면 Coordinator Agent에 차단 상태를 보고한다.
-
-주요 산출물:
-- 수정 패치
-- 실패 원인 요약
-- 재검증 요청
-
-### Reference Verification Agent
-구현 솔버 결과와 저장된 레퍼런스 결과를 비교하는 Agent이다.
-
-책임:
-- 구현 솔버 `results.h5`의 rows와 `reference/<model-id>/`의 Abaqus reference CSV rows를 비교하고, FESA debug CSV views는 row identity 검토와 report evidence로만 사용한다.
-- 절점 변위, 반력, 요소 내력, 응력의 tolerance 만족 여부를 평가한다.
-- absolute tolerance, relative tolerance, norm-based tolerance를 구분해 적용한다.
-- 결과 차이가 tolerance 밖이면 원인 후보를 분류한다.
-
-주요 산출물:
-- reference comparison report
-- 실패한 물리량과 위치
-- 최대 오차, 평균 오차, norm 오차
+정규 단계가 아닌 on-demand rework sub-agent다. 같은 실패가 반복되거나 원인이 불명확할
+때 Coordinator가 호출한다. Upstream 계약을 바꾸지 않고 최소 수정과 재검증을 수행해
+`corrections.md`에 누적하고 Implementation Agent 재실행 요청을 Coordinator에 반환한다.
 
 ### Physics Evaluation Agent
-수치 결과가 물리적으로 타당한지 검토하는 Agent이다.
 
-책임:
-- 레퍼런스와 수치적으로 비슷해도 물리적으로 이상한 결과가 있는지 확인한다.
-- 변위 방향, 반력 평형, 응력 집중, 대칭 조건, rigid body mode를 검토한다.
-- 테스트 모델이 기능을 충분히 검증하지 못하면 추가 모델을 요구한다.
-
-주요 산출물:
-- 물리 검토 결과
-- 추가 검증 모델 요구사항
-- release 가능 여부 의견
+Reference comparison 이후 equilibrium, reaction consistency, displacement direction, symmetry,
+element force balance, stress/strain sanity, rigid-body mode와 model coverage를 검토해
+`physics-evaluation.md`에 기록한다.
 
 ### Release Agent
-기능 배포 준비를 담당하는 Agent이다.
 
-책임:
-- 요구조건, 테스트, 레퍼런스 비교, 물리 검토가 모두 통과했는지 확인한다.
-- 기능 문서와 release note를 정리한다.
-- 알려진 제한사항과 tolerance 기준을 기록한다.
+Requirements부터 physics까지 gate evidence와 acceptance traceability를 audit하고 known
+limitations, release notes draft와 release verdict를 `release.md`에 기록한다. 내부 readiness
+판정은 publish, deploy, package, tag 또는 external release 권한이 아니다.
 
-주요 산출물:
-- release checklist
-- 기능 문서
-- known limitations
+## 8단계 개발 프로세스
 
-## 개발 프로세스 매핑
+| 단계 | 담당 sub-agent | 필수 skill | `docs/<feature-id>/` 산출물 | 통과 조건 |
+| --- | --- | --- | --- | --- |
+| 1. 요구조건 | `requirement-agent` | `fesa-requirements-baseline` | `requirements.md` | 승인 범위, acceptance criteria, 검증량과 tolerance가 명확함 |
+| 2. 연구 | `research-agent` | `fesa-research-evidence`, 필요 시 `fem-theory-query` | `research.md` | 이론과 검증 evidence 및 적용 한계가 충분함 |
+| 3. 정식화 | `formulation-agent` | `fesa-formulation-spec` | `formulation.md` | 구현 가능한 수치 계약이 완성됨 |
+| 4. 수치 검토 + reference model 계약 | `numerical-review-agent` | `fesa-numerical-review` | `numerical-review.md`, `reference-model.md` | 두 문서가 함께 `pass-for-io-definition`임 |
+| 5. I/O 정의 | `io-definition-agent` | `fesa-io-contract` | `io.md` | Logical reference identity가 최종 HDF5 projection과 연결됨 |
+| 6. 구현 계획 + C++ 구현 + build/test + reference comparison | `implementation-planning-agent`, `implementation-agent` | `fesa-cpp-msvc-tdd`, 계획 시 project-local `harness` | `implementation-plan.md`, `implementation-report.md`, `build-test.md`, `reference-comparison.md` | TDD, full MSVC/CTest와 blocking reference comparison이 모두 통과함 |
+| 7. 물리 검토 | `physics-evaluation-agent` | `fesa-physics-sanity` | `physics-evaluation.md` | 물리 검토가 `pass-for-release-agent`임 |
+| 8. 배포 준비 | `release-agent` | `fesa-release-readiness` | `release.md` | `ready-for-release`이며 Coordinator가 closure를 기록함 |
 
-| 개발 과정 | 담당 Agent | 필수 산출물 |
-| --- | --- | --- |
-| 1. 솔버 기능 요구조건 정의 | Requirement Agent | 요구조건, acceptance criteria |
-| 2. 연구자료 조사 | Research Agent | 자료 요약, benchmark 후보 |
-| 3. 유한요소 정식화 | Formulation Agent, Numerical Review Agent | 정식화 문서, 리뷰 결과 |
-| 4. 입출력 데이터 정의 | I/O Definition Agent | 입력/출력 schema |
-| 5. Reference case 준비 | Reference Model Agent, Implementation Planning Agent | 기존 input/required CSV inventory, 비교 mapping, tolerance |
-| 6. 코드 구현 | Implementation Agent | C++ 코드, 테스트 |
-| 7. 레퍼런스 결과 비교 검증 | Reference Verification Agent, Physics Evaluation Agent | 비교 리포트, 물리 검토 |
-| 8. tolerance 만족 시 완료 | Coordinator Agent | 기능 완료 승인 |
-| 9. 기능 배포 | Release Agent | release checklist, 문서 |
-
-## 표준 작업 흐름
+## 8단계 workflow
 
 ```mermaid
 flowchart TD
-  A["기능 요청"] --> B["Requirement Agent"]
-  B --> C["Research Agent"]
-  C --> D["Formulation Agent"]
-  D --> E["Numerical Review Agent"]
-  E --> F["I/O Definition Agent"]
-  F --> G["Reference Model Agent"]
-  G --> H["Implementation Planning Agent"]
-  H --> I["Implementation Agent"]
-  I --> J["Build/Test Executor Agent"]
-  J --> K{"빌드/테스트 통과?"}
-  K -- "아니오" --> L["Correction Agent"]
-  L --> I
-  K -- "예" --> M["Reference Verification Agent"]
-  M --> N{"tolerance 만족?"}
-  N -- "아니오" --> L
-  N -- "예" --> P["Physics Evaluation Agent"]
-  P --> Q{"물리 검토 통과?"}
-  Q -- "아니오" --> L
-  Q -- "예" --> R["Release Agent"]
+  C["Coordinator: intake 및 worklist"] --> RQ["1. Requirement Agent"]
+  RQ --> RS["2. Research Agent"]
+  RS --> FM["3. Formulation Agent"]
+  FM --> NR["4. Numerical Review Agent<br/>numerical + reference gate"]
+  NR --> IO["5. I/O Definition Agent<br/>final HDF5 projection"]
+  IO --> IP["6. Implementation Planning Agent"]
+  IP --> IM["6. Implementation Agent<br/>TDD + MSVC/CTest + reference comparison"]
+  IM --> OK{"Implementation gate pass?"}
+  OK -- "yes" --> PH["7. Physics Evaluation Agent"]
+  OK -- "repeated or unclear failure" --> CR["Correction Agent<br/>on-demand rework"]
+  CR --> IM
+  PH --> RL["8. Release Agent"]
+  RL --> CL["Coordinator: closure"]
 ```
 
-## 검증 Gate
+Compile, link, ordinary test와 명확한 implementation-owned mismatch는 Implementation Agent가
+먼저 수정한다. 반복되거나 불명확한 실패만 Correction loop로 보낸다. Upstream contract
+gap은 Coordinator가 해당 owner 단계로 되돌린다.
 
-### Gate 1: 요구조건 승인
-통과 조건:
-- 대상 기능과 제외 범위가 명확하다.
-- 입력, 출력, tolerance, 검증 물리량이 정의되어 있다.
-- 레퍼런스 비교 방식이 정해져 있다.
+## Gate 계약
 
-### Gate 2: 정식화 승인
-통과 조건:
-- 요소 정식화와 수치적분 규칙이 문서화되어 있다.
-- 좌표계, 자유도, 부호 규약이 명확하다.
-- Numerical Review Agent가 주요 수치 위험을 검토했다.
+1. Requirements gate: `requirements.md`가 범위, acceptance criteria, 검증량과 tolerance를
+   고정한다.
+2. Research gate: `research.md`가 필요한 이론, benchmark와 applicability evidence를 제공한다.
+3. Formulation gate: `formulation.md`가 구현 가능한 수치 계약을 제공한다.
+4. Numerical/reference merged gate: `numerical-review.md`와 `reference-model.md`가 함께 통과한다.
+5. I/O gate: `io.md`가 source identity와 logical quantity를 authoritative `results.h5` dataset,
+   units, coordinates, component와 CSV column에 최종 투영한다.
+6. Implementation gate: 승인 계획, RED/GREEN/VERIFY evidence, full MSVC x64 Debug build/CTest,
+   deterministic row precheck와 blocking/warning reference comparison이 모두 통과한다.
+7. Physics gate: `physics-evaluation.md`가 `pass-for-release-agent`다.
+8. Release gate: `release.md`가 `ready-for-release`이고 Coordinator가 closure를 기록한다.
 
-### Gate 3: 테스트 준비 승인
-통과 조건:
-- 구현 전 실패해야 하는 테스트가 정의되어 있다.
-- 기능이 요구하는 기존 input/CSV pair와 blocking/warning quantity가 명확하다.
-- 필요한 source ID/component matching과 tolerance가 정의되어 있다.
+Reference comparison은 source identity와 component로 행을 결정적으로 대응시키며 missing,
+extra, duplicate 또는 nonfinite required row를 tolerance 전에 거부한다. CSV는 외부 reference이고
+FESA의 authoritative output은 `results.h5`다. 공통 tolerance 값, family 구성, row/RMS
+판정, report evidence와 변경 관리는 `docs/TOLERANCE.md`를 따른다.
 
-### Gate 4: 구현 검증
-통과 조건:
-- CMake/MSVC/CTest validation이 통과한다.
-- 단위 테스트와 통합 테스트가 통과한다.
-- 관련 C++ test file이 있고 같은 구현 Step 안에 RED 실패와 후속 GREEN 성공 증거가 있다.
-- Stop의 전체 MSVC build/test 검증이 통과한다.
+## 요구사항 단위 산출물 구조
 
-### Gate 5: 레퍼런스 검증
-통과 조건:
-- 기능이 blocking으로 선언한 Abaqus CSV quantity와 구현 solver HDF5 quantity가
-  tolerance 안에 있다.
-- Warning-only quantity는 결과와 경고가 리포트에 남고 pass/fail을 바꾸지 않는다.
-- Required source row/component의 누락, 추가, 중복 또는 nonfinite 값이 없다.
+```text
+docs/<feature-id>/
+├── coordination.md
+├── requirements.md
+├── research.md
+├── formulation.md
+├── numerical-review.md
+├── reference-model.md
+├── io.md
+├── implementation-plan.md
+├── implementation-report.md
+├── build-test.md
+├── reference-comparison.md
+├── corrections.md
+├── physics-evaluation.md
+└── release.md
+```
 
-### Gate 6: 배포 승인
-통과 조건:
-- 요구조건의 acceptance criteria가 모두 만족된다.
-- 문서와 release note가 준비되어 있다.
-- 남은 제한사항이 명확히 기록되어 있다.
-
-## FESA HDF5 / Abaqus Reference CSV 비교 기준
-
-권장 비교 방식:
-- authoritative 비교는 FESA `results.h5` rows와 Abaqus reference CSV rows 기준으로 수행한다.
-- FESA HDF5에서 추출한 deterministic CSV view는 debugging/review 보조 artifact로만 사용한다.
-- scalar 값: absolute tolerance와 relative tolerance를 함께 적용한다.
-- vector 값: component-wise 비교와 norm 비교를 함께 기록한다.
-- stress tensor: component-wise 비교를 기본으로 하고, 필요한 경우 principal stress 또는 von Mises stress를 추가 비교한다.
-- 반력: 전체 하중 평형과 개별 구속 자유도 반력을 모두 확인한다.
-
-권장 리포트 항목:
-- model name
-- compared quantity
-- number of compared rows
-- maximum absolute error
-- maximum relative error
-- RMS error
-- worst node or element id
-- pass/fail
-
-## 반복 실패 처리
-
-반복 실패가 발생하면 Correction Agent가 무한 수정 루프를 계속하지 않는다. 다음 중 하나로 분류해 Coordinator Agent에 보고한다.
-
-- 요구조건 불명확
-- 정식화 오류 가능성
-- reference artifact 오류 가능성
-- I/O schema 불일치
-- 구현 결함
-- tolerance 기준 부적절
-- 테스트 모델이 기능을 과도하게 또는 불충분하게 검증함
-
-Coordinator Agent는 분류 결과에 따라 Requirement, Formulation, I/O Definition, Reference Model, Implementation Agent 중 적절한 단계로 되돌린다.
-
-## 초기 적용 우선순위
-
-1. 선형 정적 해석의 최소 골격
-2. Isoparametric 3D Euler beam element
-3. 1D truss 또는 bar element
-4. 2D plane stress/plane strain element
-5. 3D solid element
-6. material model 확장
-7. nonlinear 또는 dynamic analysis 확장
-
-각 단계는 요구조건, 정식화, 테스트모델, 구현, 레퍼런스 비교, 배포 Gate를 독립적으로 통과해야 한다.
-
-## 운영 메모
-
-- Agent 산출물은 가능한 한 문서, 테스트, 비교 리포트 형태로 남긴다.
-- 사람이 제공한 Abaqus reference artifact는 현재 path/name 그대로 read-only로 사용한다.
-  별도 provenance, README 또는 metadata가 없다는 이유로 gate를 차단하지 않는다.
-- reference artifact가 바뀌면 기능 구현 변경과 같은 수준으로 검토한다.
-- 기능 구현 완료 판정은 build/test와 기능별 blocking reference validation 통과를 기준으로
-  한다. Physics evaluation과 release readiness는 별도 후속 배포 gate다.
+산출물이 필요하지 않았거나 기존 workflow에 없었던 경우 placeholder를 만들지 않는다.
+`corrections.md`는 실제 on-demand correction이 발생했을 때만 생성한다. Reference input/CSV는
+현재 path와 name 그대로 read-only로 사용하고, 문서 정리를 위해 rename 또는 보정하지 않는다.

docs/SOLVER_SKILL_DESIGN.md

@@ -1,204 +1,193 @@
-# FESA Solver Skill Rebuild Plan
+# FESA Solver Skill Design
 
 ## 목적
 
-이 문서는 FESA 유한요소 기반 구조해석 솔버 개발에 사용할 project-local Codex skill 구성을 정의한다.
+이 문서는 FESA의 8단계 feature workflow에서 사용하는 project-local skill 구성을 정의한다.
+Agent는 역할과 책임 단위이고 skill은 여러 agent가 재사용하는 절차, 품질 gate와 handoff
+단위다. 모든 feature별 skill output은 `docs/<feature-id>/`에 기록한다.
 
-Agent는 역할과 책임 단위이고, skill은 여러 Agent가 반복적으로 사용하는 절차와 검증 도구 단위다. 따라서 skill은 Agent와 1:1로 대응하지 않는다. 대신 요구조건, 연구, 정식화, I/O 계약, reference model, C++ TDD 구현, reference 비교, 물리 검토, release readiness처럼 솔버 개발 과정에서 반복되는 작업 흐름을 기준으로 구성한다.
+실제 실행 지침의 source of truth는 각 `.codex/skills/<skill-name>/SKILL.md`다. 이 문서는
+skill inventory와 책임 분리를 사람이 읽을 수 있게 설명한다.
 
 ## 설계 원칙
 
-- Skill은 `.codex/skills/<skill-name>/SKILL.md`에 둔다.
-- 각 skill은 필수 frontmatter `name`, `description`과 UI metadata `agents/openai.yaml`을 가진다.
-- Skill 본문은 agent TOML의 역할 설명을 반복하지 않고, 입력, 절차, 산출물, 금지사항, 품질 gate, handoff를 정의한다.
-- Skill은 `AGENTS.md`와 `docs/SOLVER_AGENT_DESIGN.md`를 공통 상위 기준으로 읽는다.
-- Abaqus, Nastran 또는 reference solver 실행은 skill 범위에 포함하지 않는다.
-- Abaqus reference CSV 파일 생성/수정은 skill 범위에 포함하지 않는다.
-- C++ 구현 관련 skill은 C++17 이상, MSVC, CMake, CTest, TDD 원칙을 따른다.
-- C++ 검증 명령은 `.harness/config.json` 또는 `docs/HARNESS.md`의 자동 감지 기본값을 따른다.
-- Harness Python 변경은 `uv run --with pytest python -m pytest -v -rs`로 검증한다.
+- 각 skill은 `SKILL.md`의 `name`, `description` frontmatter와 `agents/openai.yaml` UI metadata를
+  가진다.
+- Skill은 입력, workflow, output contract, boundaries, quality gate와 handoff를 정의한다.
+- 공통 상위 계약은 `AGENTS.md`와 `docs/SOLVER_AGENT_DESIGN.md`다.
+- Abaqus, Nastran 또는 다른 reference solver 실행과 reference CSV 생성/수정은 skill 범위가
+  아니다.
+- C++ 절차는 C++17 이상, MSVC x64 Debug, CMake/CTest와 TDD를 따른다.
+- 검증 명령은 `.harness/config.json`을 우선하고 없으면 `docs/HARNESS.md`의 자동 감지
+  기본값을 따른다.
+- Skill output path에는 agent별 폴더를 만들지 않고 `docs/<feature-id>/`만 사용한다.
 
-## Skill 구성
+## 8개 workflow skill
 
-| Skill | 적용 개발 과정 | 주요 사용자 Agent | 대표 산출물 |
+| Skill | 8단계 적용 과정 | 주요 사용자 sub-agent | `docs/<feature-id>/` 대표 산출물 |
 | --- | --- | --- | --- |
-| `fesa-requirements-baseline` | 1. 솔버 기능 요구조건 정의 | Requirement Agent, Coordinator Agent | `docs/requirements/<feature-id>.md` |
-| `fesa-research-evidence` | 2. 책, 논문 등 연구자료 조사 | Research Agent, Formulation Agent | `docs/research/<feature-id>-research.md` |
-| `fesa-formulation-spec` | 3. 코드 구현을 위한 유한요소 정식화 | Formulation Agent, Implementation Planning Agent | `docs/formulations/<feature-id>-formulation.md` |
-| `fesa-numerical-review` | 3. 정식화 독립 수치 검토 | Numerical Review Agent, Coordinator Agent | `docs/numerical-reviews/<feature-id>-review.md` |
-| `fesa-io-contract` | 4. 솔버 입출력 데이터 정의 | I/O Definition Agent, Reference Verification Agent | `docs/io-definitions/<feature-id>-io.md` |
-| `fesa-reference-models` | 5. TDD/reference 테스트모델 작성 | Reference Model Agent, Implementation Planning Agent | `docs/reference-models/<feature-id>-reference-models.md` |
-| `fesa-cpp-msvc-tdd` | 6. 코드 구현 및 build/test correction | Implementation Planning Agent, Implementation Agent, Build/Test Executor Agent, Correction Agent | implementation plan/report, build/test report, correction report |
-| `fesa-reference-comparison` | 7. reference solver 결과와 구현 solver 결과 비교 | Reference Verification Agent | `docs/reference-verifications/<feature-id>-reference-verification.md` |
-| `fesa-physics-sanity` | 8. tolerance 통과 후 물리 타당성 검토 | Physics Evaluation Agent | `docs/physics-evaluations/<feature-id>-physics-evaluation.md` |
-| `fesa-release-readiness` | 9. 솔버 기능 배포 준비 | Release Agent, Coordinator Agent | `docs/releases/<feature-id>-release.md` |
+| `fesa-requirements-baseline` | 1. 요구조건 | Requirement Agent | `requirements.md` |
+| `fesa-research-evidence` | 2. 연구 | Research Agent | `research.md` |
+| `fesa-formulation-spec` | 3. 정식화 | Formulation Agent | `formulation.md` |
+| `fesa-numerical-review` | 4. 수치 검토 + reference readiness | Numerical Review Agent | `numerical-review.md`, `reference-model.md` |
+| `fesa-io-contract` | 5. I/O 정의 | I/O Definition Agent | `io.md` |
+| `fesa-cpp-msvc-tdd` | 6. 구현 계획 + C++ 구현 + build/test + reference comparison | Implementation Planning Agent, Implementation Agent, Correction Agent | `implementation-plan.md`, `implementation-report.md`, `build-test.md`, `reference-comparison.md`, 필요 시 `corrections.md` |
+| `fesa-physics-sanity` | 7. 물리 검토 | Physics Evaluation Agent | `physics-evaluation.md` |
+| `fesa-release-readiness` | 8. 배포 준비 | Release Agent | `release.md` |
 
-## 개발 과정별 사용 예
-
-예시 기능: `isoparametric-3d-euler-beam`
-
-1. Requirement Agent는 `fesa-requirements-baseline`을 사용해 기능 범위, 제외 범위, 입력, 출력, 검증 물리량, tolerance, `Requirement Verification Matrix`를 작성한다.
-2. Research Agent는 `fesa-research-evidence`를 사용해 3D Euler beam element 이론, benchmark 후보, source reliability, applicability limits를 정리한다.
-3. Formulation Agent는 `fesa-formulation-spec`을 사용해 strong form, weak form, shape functions, B matrix, element stiffness, output recovery를 정리한다.
-4. Numerical Review Agent는 `fesa-numerical-review`를 사용해 rigid body modes, patch test, stiffness symmetry, Jacobian, locking 위험을 검토하고 `pass-for-implementation-planning` 여부를 판단한다.
-5. I/O Definition Agent는 `fesa-io-contract`를 사용해 지원할 Abaqus `.inp` keyword subset, `results.h5` schema, reference CSV comparison row schema를 정의한다.
-6. Reference Model Agent는 `fesa-reference-models`를 사용해 기존 input/required CSV
-   reference-case inventory와 비교 mapping을 작성한다.
-7. Implementation Planning Agent는 먼저 project-local `harness`를 사용해 사용자 승인용
-   multi-Step 초안을 만들고, 승인 후 phase files를 생성한다. 그 뒤 Implementation Agent와
-   함께 `fesa-cpp-msvc-tdd` 계약에 따라 RED/GREEN/VERIFY를 수행한다.
-8. Reference Verification Agent는 `fesa-reference-comparison`을 사용해 구현 solver `results.h5` rows와 Abaqus reference CSV rows를 tolerance 기준으로 비교한다.
-9. Physics Evaluation Agent는 `fesa-physics-sanity`를 사용해 global equilibrium, reaction consistency, displacement direction, symmetry, model coverage를 검토한다.
-10. Release Agent는 `fesa-release-readiness`를 사용해 gate evidence, acceptance traceability, known limitations, release notes draft를 작성한다.
+통합 후 FESA workflow skill은 위 8개가 전부다.
 
 ## Skill별 핵심 계약
 
 ### `fesa-requirements-baseline`
 
-- 기능 요청을 검증 가능한 요구조건 baseline으로 만든다.
+- 기능 요청을 검증 가능한 baseline으로 만든다.
 - `shall` 문장과 `FESA-REQ-<FEATURE>-###` id를 사용한다.
-- 모든 `must` 요구조건은 verification method와 acceptance criteria를 가져야 한다.
-- FEM 정식화, C++ 구현, Abaqus reference CSV 생성 또는 수정, release readiness 판단은 하지 않는다.
+- 모든 `must` 요구조건에 verification method와 acceptance criteria를 연결한다.
+- FEM 정식화, C++ 구현, reference value 생성 또는 release 판정은 하지 않는다.
 
 ### `fesa-research-evidence`
 
-- 연구 질문, source inventory, source reliability tier, benchmark 후보를 정리한다.
-- 검증된 사실과 추론을 분리한다.
-- source gap은 open issue로 남긴다.
-- FEM 정식화 확정이나 reference value 생성을 하지 않는다.
+- Research question, source inventory, reliability tier, benchmark 후보와 applicability limit를
+  정리한다.
+- 검증된 사실과 추론을 분리하고 source gap은 open issue로 남긴다.
+- FEM 정식화나 reference value를 확정하지 않는다.
 
 ### `fesa-formulation-spec`
 
-- strong form, weak form, discretization, kinematics, constitutive contract, element equations를 구분해 작성한다.
-- Jacobian, derivative transform, numerical integration, output recovery, numerical risks를 명시한다.
-- C++ API, parser, file ownership은 설계하지 않는다.
-- Numerical Review Agent 검토 전 최종 승인 상태로 두지 않는다.
+- Strong form, weak form, discretization, kinematics, constitutive contract와 element equation을
+  구분한다.
+- Jacobian, derivative transform, numerical integration, output recovery와 numerical risk를
+  명시한다.
+- C++ API, parser ownership 또는 file layout을 설계하지 않는다.
+- Numerical Review 전에는 최종 구현 승인 상태로 두지 않는다.
 
 ### `fesa-numerical-review`
 
-- 정식화를 수치 알고리즘 계약으로 독립 검토한다.
-- dimensions, signs, DOF ordering, coordinate transforms, Jacobian, integration rule, stiffness symmetry, rigid body modes, patch test, hourglass, locking을 확인한다.
-- `pass-for-implementation-planning`은 구현 계획 가능 상태만 의미한다.
-- 정식화 문서를 직접 수정하지 않는다.
-- 이후 Reference Model 문서, artifact naming, README, metadata, provenance 또는 portfolio가
-  없다는 이유로 formulation verdict를 실패시키지 않는다.
+이 skill은 numerical correctness와 reference readiness 절차를 함께 소유한다.
+
+```text
+FORMULATION REVIEW -> REFERENCE CASE INVENTORY -> CLASSIFY -> REPORT -> I/O HANDOFF
+```
+
+- Dimensions, signs, DOF order, coordinate transform, Jacobian, integration, stiffness symmetry,
+  rigid-body mode, patch test, hourglass와 locking을 독립 검토한다.
+- 기존 exact input/required CSV를 inventory하고 blocking/warning quantity, source identity와
+  component matching, missing/extra/duplicate/nonfinite row precheck와 승인 tolerance를 정한다.
+- `numerical-review.md`와 `reference-model.md`를 함께 산출한다.
+- Logical quantity와 source identity까지만 정의한다. 최종 HDF5 dataset projection은 I/O
+  Definition Agent에 handoff한다.
+- 두 문서가 준비된 `pass-for-io-definition`만 다음 단계 진행을 허용한다.
 
 ### `fesa-io-contract`
 
-- FESA solver input이 지원할 Abaqus `.inp` subset을 정의한다.
-- model data와 history data를 구분한다.
-- 내부 semantic model 계약, HDF5 output schema, reference CSV comparison row schema를 정의한다.
-- parser 구현이나 full Abaqus compatibility claim은 하지 않는다.
-
-### `fesa-reference-models`
-
-- 기능이 실제로 사용하는 기존 reference case를 inventory한다.
-- Exact input/required CSV path, case purpose, blocking/warning quantity, HDF5 projection,
-  source ID/component matching과 tolerance만 정의한다.
-- Required comparison file이 없을 때만 `needs-reference-artifacts`로 둔다. Canonical naming,
-  README, metadata, provenance와 비교하지 않는 quantity CSV는 요구하지 않는다.
+- 지원할 Abaqus `.inp` keyword subset, semantic model mapping과 validation rule을 정의한다.
+- Authoritative `results.h5` schema, units, coordinates, step/frame, row identity와 component를
+  정의한다.
+- Reference readiness의 logical quantity/source identity를 최종 HDF5 dataset projection과
+  CSV column mapping으로 연결한다.
+- Parser 구현이나 full Abaqus compatibility를 주장하지 않는다.
 
 ### `fesa-cpp-msvc-tdd`
 
-- C++ 구현을 `RED -> GREEN -> VERIFY` 순서로 수행한다.
-- C++ production 변경에는 관련 C++ test file이 있어야 한다.
-- 기본 검증 명령:
+이 skill은 implementation planning, TDD implementation, MSVC validation, failure correction과
+reference comparison 절차를 함께 소유한다.
+
+```text
+RED -> OBSERVED FAILURE -> MINIMAL GREEN -> FOCUSED VERIFY
+    -> FULL MSVC BUILD/CTEST -> ARTIFACT CHECK
+    -> COMPARE -> CLASSIFY -> REPORT
+```
+
+- Planning 시 project-local `harness`로 user-approved multi-Step plan을 만들고
+  `implementation-plan.md`를 산출한다.
+- C++ production 변경에는 관련 C++ test가 있어야 하며 targeted RED와 후속 GREEN evidence를
+  기록한다.
+- `.harness/config.json`이 선택한 MSVC x64 Debug build/CTest를 실행하고 명령, exit code,
+  duration, output tail과 failure classification을 `build-test.md`에 기록한다.
+- Reference artifact check 뒤 HDF5/CSV row를 source identity와 component로 대응시킨다.
+  Missing, extra, duplicate와 nonfinite required row는 tolerance 전에 실패하며 warning-only
+  quantity는 blocking result를 바꾸지 않는다.
+- Implementation-owned 실패를 먼저 수정한다. 반복되거나 불명확한 실패는 Coordinator를
+  통해 Correction Agent로 보내고 `corrections.md`에 재작업 evidence를 남긴다.
+- 성공 시 `implementation-report.md`, `build-test.md`, `reference-comparison.md`와
+  `pass-for-physics-evaluation` handoff를 반환한다.
+- Requirements, formulation, numerical/reference 계약, I/O 계약, reference artifact 또는
+  tolerance policy를 변경해 결과를 맞추지 않는다.
+
+기본 validation sequence는 다음과 같다.
 
 ```powershell
 cmake -S . -B .harness/build -A x64
 cmake --build .harness/build --config Debug
 ctest --test-dir .harness/build -C Debug -R <feature-or-label> --output-on-failure
+ctest --test-dir .harness/build -C Debug --show-only=json-v1
 ctest --test-dir .harness/build -C Debug --output-on-failure
 ```
 
-Preset 또는 직접 MSBuild 프로젝트는 `.harness/config.json`의 설정을 사용한다. Harness
-Python, Hook, agent config를 변경한 경우에는 다음 명령도 실행한다.
-
-```powershell
-uv run --with pytest python -m pytest -v -rs
-```
-
-- 실패는 `configure | compile | link | test | reference-comparison | harness | environment | upstream-contract`로 분류한다.
-- 요구조건, 정식화, I/O 계약, reference artifact, tolerance policy를 바꾸지 않는다.
-
-### project-local `harness`
-
-- Implementation Planning Agent가 구현 요청을 여러 독립 Step으로 분해할 때 사용한다.
-- 한 Step은 하나의 layer/module만 소유하고 prerequisite file, TDD RED/GREEN/VERIFY,
-  exact acceptance command와 구체적 금지사항을 포함한다.
-- 사용자에게 Step 초안을 먼저 제시한다. 승인 후에만 `phases/index.json`,
-  `phases/<task-name>/index.json`, `phases/<task-name>/stepN.md`를 생성한다.
-- 계획 작성과 executor 실행을 구분하며, `scripts/execute.py`는 별도 사용자 요청 없이
-  실행하지 않는다.
-
-### `fesa-reference-comparison`
-
-- `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT` 순서로 수행한다.
-- 선언된 input, required Abaqus CSV, `results.h5`, source ID/component mapping과 tolerance를
-  확인한다. Missing/extra/duplicate/nonfinite required row는 비교 전에 실패한다.
-- max absolute error, max relative error, RMS error, norm error, missing rows, extra rows를 보고한다.
-- Reference pass는 physics validation이나 release readiness를 의미하지 않는다.
-
 ### `fesa-physics-sanity`
 
-- Reference comparison 통과 후 물리 타당성을 검토한다.
-- global equilibrium, reaction consistency, displacement direction, symmetry, element force balance, stress/strain sanity, rigid body mode, model coverage를 확인한다.
+- Implementation gate 통과 후 equilibrium, reaction consistency, displacement direction,
+  symmetry, element force balance, stress/strain sanity, rigid-body mode와 model coverage를
+  검토한다.
 - 문서화된 물리 기대값이 없으면 pass를 선언하지 않는다.
 - `pass-for-release-agent`는 Release Agent 검토 가능 상태만 의미한다.
 
 ### `fesa-release-readiness`
 
-- `GATE AUDIT -> TRACEABILITY CHECK -> RELEASE DOCUMENTATION -> RELEASE VERDICT` 순서로 수행한다.
-- `pass-for-reference-verification`, `pass-for-physics-evaluation`, `pass-for-release-agent` evidence를 요구한다.
-- Known Limitations와 Release Notes Draft를 작성한다.
-- 사용자 명시 요청 없이 publish, deploy, package, tag, commit, external release를 수행하지 않는다.
+- `GATE AUDIT -> TRACEABILITY CHECK -> RELEASE DOCUMENTATION -> RELEASE VERDICT` 순서를 따른다.
+- Requirements부터 physics까지 동일 feature evidence와 필수 pass status를 확인한다.
+- Known limitations, Release Notes Draft와 `ready-for-release` 여부를 기록한다.
+- 사용자 요청 없이 publish, deploy, package, tag, commit 또는 external release를 수행하지
+  않는다.
 
-## Agent와 Skill 관계
+## Supporting skills
 
-| Agent | 주로 사용하는 Skill |
+다음은 8개 workflow skill 수에 포함하지 않는 supporting skill이다.
+
+- `fem-theory-query`: FEM wiki의 이론, benchmark, solver manual과 verification evidence 조회.
+- project-local `harness`: Implementation Planning의 Step 초안, 사용자 승인 후 phase
+  materialization과 별도 요청에 의한 executor 실행.
+- `review`: AGENTS, architecture, tests와 build requirement에 대한 repository change review.
+
+Implementation Planning Agent는 project-local `harness`를 반드시 사용한다. 사용자에게
+자기완결적 Step 초안을 먼저 제시하고 승인 후에만 `phases/index.json`,
+`phases/<task-name>/index.json`, `phases/<task-name>/stepN.md`를 생성한다. `scripts/execute.py`는
+별도 사용자 요청 없이 실행하지 않는다.
+
+## Agent와 skill 관계
+
+| Agent | 주로 사용하는 skill |
 | --- | --- |
-| Coordinator Agent | `fesa-requirements-baseline`, `fesa-reference-models`, `fesa-release-readiness` |
+| Coordinator Agent | 단계 owner가 반환한 skill evidence를 audit하고 dispatch/closure 관리 |
 | Requirement Agent | `fesa-requirements-baseline` |
-| Research Agent | `fesa-research-evidence` |
+| Research Agent | `fesa-research-evidence`, 필요 시 `fem-theory-query` |
 | Formulation Agent | `fesa-formulation-spec` |
-| Numerical Review Agent | `fesa-numerical-review` |
+| Numerical Review Agent | `fesa-numerical-review`, 필요 시 `fem-theory-query` |
 | I/O Definition Agent | `fesa-io-contract` |
-| Reference Model Agent | `fesa-reference-models` |
-| Implementation Planning Agent | project-local `harness`, `fesa-formulation-spec`, `fesa-reference-models`, `fesa-cpp-msvc-tdd` |
+| Implementation Planning Agent | project-local `harness`, `fesa-cpp-msvc-tdd` |
 | Implementation Agent | `fesa-cpp-msvc-tdd` |
-| Build/Test Executor Agent | `fesa-cpp-msvc-tdd` |
 | Correction Agent | `fesa-cpp-msvc-tdd` |
-| Reference Verification Agent | `fesa-reference-comparison`, `fesa-io-contract` |
 | Physics Evaluation Agent | `fesa-physics-sanity` |
 | Release Agent | `fesa-release-readiness` |
 
 ## 검증 기준
 
-Skill 구성은 실제 `.codex/skills/` 파일을 source of truth로 삼아 정적 계약과 repository
-pytest suite로 검증한다.
+Skill 구성은 실제 `.codex/skills/` 파일과 repository pytest suite를 기준으로 검증한다.
 
-검증 항목:
-
-- 10개 solver skill의 `SKILL.md` 존재 여부
+- 8개 FESA workflow skill의 `SKILL.md` 존재 여부
 - YAML frontmatter의 `name`, `description`
-- 공통 섹션: `Inputs`, `Workflow`, `Output Contract`, `Boundaries`, `Quality Gate`, `Handoff`
+- 공통 section: `Inputs`, `Workflow`, `Output Contract`, `Boundaries`, `Quality Gate`, `Handoff`
 - `AGENTS.md`와 `docs/SOLVER_AGENT_DESIGN.md` 참조
-- skill-specific 핵심 문구와 산출물 경로
-- `agents/openai.yaml` UI metadata
-- 이 문서가 아니라 실제 skill 파일이 기준이 되도록 `docs/SOLVER_SKILL_DESIGN.md`에 대한 skill 본문 참조 금지
+- Skill-specific procedure와 `docs/<feature-id>/` output contract
+- `agents/openai.yaml` UI metadata와 skill name reference
+- TOML 및 YAML metadata parseability
 
-검증 명령:
+Repository validation은 다음 명령을 사용한다.
 
 ```powershell
 uv run --with pytest python -m pytest -v -rs
 ```
 
-개별 skill schema를 점검할 때는 현재 Codex 설치에 포함된 `skill-creator` validator를
-사용하되 사용자 홈을 하드코딩한 경로를 프로젝트 계약으로 두지 않는다.
-
-## v1 범위
-
-- v1은 `SKILL.md`와 `agents/openai.yaml`만 포함한다.
-- 별도 `scripts/`, `references/`, `assets/`는 만들지 않는다.
-- 반복 사용 중 절차가 안정화되면 deterministic comparison script, reference artifact template, report template 같은 resource를 별도 후속 작업으로 분리한다.
-- 이 문서는 skill 구성을 설명하는 계획 문서이며, 실제 실행 지침의 source of truth는 각 `.codex/skills/<skill-name>/SKILL.md`이다.
+개별 skill schema는 현재 Codex 설치의 `skill-creator` validator로 점검하되 사용자 홈의
+절대 경로를 프로젝트 계약으로 두지 않는다.

docs/TOLERANCE.md

@@ -0,0 +1,179 @@
+# FESA Reference Tolerance Policy
+
+## 문서 정보
+
+- policy_id: `common-reference-tolerance`
+- status: `approved-and-implemented`
+- effective_date: `2026-08-18`
+- decision_record: `docs/ADR.md`의 ADR-022
+
+## 목적
+
+이 문서는 FESA 결과와 외부 reference 결과를 비교할 때 사용하는 공통 tolerance 값과
+검증 방법을 정의한다. 현재 기능뿐 아니라 앞으로 추가되는 요소, 재료, 해석, 하중 및
+경계조건의 reference comparison에도 같은 규칙을 적용한다.
+
+이 문서는 공통 수치 판정만 정의한다. 각 기능의 비교 대상, 단위, 좌표계, row identity,
+component 구성과 최종 판정 영향은 해당 기능의 `requirements.md`, `reference-model.md`와
+`io.md`에서 정의한다.
+
+## 적용 범위
+
+이 정책은 승인된 외부 reference 값과 FESA 공식 결과인 `results.h5`의 값을 비교하는 데
+사용한다. Parser schema, row identity, 물리 평형, 수렴성 및 정식화 검증에는 각각의 별도
+계약을 적용한다. 이러한 검증 실패를 수치 tolerance로 완화해서는 안 된다.
+
+## 공통 tolerance 값
+
+| 항목 | 값 | 의미 |
+| --- | ---: | --- |
+| Near-zero 비율 | `0.01` | Reference family 최대값의 1% 이하를 near-zero로 분류 |
+| 상대오차 tolerance | `0.05` | 일반 행의 상대오차를 5% 이하로 제한 |
+| Relative RMS tolerance | `0.01` | Family 전체 RMS 오차를 reference scale의 1% 이하로 제한 |
+
+독립적인 absolute-error tolerance는 사용하지 않는다. Absolute error는 near-zero 행을
+판정하고 결과를 진단하기 위해서만 사용한다.
+
+## Comparison family
+
+수치 scale은 개별 행이나 component마다 만들지 않고 comparison family마다 계산한다.
+하나의 family에는 다음 조건이 같은 값만 포함한다.
+
+- 같은 model 또는 reference case
+- 같은 step과 frame
+- 같은 logical quantity
+- 같은 단위 차원
+- 같은 좌표계
+- 같은 최종 판정 영향(`blocking` 또는 `warning-only`)
+
+각 기능 문서는 family 이름, 포함 component와 위 항목을 명시해야 한다. 서로 다른 단위,
+좌표계 또는 판정 영향을 가진 값은 같은 family에 포함할 수 없다.
+
+## 검증 방법
+
+### 1. 비교 입력 확정
+
+기능 문서가 승인한 reference artifact와 FESA `results.h5`를 사용한다. Reference artifact는
+비교를 통과시키기 위해 이름을 바꾸거나 값을 수정, 보정 또는 zero-clamp하지 않는다.
+
+### 2. Row 대응 및 사전검사
+
+Reference와 FESA 값을 기능 문서가 정의한 stable source identity와 component로 일대일
+대응시킨다. 다음 오류는 tolerance 계산 전에 comparison을 실패시킨다.
+
+- 필요한 파일, dataset 또는 component 누락
+- missing, extra 또는 duplicate row
+- source identity 불일치
+- 비유한 값(`NaN`, `Inf`)
+
+Tolerance는 schema 또는 identity 오류를 허용하는 수단이 아니다.
+
+### 3. Reference scale 계산
+
+Family의 reference 값 `r_i`만 사용해 scale `S`와 near-zero band `B`를 계산한다.
+
+\[
+S = \max_i |r_i|
+\]
+
+\[
+B = 0.01S
+\]
+
+FESA 값은 scale 계산에 사용하지 않는다. 임의의 absolute floor나 `max(1, S)`도 추가하지
+않는다.
+
+### 4. 행별 오차 판정
+
+FESA 값 `f_i`와 reference 값 `r_i`의 absolute error를 계산한다.
+
+\[
+e_i = |f_i-r_i|
+\]
+
+Reference 값이 near-zero band 안에 있으면 absolute error로 판정한다.
+
+\[
+|r_i| \le B \quad\Rightarrow\quad e_i \le B
+\]
+
+그 외 행은 상대오차로 판정한다.
+
+\[
+|r_i| > B \quad\Rightarrow\quad \frac{e_i}{|r_i|} \le 0.05
+\]
+
+경계값은 통과에 포함하며 모든 대응 행을 검사한다.
+
+### 5. Family Relative RMS 판정
+
+Family의 모든 absolute error로 RMS를 계산하고 reference scale로 정규화한다.
+
+\[
+\operatorname{relative\_rms} =
+\frac{\sqrt{\frac{1}{n}\sum_i e_i^2}}{S}
+\]
+
+다음을 만족해야 RMS 판정을 통과한다.
+
+\[
+\operatorname{relative\_rms} \le 0.01
+\]
+
+Family가 통과하려면 모든 행과 Relative RMS가 모두 통과해야 한다.
+
+### 6. Reference scale이 0인 경우
+
+`S = 0`이면 family의 모든 reference 값이 정확히 0이다.
+
+- 모든 FESA 값도 정확히 0이면 통과한다.
+- 하나라도 0이 아니면 실패한다.
+- 결과에는 비유한 metric 대신 `zero-reference-scale-nonzero-error`를 기록한다.
+
+### 7. 최종 판정
+
+- 사전검사 실패는 항상 전체 comparison을 실패시킨다.
+- `blocking` family의 행 또는 RMS 실패는 전체 comparison을 실패시킨다.
+- `warning-only` family의 실패는 warning을 기록하되 전체 blocking 판정은 변경하지 않는다.
+- Warning은 행 실패와 RMS 실패를 구분해 결정적인 순서로 기록한다.
+
+## 결과 기록
+
+Comparison 결과는 원본 값에서 판정을 재현할 수 있어야 한다. 최소한 다음 정보를
+기록한다.
+
+- 사용한 input, reference artifact와 FESA 결과 identity
+- 사전검사 결과
+- Family identity, component, row 수와 reference scale
+- 각 행의 FESA 값, reference 값, error, 적용 판정과 통과 여부
+- Family Relative RMS와 통과 여부
+- Blocking failure, warning과 전체 verdict
+
+동일한 입력을 반복 비교하면 row, family, warning과 결과 출력 순서가 같아야 한다.
+
+## 새 기능에 적용하는 방법
+
+새 기능의 comparator를 구현하기 전에 기능 문서에서 다음 항목을 승인한다.
+
+1. 사용할 input, reference artifact와 FESA HDF5 위치
+2. 비교할 quantity와 component
+3. 단위와 좌표계
+4. Stable source row identity와 일대일 mapping
+5. Comparison family 구성
+6. `blocking` 또는 `warning-only` 판정 영향
+
+구현 시에는 사전검사, tolerance 경계값, zero-scale, 행별 판정, Relative RMS와 결과 기록을
+테스트한다. 이후 기능별 reference comparison을 다시 실행해 evidence를 남긴다.
+
+## 변경 관리
+
+Tolerance 값, 계산식 또는 family 구성 규칙을 변경하려면 다음 절차를 따른다.
+
+1. 변경 이유와 영향을 검토하고 사용자 승인을 받는다.
+2. 이 문서와 ADR을 갱신한다.
+3. 경계값과 실패 동작을 테스트로 먼저 고정한다.
+4. 영향받는 comparator와 기능 문서를 수정한다.
+5. 전체 테스트와 영향받는 reference comparison을 다시 실행한다.
+
+기존 comparison report의 과거 수치를 소급 수정하지 않는다. 변경된 정책으로 새 evidence를
+생성하며 reference artifact 자체는 변경하지 않는다.

docs/build-test-reports/README.md

@@ -1,160 +0,0 @@
-# Build/Test Report 문서 작성 가이드
-
-이 디렉터리는 Build/Test Executor Agent가 작성하거나 제안하는 기능별 build/test 실행 리포트를 보관하는 위치다.
-
-Build/Test Executor Agent는 Implementation Agent 이후 독립적으로 C++/MSVC/CMake/CTest 검증을 실행하고, 실패를 분류해 다음 agent로 handoff한다. 이 agent는 source code, tests, CMake files, requirements, formulations, I/O contracts, reference artifacts, tolerance policies를 수정하지 않는다. 기본 build artifact는 `.harness/build/` 아래 생성된다.
-
-기본 문서명은 `docs/build-test-reports/<feature-id>-build-test.md` 형식을 사용한다.
-
-## Build/Test Executor Agent 역할
-
-수행한다:
-- `.harness/config.json`과 프로젝트 자동 감지 결과를 확인하고 같은 build/test 경로를 독립 실행한다.
-- implementation plan/report에 명시된 feature-specific CTest를 전체 검증 전에 실행한다.
-- Harness Python, Hook, agent config 변경이 포함되면 `uv run --with pytest python -m pytest -v -rs`를 실행한다.
-- CMake preset, 직접 MSBuild, 기본 CMake/MSVC x64 Debug 중 어떤 검증 경로가 사용되었는지 기록한다.
-- configure, compile, link, test, reference-comparison, harness, environment, upstream-contract 실패를 구분한다.
-- command, exit code, duration, stdout/stderr tail, failed test name을 요약한다.
-- 실패 원인에 따라 Implementation Agent, Correction Agent, Reference Verification Agent, Implementation Planning Agent 중 handoff 대상을 제안한다.
-
-수행하지 않는다:
-- source code를 수정하지 않는다.
-- tests를 수정하지 않는다.
-- CMake files를 수정하지 않는다.
-- requirements, formulations, I/O contracts, reference artifacts, tolerance policies를 수정하지 않는다.
-- Abaqus, Nastran 또는 reference solver를 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- release readiness, reference tolerance success, physics validation success를 승인하지 않는다.
-- 최종 reference verification report를 작성하지 않는다.
-
-## 실행 순서
-
-기본 순서는 implementation plan/report에 따라 다음 중 필요한 항목을 실행한다.
-
-```powershell
-cmake -S . -B .harness/build -A x64
-cmake --build .harness/build --config Debug
-ctest --test-dir .harness/build -C Debug -R <feature-or-label> --output-on-failure
-ctest --test-dir .harness/build -C Debug --show-only=json-v1
-ctest --test-dir .harness/build -C Debug --output-on-failure
-```
-
-Harness Python, Hook, agent config 변경이 검증 범위에 포함되면 다음 명령을 먼저 실행한다.
-
-```powershell
-uv run --with pytest python -m pytest -v -rs
-```
-
-프로젝트 선택 우선순위는 다음과 같다.
-
-1. `.harness/config.json`의 명시적 `projectType`
-2. 루트의 CMake metadata
-3. 루트의 단일 `.sln`
-4. 루트의 단일 `.vcxproj`
-
-기본 CMake/MSVC x64 Debug 명령은 다음과 같다.
-
-```powershell
-cmake -S . -B .harness/build -A x64
-cmake --build .harness/build --config Debug
-ctest --test-dir .harness/build -C Debug --show-only=json-v1
-ctest --test-dir .harness/build -C Debug --output-on-failure
-```
-
-CMake preset을 사용하면 configure/build/test preset과 `binaryDir`를 모두
-`.harness/config.json`에 지정한다. 직접 MSBuild는 solution/project와
-`msbuild.testCommand`를 지정한다. C/C++와 build metadata가 모두 없으면 검증 대상이
-없으므로 통과하지만, C/C++ 파일만 있고 build metadata가 없으면 오류다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Build/Test Report
-
-## Metadata
-- feature_id: <feature-id>
-- source_implementation_report: <path or N/A>
-- source_implementation_plan: docs/implementation-plans/<feature-id>-implementation-plan.md
-- status: pass-for-reference-verification | needs-correction | needs-environment-fix | needs-upstream-decision | blocked
-- owner_agent: build-test-executor-agent
-- date: <YYYY-MM-DD>
-
-## Execution Environment
-- os: <OS and version>
-- generator: Visual Studio 17 2022 | <observed generator>
-- platform: x64 | <observed platform>
-- config: Debug | <observed config>
-- build_dir: .harness/build | <configured/observed build dir>
-- harness_config: .harness/config.json | absent-defaults
-- project_selection: configured cmake | configured msbuild | auto CMake | auto MSBuild | no C/C++ project
-- command_discovery_path: CMake preset | direct MSBuild | default CMake/MSVC x64 Debug | no C/C++ project
-
-## Command Log Summary
-
-| order | command | exit_code | duration | stdout_stderr_tail |
-| --- | --- | --- | --- | --- |
-| 1 | uv run --with pytest python -m pytest -v -rs (when applicable) | <code or skipped> | <duration> | <tail summary> |
-| 2 | <config-resolved configure/build commands> | <code> | <duration> | <tail summary> |
-| 3 | <feature-specific test command when applicable> | <code or skipped> | <duration> | <tail summary> |
-| 4 | <config-resolved test discovery/full test commands> | <code> | <duration> | <tail summary> |
-
-## Validation Results
-
-| validation_stage | result | evidence |
-| --- | --- | --- |
-| harness self-test | pass | fail | skipped | <summary> |
-| configure | pass | fail | skipped | <summary> |
-| build | pass | fail | skipped | <summary> |
-| CTest | pass | fail | skipped | <summary> |
-| feature-specific tests | pass | fail | skipped | <summary> |
-
-## Failure Classification
-
-- classification: configure | compile | link | test | reference-comparison | harness | environment | upstream-contract | N/A
-- primary_failure: <short reason>
-- first_failed_command: <command or N/A>
-- evidence_tail: <short excerpt or summary>
-
-## Failed Test Inventory
-
-| test_name | label | command | failure_summary |
-| --- | --- | --- | --- |
-| <test name> | <label or N/A> | <command> | <summary> |
-
-## Handoff Recommendation
-
-| target_agent | reason | required_input |
-| --- | --- | --- |
-| Implementation Agent | <when implementation-owned failure is found> | <command log and failing test> |
-| Correction Agent | <when focused repair/rollback is needed> | <failure classification and changed files from implementation report> |
-| Reference Verification Agent | <when build/test passes and reference comparison report is next> | <passing command evidence> |
-| Implementation Planning Agent | <when plan/test contract is incomplete> | <missing or contradictory plan item> |
-
-## No-Change Assertion
-- source_files_modified: false
-- test_files_modified: false
-- cmake_files_modified: false
-- reference_artifacts_modified: false
-- notes: <observed no-change evidence or exception>
-
-## Open Issues
-- <environment gap, missing CMake preset, missing reference artifact, repeated failure, or upstream-contract issue>
-```
-
-## 상태 값
-
-- `pass-for-reference-verification`: build/test 검증이 통과해 Reference Verification Agent로 넘길 수 있다.
-- `needs-correction`: compile, link, ordinary test, implementation-owned failure가 있어 Correction Agent 또는 Implementation Agent 작업이 필요하다.
-- `needs-environment-fix`: MSVC, CMake generator, Python, path, permission 등 로컬 환경 문제로 검증이 막혔다.
-- `needs-upstream-decision`: implementation plan, requirements, formulation, I/O, reference artifact, tolerance policy가 불완전하거나 충돌한다.
-- `blocked`: 반복 실패 또는 외부 조건 때문에 사용자나 Coordinator Agent 결정 없이는 진행할 수 없다.
-
-## 품질 기준
-
-- 모든 실행 명령과 exit code를 기록해야 한다.
-- 실패 로그는 전체 원문을 복제하지 않고 마지막 핵심 구간과 실패 원인을 요약한다.
-- configure, compile, link, test, reference-comparison, harness, environment, upstream-contract 실패를 구분한다.
-- C/C++와 build metadata가 모두 없는 상황만 `no C/C++ project` 성공으로 기록한다. C/C++ 파일이 있는데 build metadata가 없으면 `environment` 또는 `configure` 실패로 기록한다.
-- 성공 판정은 build/test 통과까지만 의미한다.
-- reference tolerance, physics validation, release readiness는 판정하지 않는다.
-- upstream 계약 문제는 Implementation Agent에 임의 수정으로 넘기지 않고 적절한 upstream agent로 handoff한다.

docs/coordination/README.md

@@ -1,190 +0,0 @@
-# Coordination Report 문서 작성 가이드
-
-이 디렉터리는 Coordinator Agent가 작성하거나 제안하는 기능별 workflow coordination report를 보관하는 위치다.
-
-Coordinator Agent는 FESA solver 기능 개발의 전체 lifecycle에서 gate evidence, handoff, rework loop, blocker, user decision을 관리한다. 이 Agent는 specialist agent의 기술 판정을 대체하지 않고, 다음 agent가 어떤 입력으로 무엇을 산출해야 하는지 명확히 기록한다.
-
-기본 문서명은 `docs/coordination/<feature-id>-coordination.md` 형식을 사용한다.
-
-## Coordinator Agent 역할
-
-수행한다:
-- feature request를 `feature_id`, target capability, initial priority, expected first agent로 분류한다.
-- existing docs/reports/artifacts를 읽고 workflow state를 audit한다.
-- gate별 source evidence, missing evidence, contradictory evidence를 inventory로 만든다.
-- 다음 agent handoff package를 작성한다.
-- repeated failure와 blocker를 추적하고 escalation target을 정한다.
-- final workflow closure를 기록한다.
-
-수행하지 않는다:
-- source code를 수정하지 않는다.
-- tests를 수정하지 않는다.
-- CMake files 또는 build configuration을 수정하지 않는다.
-- build/test validation을 실행하지 않는다.
-- reference comparison을 실행하지 않는다.
-- physics evaluation을 실행하지 않는다.
-- requirements, formulations, I/O contracts, numerical review reports를 수정하지 않는다.
-- reference artifacts 또는 tolerance policies를 수정하지 않는다.
-- Abaqus, Nastran 또는 reference solver를 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- subagents를 자동 spawn하지 않는다.
-- release readiness를 독립적으로 승인하지 않는다.
-
-## 실행 순서
-
-Coordinator Agent는 다음 순서를 따른다.
-
-```text
-INTAKE -> STATE AUDIT -> GATE DECISION -> HANDOFF PACKAGE -> STATUS REPORT
-```
-
-`STATE AUDIT`에서는 다음 evidence를 확인한다.
-
-- Requirement Agent output
-- Research Agent output
-- Formulation Agent output
-- Numerical Review Agent output
-- I/O Definition Agent output
-- Reference Model Agent output
-- Implementation Planning Agent output
-- Implementation Agent report
-- Build/Test Executor Agent report
-- Correction Agent report
-- Reference Verification Agent report
-- Physics Evaluation Agent report
-- Release Agent report
-- Build/Test report의 `.harness/config.json` 또는 자동 감지 기반 MSVC build/test command evidence
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Coordination Report
-
-## Metadata
-- feature_id: <feature-id>
-- status: intake | needs-requirements | needs-research | needs-formulation | needs-numerical-review | needs-io-definition | needs-reference-model | needs-implementation-plan | ready-for-implementation | needs-build-test | needs-correction | needs-reference-verification | needs-physics-evaluation | needs-release | ready-for-release | completed | needs-user-decision | blocked
-- owner_agent: coordinator-agent
-- date: <YYYY-MM-DD>
-- source_docs: <docs/reports used>
-
-## Feature Request Summary
-- requested_feature: <short summary>
-- current_goal: <current coordination goal>
-- included_scope: <included scope>
-- excluded_scope: <excluded scope>
-- priority: <initial priority>
-
-## Current Workflow State
-
-| item | value | notes |
-| --- | --- | --- |
-| current_gate | <gate> | <notes> |
-| completed_outputs | <outputs> | <notes> |
-| missing_outputs | <outputs> | <notes> |
-| active_blockers | <blockers> | <notes> |
-| next_eligible_gate | <gate> | <notes> |
-
-## Gate Evidence Inventory
-
-| gate | owning_agent | expected_evidence | observed_evidence | status | notes |
-| --- | --- | --- | --- | --- | --- |
-| requirements | Requirement Agent | docs/requirements/<feature-id>.md | <path/status> | pass | fail | missing | <notes> |
-| research | Research Agent | docs/research/<feature-id>-research.md | <path/status> | pass | fail | missing | <notes> |
-| formulation | Formulation Agent | docs/formulations/<feature-id>-formulation.md | <path/status> | pass | fail | missing | <notes> |
-| numerical_review | Numerical Review Agent | docs/numerical-reviews/<feature-id>-review.md | <path/status> | pass | fail | missing | <notes> |
-| io_definition | I/O Definition Agent | docs/io-definitions/<feature-id>-io.md | <path/status> | pass | fail | missing | <notes> |
-| reference_model | Reference Model Agent | docs/reference-models/<feature-id>-reference-models.md | <path/status> | pass | fail | missing | <notes> |
-| implementation_planning | Implementation Planning Agent | docs/implementation-plans/<feature-id>-implementation-plan.md | <path/status> | pass | fail | missing | <notes> |
-| implementation | Implementation Agent | implementation report | <path/status> | pass | fail | missing | <notes> |
-| build_test | Build/Test Executor Agent | pass-for-reference-verification | <path/status> | pass | fail | missing | <notes> |
-| correction | Correction Agent | correction report when needed | <path/status> | pass | fail | missing | <notes> |
-| reference_verification | Reference Verification Agent | pass-for-physics-evaluation | <path/status> | pass | fail | missing | <notes> |
-| physics_evaluation | Physics Evaluation Agent | pass-for-release-agent | <path/status> | pass | fail | missing | <notes> |
-| release | Release Agent | ready-for-release | <path/status> | pass | fail | missing | <notes> |
-
-## Decision Log
-
-| date | decision_type | decision | source_evidence | rationale |
-| --- | --- | --- | --- | --- |
-| <YYYY-MM-DD> | gate transition | blocker | user decision | rework decision | <decision> | <path/status> | <reason> |
-
-## Next Agent Handoff
-
-| field | value |
-| --- | --- |
-| target_agent | <agent name> |
-| reason | <why this agent is next> |
-| required_inputs | <docs/reports/artifacts> |
-| expected_output | <expected report or artifact contract> |
-| acceptance_gate | <status or gate required after handoff> |
-| stop_condition | <when the agent should stop and hand back> |
-| missing_evidence | <missing inputs or decisions> |
-
-## Traceability Snapshot
-
-| requirement_id | gate | report | artifact | status | current_disposition |
-| --- | --- | --- | --- | --- | --- |
-| <req-id> | <gate> | <report path> | <artifact path> | <status> | <released | deferred | blocked | pending> |
-
-## Risk and Blocker Register
-
-| risk_or_blocker | category | owner | status | next_action |
-| --- | --- | --- | --- | --- |
-| <issue> | upstream ambiguity | repeated failure | reference artifact gap | environment blocker | <agent/user> | open | mitigated | blocked | <action> |
-
-## Rework Loop Control
-
-| failure_classification | correction_attempt_count | escalation_target | stop_condition | notes |
-| --- | --- | --- | --- | --- |
-| <classification> | <count> | <agent/user> | <condition> | <notes> |
-
-## No-Change Assertion
-- source_files_modified: false
-- test_files_modified: false
-- cmake_files_modified: false
-- reference_artifacts_modified: false
-- tolerance_policies_modified: false
-- notes: <observed no-change evidence or exception>
-
-## Open Issues
-- <unresolved user decision, missing evidence, contradictory report, or blocked transition>
-```
-
-## 상태 값
-
-- `intake`: 기능 요청은 들어왔지만 첫 handoff가 완료되지 않았다.
-- `needs-requirements`: Requirement Agent가 요구조건을 정의하거나 수정해야 한다.
-- `needs-research`: Research Agent가 source-backed research evidence를 제공하거나 수정해야 한다.
-- `needs-formulation`: Formulation Agent가 FEM 정식화를 작성하거나 수정해야 한다.
-- `needs-numerical-review`: Numerical Review Agent가 정식화를 검토하거나 재검토해야 한다.
-- `needs-io-definition`: I/O Definition Agent가 Abaqus input/output 계약을 정의하거나 수정해야 한다.
-- `needs-reference-model`: Reference Model Agent가 lightweight reference-case inventory,
-  required input/CSV presence, source-ID/component matching 또는 tolerance를 정의해야 한다.
-- `needs-implementation-plan`: Implementation Planning Agent가 TDD implementation plan을 작성하거나 수정해야 한다.
-- `ready-for-implementation`: implementation plan이 준비되었고 downstream 구현을 막는 upstream gate가 없다.
-- `needs-build-test`: 구현 이후 독립 Build/Test Executor 검증이 필요하다.
-- `needs-correction`: implementation-owned failure가 있어 Correction Agent가 필요하다.
-- `needs-reference-verification`: Build/Test evidence가 `pass-for-reference-verification`이다.
-- `needs-physics-evaluation`: Reference Verification report가 `pass-for-physics-evaluation`이다.
-- `needs-release`: Physics Evaluation report가 `pass-for-release-agent`이다.
-- `ready-for-release`: Release Agent report가 `ready-for-release`이고 final closure 기록이 필요하다.
-- `completed`: Release Agent report가 `ready-for-release`이고 Coordinator가 final workflow closure를 기록했다.
-- `needs-user-decision`: 사용자 또는 project decision 없이는 안전하게 진행할 수 없다.
-- `blocked`: 사용자 결정, 환경 변경, upstream correction 없이는 진행할 수 없다.
-
-## Handoff 원칙
-
-- 다음 단계 handoff는 source evidence, missing evidence, expected output, acceptance gate, stop condition을 포함해야 한다.
-- specialist agent가 소유한 기술 판정을 Coordinator가 대체하지 않는다.
-- `ready-for-implementation`은 Implementation Planning report가 `ready-for-implementation`일 때만 가능하다.
-- `needs-reference-verification`은 Build/Test evidence가 `pass-for-reference-verification`일 때만 가능하다.
-- `needs-physics-evaluation`은 Reference Verification report가 `pass-for-physics-evaluation`일 때만 가능하다.
-- `needs-release`는 Physics Evaluation report가 `pass-for-release-agent`일 때만 가능하다.
-- `completed`는 Release Agent report가 `ready-for-release`이고 final workflow closure가 기록된 경우에만 가능하다.
-- 동일 failure classification이 두 번 이상 반복되거나 upstream 계약 변경이 필요하면 `needs-user-decision` 또는 `blocked`로 전환한다.
-
-## 검증 기준
-
-- Coordinator Agent config와 문서 템플릿에 자동화된 Python 검증이 있으면 `uv run --with pytest python -m pytest -v -rs`로 실행한다.
-- C++ build/test evidence는 `.harness/config.json` 또는 Harness 자동 감지 기본값에 따른 Build/Test report에서 확인한다.
-- C/C++ 파일과 build metadata가 모두 없을 때만 `no C/C++ project` 성공을 허용한다. C/C++ 파일만 있고 build metadata가 없으면 통과 evidence로 사용하지 않는다.

docs/corrections/README.md

@@ -1,158 +0,0 @@
-# Correction Report 문서 작성 가이드
-
-이 디렉터리는 Correction Agent가 작성하거나 제안하는 기능별 correction report를 보관하는 위치다.
-
-Correction Agent는 Build/Test Executor Agent, Reference Verification Agent, Physics Evaluation Agent가 전달한 실패를 triage하고, implementation-owned failure만 최소 수정으로 복구한다. 이 agent는 source, header, test, CMake 수정은 수행할 수 있지만 requirements, formulations, I/O contracts, reference artifacts, tolerance policies는 수정하지 않는다.
-
-기본 문서명은 `docs/corrections/<feature-id>-correction.md` 형식을 사용한다.
-
-## Correction Agent 역할
-
-수행한다:
-- 실패 로그와 implementation report를 읽고 failure classification을 먼저 확정한다.
-- configure, compile, link, test, reference-comparison, harness, environment, upstream-contract 실패를 구분한다.
-- implementation-owned failure에 한해 source/header/test/CMake를 최소 수정한다.
-- 수정 후 targeted command를 먼저 실행하고 `.harness/config.json` 또는 자동 감지 기본값에 따른 전체 MSVC build/test를 실행한다.
-- Harness Python, Hook, agent config 관련 수정에서는 `uv run --with pytest python -m pytest -v -rs`도 실행한다.
-- 반복 실패 또는 upstream 계약 문제를 Coordinator Agent나 관련 upstream agent로 handoff한다.
-
-수행하지 않는다:
-- requirements를 수정하지 않는다.
-- formulations를 수정하지 않는다.
-- I/O contracts를 수정하지 않는다.
-- numerical review reports를 수정하지 않는다.
-- reference artifacts를 수정하지 않는다.
-- tolerance policies를 수정하지 않는다.
-- Abaqus, Nastran 또는 reference solver를 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- release readiness, reference tolerance success, physics validation success를 승인하지 않는다.
-- 최종 reference verification report 또는 physics validation report를 작성하지 않는다.
-
-## 실행 순서
-
-Correction Agent는 항상 다음 순서를 따른다.
-
-```text
-TRIAGE -> MINIMAL FIX -> VERIFY -> REPORT
-```
-
-기본 검증 명령은 다음과 같다.
-
-```powershell
-<targeted command that reproduced the failure>
-cmake -S . -B .harness/build -A x64
-cmake --build .harness/build --config Debug
-ctest --test-dir .harness/build -C Debug --show-only=json-v1
-ctest --test-dir .harness/build -C Debug --output-on-failure
-```
-
-Preset 또는 직접 MSBuild 프로젝트는 `.harness/config.json`의 설정을 따른다. Harness
-Python, Hook, agent config, Python validation behavior가 correction 범위에 포함될 때는
-`uv run --with pytest python -m pytest -v -rs`도 실행한다. Stop 검증은 응답 종료 전에
-같은 전체 프로젝트 검증을 다시 수행한다.
-
-## Failure Classification
-
-- `configure`: CMake configure, preset, generator, cache setup 실패
-- `compile`: C++ compilation 실패
-- `link`: linker, symbol resolution, target dependency 실패
-- `test`: CTest, unit, integration, parser/I/O, ordinary regression test 실패
-- `reference-comparison`: 저장된 reference artifact와 deterministic comparison 실패
-- `harness`: Python Harness test, PreToolUse/Stop Hook, config loading, discovery, adapter validation 실패
-- `environment`: MSVC, CMake, Python, path, permission, generator, local dependency 문제
-- `upstream-contract`: requirements, formulation, I/O, reference artifact, tolerance, implementation plan 불일치 또는 누락
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Correction Report
-
-## Metadata
-- feature_id: <feature-id>
-- source_failure_report: docs/build-test-reports/<feature-id>-build-test.md | <reference/physics report path>
-- source_implementation_report: <path or N/A>
-- source_implementation_plan: docs/implementation-plans/<feature-id>-implementation-plan.md
-- status: corrected-for-build-test | corrected-for-reference-verification | needs-build-test-rerun | needs-environment-fix | needs-upstream-decision | blocked
-- owner_agent: correction-agent
-- date: <YYYY-MM-DD>
-
-## Failure Triage
-- classification: configure | compile | link | test | reference-comparison | harness | environment | upstream-contract
-- first_failed_command: <command>
-- failed_target_or_test: <target, test, label, or N/A>
-- evidence_tail: <short relevant tail, not full log>
-- triage_decision: implementation-owned | environment-owned | upstream-owned | blocked
-
-## Root Cause Summary
-- root_cause_type: implementation defect | test defect | CMake registration issue | environment issue | upstream-contract issue
-- summary: <short explanation>
-- why_minimal_fix_is_allowed: <contract or failure evidence>
-
-## Correction Scope
-
-| file | change_type | reason | in_scope |
-| --- | --- | --- | --- |
-| include/fesa/<module>/<file>.hpp | source/header | <reason> | true |
-| src/<module>/<file>.cpp | source | <reason> | true |
-| tests/<module>/<file>_test.cpp | test | <reason> | true |
-| CMakeLists.txt | CMake | <reason> | true |
-
-Excluded files:
-- requirements: unchanged
-- formulations: unchanged
-- I/O contracts: unchanged
-- reference artifacts: unchanged
-- tolerance policies: unchanged
-
-## Verification Evidence
-
-| order | command | exit_code | result | evidence |
-| --- | --- | --- | --- | --- |
-| 1 | <targeted command> | <code> | pass | fail | <summary> |
-| 2 | <config-resolved full MSVC build/test commands> | <code> | pass | fail | <summary> |
-| 3 | uv run --with pytest python -m pytest -v -rs | <code or skipped> | pass | fail | skipped | <summary> |
-
-## Traceability
-
-| requirement_id | task_id | test_id | failing_command | corrected_file | acceptance_criterion |
-| --- | --- | --- | --- | --- | --- |
-| <req-id or N/A> | <task-id or N/A> | <test-id or N/A> | <command> | <file> | <criterion> |
-
-## Handoff Recommendation
-
-| target_agent | reason | required_input |
-| --- | --- | --- |
-| Build/Test Executor Agent | <independent full validation required> | <commands and correction summary> |
-| Reference Verification Agent | <reference comparison rerun required> | <corrected-for-reference-verification evidence> |
-| Physics Evaluation Agent | <physics sanity rerun required> | <corrected solver behavior evidence> |
-| Implementation Agent | <new implementation task required> | <unfixed implementation gap> |
-| upstream agent | <contract issue> | <required upstream decision> |
-| Coordinator Agent | <repeated failure or blocked state> | <classification history and stop condition> |
-
-## Stop Condition
-- repeated_failure: true | false
-- upstream_ambiguity: true | false
-- reference_artifact_gap: true | false
-- environment_blocker: true | false
-- next_required_decision: <decision or N/A>
-```
-
-## 상태 값
-
-- `corrected-for-build-test`: correction이 Build/Test Executor Agent 재실행 단계로 넘어갈 수 있다.
-- `corrected-for-reference-verification`: correction이 Reference Verification Agent 재실행 단계로 넘어갈 수 있다.
-- `needs-build-test-rerun`: targeted correction은 통과했지만 독립 build/test 재실행이 필요하다.
-- `needs-environment-fix`: 로컬 toolchain, generator, Python, path, permission 문제가 correction 또는 verification을 막는다.
-- `needs-upstream-decision`: upstream contract, reference artifact, tolerance, formulation ambiguity가 안전한 수정을 막는다.
-- `blocked`: 사용자 또는 Coordinator Agent 결정 없이는 안전하게 진행할 수 없다.
-
-## 품질 기준
-
-- 수정 전 failure classification을 기록해야 한다.
-- 모든 변경은 실패 로그 또는 implementation plan acceptance criterion에 trace되어야 한다.
-- production C++ 수정에는 관련 테스트 또는 기존 실패 테스트가 있어야 한다.
-- requirements, formulations, I/O contracts, reference artifacts, tolerance policies는 수정하지 않는다.
-- 실패 로그는 전체 원문을 복제하지 않고 핵심 tail과 원인 요약만 기록한다.
-- 동일 classification이 두 번 반복되면 Coordinator Agent 또는 관련 upstream agent로 handoff한다.
-- 성공 판정은 correction verification까지만 의미한다.
-- release readiness, reference tolerance success, physics validation success는 판정하지 않는다.

docs/cpp-object-oriented-modular-refactoring/build-test.md

@@ -0,0 +1,100 @@
+# C++ Object-Oriented Modular Refactoring Build/Test Report
+
+## Metadata
+
+- owner_agent: `implementation-agent`
+- feature_id: `cpp-object-oriented-modular-refactoring`
+- report_status: `passed`
+- date: `2026-08-16`
+- workspace: `C:\git\FESADev\.worktrees\cpp-object-oriented-modular-refactoring`
+- branch: `feat-cpp-object-oriented-modular-refactoring`
+- head: `f84ebb541f4717ab8300cf0d80497e1277d3bb48`
+- `.harness/config.json`: absent; Harness/CMake defaults and Step 24 explicit commands used
+- build generator: `Visual Studio 18 2026`
+- platform/configuration: `x64` / `Debug`
+- compiler observed by configure: `MSVC 19.51.36252.0`
+- inherited environment note: `FESA_HARNESS_CODEX_SANDBOX=danger-full-access`
+
+## Execution environment
+
+Required dependency paths all existed:
+
+| Path | Status |
+| --- | --- |
+| `C:/git/googletest` | found |
+| `C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl` | found |
+| `C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb` | found |
+| `C:/Program Files/HDF_Group/HDF5/2.1.1/cmake` | found |
+
+LLVM tools:
+
+- `clang-format version 22.1.8`
+- `clang-tidy LLVM version 22.1.8`
+- `clang-tidy --verify-config`: `No config errors detected.`
+
+## Command log summary
+
+| Command | Exit | Duration | Output tail / result |
+| --- | ---: | ---: | --- |
+| `uv run --with pytest python -m pytest -v -rs` | 1 | 2.700s | 20 passed, 1 failed: `test_invoke_codex_uses_utf8_for_unicode_prompt` saw inherited sandbox override `danger-full-access` instead of default `workspace-write`. Classified as environment-specific diagnostic, not product failure. |
+| Clean child process without `FESA_HARNESS_CODEX_SANDBOX`; `uv run --with pytest python -m pytest -v -rs` | 0 | 0.751s | `21 passed in 0.13s`. |
+| `clang-format --dry-run --Werror` over `@(rg --files include src tests -g "*.h" -g "*.cpp")` | 1 | 0.892s | Reported four formatting findings in `src/fesa/math/sparse_matrix.cpp`, `tests/unit/math/sparse_matrix_test.cpp`, `tests/unit/solvers/linear/linear_solver_test.cpp`, `tests/unit/solvers/linear/mkl_pardiso_solver_test.cpp`. |
+| `clang-format -i` on the four reported files | 0 | tool wall 0.5s | Normalized formatting/stat state; `git diff --raw` and `git diff --numstat` remained empty for those files. |
+| `clang-format --dry-run --Werror` over 163 C++ files | 0 | 0.888s | `CPP_FILE_COUNT: 163`; no violations. |
+| `clang-tidy --config-file=.clang-tidy <publicHeader> -- -x c++ -std=c++17 -Iinclude` | 0 | 41.822s | `PUBLIC_HEADER_COUNT: 46`; 45 diagnostics were required trailing-underscore header guards and two were `const` parameter classifications; zero other naming diagnostics. |
+| Read-only production Doxygen/header-guard and test-tag scan | 0 | 0.302s | 63 production headers, 0 missing guards, 0 missing Doxygen-contract headers, and 0 test files with imposed Doxygen tags. |
+| `.hpp` scan under `include src tests` | 0 | 0.162s | `LEGACY_HPP_COUNT: 0`; `rg` returned 1 for empty result but count-based AC passed. |
+| Required dependency path check | 0 | 0.165s | All four declared dependency paths found. |
+| `cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 "-DFESA_GTEST_SOURCE_DIR=C:/git/googletest" "-DMKL_DIR=C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl" "-DTBB_DIR=C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb" "-DHDF5_DIR=C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"` | 0 | 6.204s | Configure/generate complete; MKL 2026.1.0 found; build files written to `.harness/build`. |
+| `cmake --build .harness/build --config Debug --target fesa_tests` | 0 | 9.639s | `fesa_solver.lib`, `fesa_integration_tests.exe`, `fesa_reference_tests.exe`, and `fesa_unit_tests.exe` built. |
+| `ctest --test-dir .harness/build -C Debug --show-only=json-v1` | 0 | 0.235s | `DISCOVERED_TEST_COUNT: 206`. |
+| `ctest --test-dir .harness/build -C Debug --output-on-failure` | 0 | 9.908s | `100% tests passed out of 206`; labels: unit 182, integration 11, reference 13. |
+| `ctest --test-dir .harness/build -C Debug -R "B33ReferenceComparison|Mitc4S4Reference" --output-on-failure` | 0 | 1.260s | `100% tests passed out of 3`. |
+| `git diff --check` | 0 | 0.199s | `<no whitespace errors>`. |
+
+## Validation results
+
+| Validation | Result |
+| --- | --- |
+| Harness Python/policy tests | Pass in clean child process: 21/21. |
+| LLVM tool availability/config | Pass. |
+| clang-format repository dry-run | Pass: 163 files. |
+| clang-tidy selected public-header check | Pass: 46 headers, exit 0, zero naming diagnostics outside the two documented contract exceptions. |
+| Production Doxygen/header-guard policy | Pass: 63/63 headers; tests contain no imposed Doxygen boilerplate tags. |
+| Header extension policy | Pass: 0 `.hpp` under `include`, `src`, `tests`. |
+| Fresh MSVC x64 Debug configure | Pass. |
+| MSVC x64 Debug build | Pass: `fesa_tests`. |
+| CTest discovery | Pass: 206 tests. |
+| Full CTest | Pass: 206/206. |
+| Approved reference suites | Pass: 3/3. |
+| Reference tree no-change | Pass before and after compare: `git diff --exit-code 1e5758f -- reference` exit 0. |
+
+## Failure classification and failed test inventory
+
+Blocking classification: `none`.
+
+Nonblocking diagnostics:
+
+- `environment`: inherited `FESA_HARNESS_CODEX_SANDBOX=danger-full-access` caused the raw pytest command to fail one default-sandbox test. The same pytest command passed 21/21 in a child process with only that override removed.
+- `style`: initial clang-format dry-run reported four whitespace/line-ending findings. Formatting normalization introduced no tracked source-content diff, and the rerun passed.
+- `static-policy`: clang-tidy reported header-guard trailing underscores and two
+  `const` parameters because its generic macro/constant categories differ from the
+  higher-priority FESA guard and parameter conventions. It reported no other naming
+  diagnostics, and the required command exited 0.
+
+Failed blocking tests after clean verification: none.
+
+## Handoff recommendation
+
+Proceed to Physics Evaluation Agent. Build/test evidence is sufficient for the final
+Implementation-owned gate; no compile, link, test, reference-comparison, or
+environment blocker remains.
+
+## No-change assertion
+
+No production behavior, reference artifact, reference tolerance, comparator contract,
+or generated Doxygen output was changed in Step 24.
+
+## Open issues
+
+None blocking.

docs/cpp-object-oriented-modular-refactoring/implementation-plan.md

@@ -0,0 +1,410 @@
+# C++ Object-Oriented Modular Refactoring Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use
+> `superpowers:subagent-driven-development` (recommended) or
+> `superpowers:executing-plans` to implement this plan task-by-task. Steps use
+> checkbox (`- [ ]`) syntax for tracking.
+>
+> In FESA, those task-by-task semantics are mediated by the project Harness. Do not
+> invoke an implementation skill or select a Step directly; a separate user request
+> must start `scripts/execute.py`, which selects exactly one pending Step.
+
+**Goal:** Preserve the current B33, MITC4, and linear-static numerical and external
+contracts while converting the FESA C++ production code to explicit object-oriented
+boundaries, focused modules, shared utilities, Google C++ style, and production-only
+Doxygen documentation.
+
+**Architecture:** Domain owns immutable polymorphic semantic definitions through
+`std::unique_ptr` and stable `EntityIndex` positions. `ElementDefinition` remains
+separate from runtime numerical `Element`, and load, boundary-condition, analysis,
+material, and property abstractions each have independent hierarchies. Existing
+deterministic assembly, result identity, HDF5 schema, and reference comparison
+contracts remain unchanged.
+
+**Tech Stack:** C++17, MSVC x64 Debug, CMake, CTest, GoogleTest, Intel oneMKL,
+Intel oneTBB, HDF5, clang-format, clang-tidy, and optional Doxygen configuration.
+
+## Global Constraints
+
+- Follow `/docs/CODINGSTYLE.md` and the official Google C++ Style Guide baseline.
+- Use PascalCase for every C++ function and accessor; use `.h` production headers
+  with full-path include guards; retain `.cpp` as the FESA source-file exception.
+- Add Doxygen comments only to production code. Do not add Doxygen coverage to tests.
+- Keep C++17 and MSVC x64 Debug compatibility and add no compiler warnings under
+  `/W4 /WX`.
+- Preserve the approved B33 and MITC4 formulations, signs, units, coordinate systems,
+  reduction order, result row identity, HDF5 schema, tolerances, and reference files.
+- Do not implement MITC3, solid elements, dynamics, eigenvalue analysis, response
+  spectrum, random vibration, density, plasticity, anisotropy, distributed load, body
+  force, or MPC behavior.
+- Do not expose MKL, TBB, HDF5, Win32, or vendor integer types from public solver-core
+  headers.
+- Every C++ production change requires a related C++ test and an in-Step
+  `RED -> observed failure -> minimal GREEN -> focused/full VERIFY` cycle.
+- Do not run `scripts/execute.py` until the user gives a separate explicit execution
+  request.
+- Doxygen comments and `Doxyfile` configuration are in scope; generated Doxygen
+  output is deferred and is not a blocking command for this phase.
+
+---
+
+## 1. Metadata
+
+| Field | Value |
+| --- | --- |
+| `feature_id` | `cpp-object-oriented-modular-refactoring` |
+| `source_requirement` | `/docs/superpowers/specs/2026-08-16-cpp-object-oriented-modular-refactoring-design.md` |
+| `source_research` | Existing repository duplication and ownership audit captured by the approved design; no new FEM research is required |
+| `source_formulation` | `/docs/linear-static-3d-euler-beam/formulation.md`; `/docs/linear-static-mitc4-shell/formulation.md` |
+| `source_numerical_review` | `/docs/linear-static-3d-euler-beam/numerical-review.md`; `/docs/linear-static-mitc4-shell/numerical-review.md` |
+| `source_io_definition` | `/docs/linear-static-3d-euler-beam/io.md`; `/docs/linear-static-mitc4-shell/io.md` |
+| `source_reference_models` | `/docs/linear-static-3d-euler-beam/reference-model.md`; `/docs/linear-static-mitc4-shell/reference-model.md` |
+| `status` | `ready-for-implementation` |
+| `owner_agent` | `implementation-planning-agent` |
+| `date` | `2026-08-16` |
+
+## 2. Readiness Check
+
+- The written refactoring design and the 25-Step draft were explicitly approved on
+  2026-08-16.
+- B33 and MITC4 requirements, formulations, numerical reviews, I/O projections, and
+  reference contracts already exist and remain upstream read-only inputs.
+- Required reference inputs and CSVs are present under
+  `/reference/cantilever beam/` and `/reference/shell/`.
+- `clang-format.exe` and `clang-tidy.exe` are present at
+  `C:/Program Files/LLVM/bin/`; the current long-lived process PATH need not contain
+  that directory because the plan uses the absolute paths.
+- Doxygen generation is intentionally deferred by user decision. The implementation
+  still adds production comments and a warning-strict `Doxyfile` for later use.
+- No missing formulation, tolerance, HDF5 projection, or artifact decision prevents
+  implementation planning.
+
+## 3. Implementation Scope
+
+### Included
+
+- Repository policy/tooling and Implementation Agent enforcement.
+- Mechanical `.hpp` to `.h`, header guard, PascalCase, formatting, and production
+  Doxygen conversion in reviewable module slices.
+- Shared `Vector3`, dense-BLAS internal adapter, ASCII utilities,
+  `SourceTargetResolver`, and owner-based DOF invariant validation.
+- Independent abstract boundaries for material, element property, semantic element
+  definition, runtime element, load, boundary condition, and analysis.
+- Current concrete B33, MITC4, isotropic linear elasticity, beam/shell property,
+  concentrated nodal load, prescribed displacement, and linear-static behavior.
+- Responsibility-based splits of domain mapping, result recovery, and HDF5 writing.
+- Full MSVC/CTest and existing B33/MITC4 external reference verification.
+
+### Excluded and non-goals
+
+- New physics, input keywords, output datasets, tolerances, reference artifacts, or
+  runtime performance optimization.
+- A common root base shared by unrelated element, load, material, and analysis types.
+- A giant material interface containing density, plasticity, and anisotropy options.
+- Registry/plugin frameworks, global static registration, speculative `Clone()`, or
+  unnecessary shared ownership.
+
+## 4. Refactoring Requirements
+
+| ID | Requirement |
+| --- | --- |
+| `R-PRESERVE-001` | Current B33/MITC4/linear-static numerical and external results shall remain unchanged within their approved contracts. |
+| `R-STYLE-001` | Production and test C++ shall use approved Google-style naming and formatting; production headers shall use `.h` and header guards. |
+| `R-DOC-001` | Production functions and classes shall carry useful Doxygen contracts; tests shall not require Doxygen comments. |
+| `R-DUP-001` | Repeated fixed-size 3D vector operations shall be implemented once by `Vector3`. |
+| `R-DUP-002` | Repeated dense-BLAS conversion/copy, ASCII/source resolution, and DOF invariant logic shall have one owner. |
+| `R-MODEL-001` | Material, element-property, and element-definition semantic objects shall have independent abstractions and Domain-owned stable lifetime. |
+| `R-ELEMENT-001` | Semantic `ElementDefinition` and runtime numerical `Element` shall remain separate and be connected by a fail-closed factory. |
+| `R-PIPELINE-001` | DofManager, SparseAssembler, and ResultRecovery shall consume runtime `Element` interfaces without scattered B33/MITC4 type branches. |
+| `R-LOAD-001` | A `Load` shall emit ordered contributions and only `LoadAssembler` shall accumulate the global vector. |
+| `R-BC-001` | A `BoundaryCondition` shall emit definitions and an essential-constraint policy shall enforce prescribed displacement. |
+| `R-ANALYSIS-001` | `Analysis` shall expose only `Run()` and `LinearStaticAnalysis` shall own its approved lifecycle. |
+| `R-MODULE-001` | Domain mapping, recovery, and HDF5 writing shall be split by their approved responsibilities. |
+| `R-AGENT-001` | Implementation Agent shall read `/docs/CODINGSTYLE.md` as a mandatory global input. |
+| `R-SCOPE-001` | No excluded future feature or runtime-performance change shall be introduced. |
+
+## 5. Work Breakdown
+
+| Task | Name | Depends on | Deliverable |
+| --- | --- | --- | --- |
+| `T00` | coding-style-agent-contract | none | Agent profile and Python contract enforce `CODINGSTYLE.md`. |
+| `T01` | cpp-style-tooling | `T00` | clang-format/tidy configuration and deferred Doxygen configuration. |
+| `T02` | architecture-boundaries | `T00` | Architecture and ADR record the approved responsibility graph. |
+| `T03` | foundation-google-style | `T01` | Core/math/linear-solver APIs use the approved style. |
+| `T04` | model-element-google-style | `T03` | Model and current element APIs use the approved style. |
+| `T05` | solver-workflow-google-style | `T04` | FEM/assembly/constraint/analysis/result APIs use the approved style. |
+| `T06` | io-application-google-style | `T05` | I/O, application, and test helper APIs use the approved style. |
+| `T07` | vector3-value-type | `T03` | Tested fixed-size vector value type. |
+| `T08` | element-geometry-vector3 | `T04`, `T07` | Element/model geometry duplicate helpers removed. |
+| `T09` | result-io-vector3 | `T06`, `T08` | Result/I/O vector duplicate helpers removed. |
+| `T10` | dense-blas-adapter | `T03` | Matrix/Vector share private MKL conversion and copy helpers. |
+| `T11` | source-target-resolver | `T06` | Shared ASCII and source-target resolution module. |
+| `T12` | material-property-hierarchy | `T04` | Independent semantic material and property abstractions. |
+| `T13` | element-definition-domain | `T11`, `T12` | Domain-owned polymorphic semantic element definitions. |
+| `T14` | runtime-element-factory | `T08`, `T13` | Runtime element abstraction and fail-closed factory. |
+| `T15` | generic-dof-manager | `T14` | DofManager consumes element DOF layouts and owns invariant checks. |
+| `T16` | generic-sparse-assembler | `T15` | SparseAssembler consumes element stiffness contributions. |
+| `T17` | generic-result-recovery | `T16` | ResultRecovery consumes element result bundles. |
+| `T18` | load-hierarchy | `T11`, `T15` | Ordered load contribution hierarchy. |
+| `T19` | boundary-condition-policy | `T15` | Constraint definition hierarchy and essential policy. |
+| `T20` | analysis-hierarchy | `T17`, `T18`, `T19` | Minimal Analysis base and unchanged linear-static lifecycle. |
+| `T21` | domain-mapper-modules | `T11`, `T13`, `T18`, `T19` | Mapper split by semantic responsibility. |
+| `T22` | result-recovery-modules | `T17` | Recovery split into global, beam, shell, and commit responsibilities. |
+| `T23` | hdf5-writer-modules | `T17`, `T22` | HDF5 writer split without schema changes. |
+| `T24` | final-quality-reference-gate | all prior tasks | Full style, build/test, HDF5, determinism, and reference evidence. |
+
+Each task maps one-to-one to `/phases/cpp-object-oriented-modular-refactoring/stepN.md`.
+
+## 6. TDD Test Plan
+
+| Test ID | First failing evidence | GREEN evidence |
+| --- | --- | --- |
+| `P-AGENT-001` | Python contract reports missing mandatory `CODINGSTYLE.md` input. | Agent workflow contract passes. |
+| `P-STYLE-001` | Policy test reports missing or incorrect clang/Doxygen configuration. | Policy and full Harness Python tests pass. |
+| `C-STYLE-001..004` | Test includes/calls use `.h` and PascalCase before production conversion, causing a compile failure. | Focused module suites and full CTest pass. |
+| `C-VEC3-001` | `vector3_test.cpp` cannot compile because `Vector3` is absent. | Arithmetic, finite, and normalization-boundary tests pass. |
+| `C-DUP-001..004` | Tests reference the new shared seam before it exists. | Shared seam passes and old duplicate helper definitions are absent by `rg` checks. |
+| `C-MODEL-001..002` | Polymorphic ownership and const stable-index tests fail before semantic bases exist. | Material/property/definition tests and Domain mapping tests pass. |
+| `C-ELEMENT-001` | Base-interface creation and incompatibility tests fail before `ElementFactory`. | B33/MITC4 creation, rejection, stiffness, and recovery tests pass. |
+| `C-DOF-001` | Fake runtime element layout is not accepted by DofManager. | Stable scatter/pattern and invariant tests pass. |
+| `C-ASSEMBLY-001` | Fake runtime contribution is not assembled. | Serial/TBB/repeated CSR outputs remain byte-identical. |
+| `C-RECOVERY-001` | Fake result bundle cannot flow through recovery. | Beam/shell identities, signs, energy, and atomic rollback pass. |
+| `C-LOAD-001` | A fake Load cannot emit ordered full-DOF contributions. | Source-order accumulation and current load validation pass. |
+| `C-BC-001` | A fake BoundaryCondition cannot resolve constraint definitions. | Nonzero prescribed displacement and reconstruction pass. |
+| `C-ANALYSIS-001` | LinearStaticAnalysis cannot be invoked through `Analysis`. | Approved factorization/load/solve/recovery lifecycle passes. |
+| `C-MODULE-001..003` | Tests reference extracted mapper/recovery/HDF5 responsibilities before their seams exist. | Existing public behavior and atomicity suites pass after extraction. |
+| `C-REF-B33-001` | No new intentional failure; final gate reuses the approved external comparison. | B33 comparison passes under its existing component-scale tolerance. |
+| `C-REF-MITC4-001` | No new intentional failure; final gate reuses the approved external comparison. | MITC4 translations pass at fixed `1.0e-5`; rotations remain warning-only. |
+
+RED and GREEN evidence, command, exit code, duration, output tail, and failed test names
+must be recorded during execution in the Implementation-owned reports. A final reference
+gate does not manufacture an artificial RED because it verifies an unchanged approved
+external contract after all refactoring tasks.
+
+## 7. CMake/CTest Plan
+
+- Keep the existing `fesa_solver`, `fesa_cli`, `fesa_unit_tests`,
+  `fesa_integration_tests`, `fesa_reference_tests`, and `fesa_tests` targets.
+- Register new production/test files in `/src/fesa/CMakeLists.txt` and
+  `/tests/CMakeLists.txt` in their owning task.
+- Do not create a new test executable or change existing test labels.
+- `.harness/config.json` is absent, so use `.harness/build`, MSVC x64, Debug, and the
+  explicit local dependency paths recorded in each Step.
+- Every C++ task runs a focused CTest regular expression and the full CTest discovery
+  and execution sequence.
+- Step `T24` performs a fresh configure and the final B33/MITC4 reference tests.
+
+## 8. Candidate Files and Ownership
+
+| Responsibility | Candidate files |
+| --- | --- |
+| Policy/tooling | `.codex/agents/implementation-agent.toml`, `.clang-format`, `.clang-tidy`, `Doxyfile`, `tests/test_agent_skill_workflow_contract.py`, `tests/test_cpp_policy_contract.py` |
+| Fixed/dynamic math | `include/fesa/math/vector3.h`, `include/fesa/math/vector.h`, `include/fesa/math/matrix.h`, `src/fesa/math/dense_blas_internal.h`, matching `.cpp` and unit tests |
+| Semantic material/property | `include/fesa/materials/*.h`, `include/fesa/properties/*.h`, `src/fesa/materials/*.cpp`, `src/fesa/properties/*.cpp`, matching unit tests |
+| Semantic element definitions | `include/fesa/elements/element_definition.h`, concrete definition headers, `include/fesa/model/domain.h`, `src/fesa/model/domain.cpp` |
+| Runtime elements | `include/fesa/elements/element.h`, `element_factory.h`, existing B33/MITC4 kernels and new factory implementation/tests |
+| Source resolution | `include/fesa/model/source_target_resolver.h`, `src/fesa/model/source_target_resolver.cpp`, focused tests |
+| Solver consumers | DofManager, SparseAssembler, ResultRecovery headers/sources/tests |
+| Loads | `include/fesa/loads/load.h`, `concentrated_nodal_load.h`, sources, LoadAssembler and tests |
+| Constraints | `boundary_condition.h`, `prescribed_displacement.h`, `essential_constraint_policy.h`, sources and tests |
+| Analysis | `analysis.h`, `linear_static_analysis.h`, sources and integration tests |
+| Mapper split | focused private mapper modules under `src/fesa/io/abaqus/` with one public `domain_mapper.h` facade |
+| Recovery split | focused modules under `src/fesa/results/` with one public `result_recovery.h` facade |
+| HDF5 split | private modules under `src/fesa/io/hdf5/` with one public `hdf5_results_writer.h` facade |
+
+These are implementation candidates, not permission to introduce extra public API. Each Step
+must choose the minimum files consistent with the approved boundaries.
+
+## 9. Candidate Interface Contracts
+
+The implementation may refine parameter carrier names while preserving these semantic contracts:
+
+```cpp
+struct AnalysisRequest {
+  std::filesystem::path input_path;
+  std::filesystem::path output_path;
+};
+
+class Analysis {
+ public:
+  virtual ~Analysis() = default;
+  virtual Status Run(const AnalysisRequest& request) = 0;
+};
+
+class ElementDefinition {
+ public:
+  virtual ~ElementDefinition() = default;
+  virtual ElementDefinitionKind Kind() const noexcept = 0;
+  virtual const SourceEntityId& SourceId() const noexcept = 0;
+  virtual const std::vector<EntityIndex>& NodeIndices() const noexcept = 0;
+  virtual EntityIndex PropertyIndex() const noexcept = 0;
+};
+
+class Element {
+ public:
+  virtual ~Element() = default;
+  virtual const ElementDofLayout& DofLayout() const noexcept = 0;
+  virtual Result<ElementStiffnessContribution> ComputeStiffness() const = 0;
+  virtual Result<ElementResultBundle> Recover(
+      const Vector& full_displacement) const = 0;
+};
+
+class Load {
+ public:
+  virtual ~Load() = default;
+  virtual Result<std::vector<LoadContribution>> ComputeContributions(
+      const LoadContext& context) const = 0;
+};
+
+class BoundaryCondition {
+ public:
+  virtual ~BoundaryCondition() = default;
+  virtual Result<std::vector<ConstraintDefinition>> ResolveConstraints(
+      const BoundaryConditionContext& context) const = 0;
+};
+```
+
+Do not add future-only methods to these bases. Factory compatibility may use a centralized,
+explicit kind discriminator followed by a checked concrete access; consumers must not scatter
+`dynamic_cast` or B33/MITC4 switches.
+
+## 10. Data Flow Contract
+
+```text
+existing Abaqus .inp
+  -> syntax reader
+  -> responsibility-split semantic mappers
+  -> immutable Domain-owned definitions
+  -> AnalysisModel non-owning active view
+  -> ElementFactory runtime elements
+  -> DofManager / deterministic assembly / constraints
+  -> LinearStaticAnalysis
+  -> result recovery candidate and validation
+  -> authoritative results.h5 atomic commit
+  -> test-only deterministic projection
+  -> existing Abaqus CSV comparison by source identity and component
+```
+
+- B33 input and CSVs remain under `/reference/cantilever beam/` with their current
+  names and component-scale tolerance.
+- Blocking MITC4 S4 input/displacement CSV remains under `/reference/shell/` with
+  fixed absolute tolerance `1.0e-5` for U1/U2/U3 and warning-only UR1/UR2/UR3.
+- `/reference/shellR/` is not promoted into a blocking comparison.
+- No reference artifact is renamed, rewritten, regenerated, or normalized.
+
+## 11. Acceptance Traceability Matrix
+
+| Requirement | Tasks | Tests/evidence | Acceptance |
+| --- | --- | --- | --- |
+| `R-PRESERVE-001` | `T03..T24` | all current suites, `C-REF-B33-001`, `C-REF-MITC4-001` | Full CTest and blocking references pass. |
+| `R-STYLE-001` | `T01`, `T03..T06`, `T24` | `P-STYLE-001`, clang-format, clang-tidy config, legacy-header scan | Style commands and full build pass. |
+| `R-DOC-001` | `T03..T24` | policy scan and configured warning-strict Doxyfile | Production comments exist; tests are excluded. |
+| `R-DUP-001` | `T07..T09` | `C-VEC3-001`, element/result/I/O suites, duplicate scan | One Vector3 implementation remains. |
+| `R-DUP-002` | `T10`, `T11`, `T15` | `C-DUP-001..004` | Shared owners pass focused tests. |
+| `R-MODEL-001` | `T12`, `T13` | `C-MODEL-001..002` | Polymorphic stable ownership passes. |
+| `R-ELEMENT-001` | `T13`, `T14` | `C-ELEMENT-001` | Factory creates current kinds and rejects incompatible combinations. |
+| `R-PIPELINE-001` | `T15..T17` | `C-DOF-001`, `C-ASSEMBLY-001`, `C-RECOVERY-001` | Generic consumer and deterministic tests pass. |
+| `R-LOAD-001` | `T18` | `C-LOAD-001` | Ordered accumulation and current validations pass. |
+| `R-BC-001` | `T19` | `C-BC-001` | Prescribed displacement partition/reconstruction passes. |
+| `R-ANALYSIS-001` | `T20` | `C-ANALYSIS-001` | Lifecycle and factorization count pass. |
+| `R-MODULE-001` | `T21..T23` | `C-MODULE-001..003` | Facade behavior and atomicity suites pass. |
+| `R-AGENT-001` | `T00` | `P-AGENT-001` | Python workflow contract passes. |
+| `R-SCOPE-001` | every task | diff review and final reference/artifact checks | No excluded behavior or artifact change appears. |
+
+## 12. Validation Commands
+
+Harness Python and policy validation:
+
+```powershell
+uv run --with pytest python -m pytest -v -rs
+& "C:/Program Files/LLVM/bin/clang-format.exe" --version
+& "C:/Program Files/LLVM/bin/clang-tidy.exe" --version
+& "C:/Program Files/LLVM/bin/clang-tidy.exe" --verify-config
+```
+
+MSVC clean configure and full verification:
+
+```powershell
+$requiredBuildPaths = @(
+  "C:/git/googletest",
+  "C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl",
+  "C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb",
+  "C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+)
+foreach ($requiredBuildPath in $requiredBuildPaths) {
+  if (-not (Test-Path -LiteralPath $requiredBuildPath)) {
+    throw "Missing $requiredBuildPath"
+  }
+}
+cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 `
+  "-DFESA_GTEST_SOURCE_DIR=C:/git/googletest" `
+  "-DMKL_DIR=C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl" `
+  "-DTBB_DIR=C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb" `
+  "-DHDF5_DIR=C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+cmake --build .harness/build --config Debug --target fesa_tests
+ctest --test-dir .harness/build -C Debug --show-only=json-v1
+ctest --test-dir .harness/build -C Debug --output-on-failure
+ctest --test-dir .harness/build -C Debug `
+  -R "B33ReferenceComparison|Mitc4S4Reference" --output-on-failure
+```
+
+Repository style and artifact checks:
+
+```powershell
+$cppFiles = @(rg --files include src tests -g "*.h" -g "*.cpp")
+& "C:/Program Files/LLVM/bin/clang-format.exe" --dry-run --Werror $cppFiles
+$publicHeaders = @(rg --files include/fesa -g "*.h")
+foreach ($publicHeader in $publicHeaders) {
+  & "C:/Program Files/LLVM/bin/clang-tidy.exe" --config-file=.clang-tidy `
+    $publicHeader -- -x c++ -std=c++17 -Iinclude
+  if ($LASTEXITCODE -ne 0) {
+    throw "clang-tidy failed for $publicHeader"
+  }
+}
+$legacyHeaders = @(rg --files include tests -g "*.hpp")
+if ($legacyHeaders.Count -ne 0) {
+  $legacyHeaders
+  throw "Legacy .hpp headers remain"
+}
+git diff --exit-code 1e5758f -- reference
+```
+
+Doxygen generation is deliberately absent from the blocking commands. When the user
+requests documentation generation later, execute `doxygen Doxyfile` and treat warnings
+as failures without committing generated HTML.
+
+## 13. Risks and Downstream Handoff
+
+- Global API/header renaming has a wide compile blast radius. Mechanical style Steps
+  are isolated from semantic restructuring to keep failures attributable.
+- Domain polymorphism can accidentally destabilize vector indices or lifetimes. Tests
+  must prove insertion order, const access, and AnalysisModel non-owning lifetime.
+- Virtual element recovery can tempt a giant result record. Preserve distinct beam and
+  shell rows in a backend-neutral bundle rather than adding meaningless common fields.
+- Moving vector helpers can change floating-point operation order. Preserve each
+  formulation expression order and use exact regression where no approved tolerance
+  applies.
+- File splits can leak vendor dependencies through public headers. Keep all HDF5/MKL/TBB
+  types in private implementation modules.
+
+Downstream handoff is one bounded handoff to `implementation-agent` through the
+Coordinator: execute only the Executor-selected `stepN.md`, read `/docs/CODINGSTYLE.md`
+before C++ work, record RED/GREEN/VERIFY evidence, and do not advance another Step.
+
+## 14. Harness Step Draft
+
+- Task name: `cpp-object-oriented-modular-refactoring`
+- Steps: `step0.md` through `step24.md` in dependency order shown in Work Breakdown.
+- Every Step contains its own prerequisite files, test-first failure, candidate
+  interfaces, exact focused/full commands, and prohibitions.
+- Stop conditions are an upstream contract conflict, a missing declared artifact at
+  final comparison, an unresolved environment dependency, or repeated build/test
+  failure. In each case only the current Step status payload is changed.
+- Planning approval materializes these files but does not authorize
+  `python scripts/execute.py cpp-object-oriented-modular-refactoring`.
+
+## 15. Open Issues
+
+- No blocking architecture, formulation, I/O, reference, or tolerance issue remains.
+- Doxygen executable use and generated documentation are deferred by explicit user
+  decision; this does not waive production Doxygen comments or `Doxyfile` configuration.

docs/cpp-object-oriented-modular-refactoring/implementation-report.md

@@ -0,0 +1,120 @@
+# C++ Object-Oriented Modular Refactoring Implementation Report
+
+## Metadata
+
+- feature_id: `cpp-object-oriented-modular-refactoring`
+- owner_agent: `implementation-agent`
+- final_step: `24 final-quality-reference-gate`
+- source_plan: `docs/cpp-object-oriented-modular-refactoring/implementation-plan.md`
+- phase_index: `phases/cpp-object-oriented-modular-refactoring/index.json`
+- evidence_date: `2026-08-16`
+- head: `f84ebb541f4717ab8300cf0d80497e1277d3bb48`
+- reference_baseline: `1e5758f3e482fae4c3d58cac680abe0aac02e549`
+- classification: `pass-for-physics-evaluation`
+
+## Scope and no-change assertion
+
+Step 24 added no production behavior, no comparator/tolerance changes, and no
+reference artifact changes. The only implementation-owned source action during this
+step was running `clang-format -i` on four files that the dry-run style gate reported;
+`git diff --raw`, `git diff --numstat`, and `git diff --check` for those files were
+empty after the run, so no tracked source-content diff was introduced.
+
+No Doxygen executable, hook entry point, `scripts/execute.py`, Abaqus, Nastran,
+reference solver, commit, or push was run.
+
+## Prior Step RED/GREEN/VERIFY evidence
+
+The final gate reviewed the Executor-recorded Step 0-23 summaries. Each prior step is
+already `completed` in the phase index and retains its Executor-owned timestamps.
+
+| Step | Task | Evidence summary |
+| ---: | --- | --- |
+| 0 | `coding-style-agent-contract` | RED missing mandatory `CODINGSTYLE.md`; GREEN/VERIFY pytest 13/13, clean-env pytest 20/20, VS18 Debug build and CTest 144/144. |
+| 1 | `cpp-style-tooling` | RED missing style/Doxygen config; GREEN/VERIFY policy pytest, clean-env pytest 21/21, LLVM 22.1.8 config checks, VS18 Debug build, CTest 144/144. |
+| 2 | `architecture-boundaries` | Documentation-only architecture/ADR update; contract grep, diff check, MSVC Debug build, CTest discovery and 144/144 passed. |
+| 3 | `foundation-google-style` | RED missing `.h` header; GREEN/VERIFY focused build/CTest 22/22, format, full build, CTest 144/144. |
+| 4 | `model-element-google-style` | RED missing model `.h`; GREEN/VERIFY focused build/CTest 37/37, format 14 files, full build, CTest 144/144. |
+| 5 | `solver-workflow-google-style` | RED missing analysis model `.h`; GREEN/VERIFY focused build/CTest 57/57, format 31 files, full build, CTest 144/144. |
+| 6 | `io-application-google-style` | RED missing input reader `.h`; GREEN/VERIFY focused I/O/HDF5/app/reference CTest 36/36, format 23 files, full build, CTest 144/144. |
+| 7 | `vector3-value-type` | RED missing `vector3.h`; GREEN/VERIFY Vector3 tests 7/7, format/tidy, full build, CTest 151/151. |
+| 8 | `element-geometry-vector3` | RED typed Vector3 seam compile failure; GREEN/VERIFY focused CTest 44/44, duplicate scan 0, format, full build, CTest 155/155. |
+| 9 | `result-io-vector3` | RED duplicate helpers; GREEN/VERIFY ResultRecovery/InpDomainMapping/HDF5 33/33, duplicate scan 0, format, full build, CTest 156/156. |
+| 10 | `dense-blas-adapter` | RED missing dense BLAS internal header; GREEN/VERIFY focused CTest 4/4, public vendor scan 0, format, full build, CTest 158/158. |
+| 11 | `source-target-resolver` | RED missing ASCII/source-target modules and plus-label failure; GREEN/VERIFY targeted/focused tests 44/44, helper scan, format, full build, CTest 164/164. |
+| 12 | `material-property-hierarchy` | RED missing material/property bases; GREEN/VERIFY focused CTest 11/11, format, full build, CTest 170/170. |
+| 13 | `element-definition-domain` | RED missing element definition/ownership APIs; GREEN/VERIFY focused CTest 20/20, format, full build, CTest 172/172. |
+| 14 | `runtime-element-factory` | RED missing runtime element API; GREEN/VERIFY focused CTest 35/35, dynamic_cast scan 0, format, full build, CTest 176/176. |
+| 15 | `generic-dof-manager` | RED fake element/layout seam failures; GREEN/VERIFY focused CTest 21/21, concrete/helper branches 0, format, full build, CTest 178/178. |
+| 16 | `generic-sparse-assembler` | RED fake runtime contribution seam missing; GREEN/VERIFY SparseAssembly 9/9, concrete branch count 0, format, full build, CTest 179/179. |
+| 17 | `generic-result-recovery` | RED missing generic recovery seam; GREEN/VERIFY focused CTest 53/53, concrete branch count 0, format, full build, CTest 182/182. |
+| 18 | `load-hierarchy` | RED missing Load APIs; GREEN/VERIFY focused CTest 29/29, format, full build, CTest 186/186. |
+| 19 | `boundary-condition-policy` | RED missing BoundaryCondition APIs; GREEN/VERIFY focused CTest 29/29, format, full build, discovery/full CTest 191/191. |
+| 20 | `analysis-hierarchy` | RED missing `analysis.h`; GREEN/VERIFY focused CTest 12/12, scans, full build, discovery/full CTest 193/193. |
+| 21 | `domain-mapper-modules` | RED missing private mapper seam; GREEN/VERIFY focused CTest 17/17, format, full build, CTest 197/197. |
+| 22 | `result-recovery-modules` | RED missing recovery component seam; GREEN/VERIFY focused CTest 28/28, format, full build, CTest 203/203. |
+| 23 | `hdf5-writer-modules` | RED missing HDF5 component seam; GREEN/VERIFY focused schema/atomicity CTest 13/13, scans, format, full build, CTest 206/206. |
+
+Step 24 is a final verification gate and did not manufacture a new RED condition;
+it reused the approved B33 and MITC4 reference comparisons after style and full
+build/test verification.
+
+## Step 24 command evidence
+
+| Stage | Command | Exit | Duration | Result |
+| --- | --- | ---: | ---: | --- |
+| Environment diagnostic | `uv run --with pytest python -m pytest -v -rs` with inherited `FESA_HARNESS_CODEX_SANDBOX=danger-full-access` | 1 | 2.700s | Environment-specific failure: default-sandbox test observed the explicit override. |
+| Policy verify | `Remove FESA_HARNESS_CODEX_SANDBOX` in child process; `uv run --with pytest python -m pytest -v -rs` | 0 | 0.751s | 21/21 passed. |
+| Tool verify | `& "C:/Program Files/LLVM/bin/clang-format.exe" --version` | 0 | 0.026s | clang-format 22.1.8. |
+| Tool verify | `& "C:/Program Files/LLVM/bin/clang-tidy.exe" --version` | 0 | 0.030s | clang-tidy 22.1.8. |
+| Tool verify | `& "C:/Program Files/LLVM/bin/clang-tidy.exe" --verify-config` | 0 | 0.028s | No config errors. |
+| Style RED | `clang-format --dry-run --Werror` over 163 files | 1 | 0.892s | Four whitespace/line-ending format findings. |
+| Style normalization | `clang-format -i` on the four reported files | 0 | tool wall 0.5s | No tracked content diff after formatting normalization. |
+| Style verify | `clang-format --dry-run --Werror` over 163 files | 0 | 0.888s | Passed. |
+| Public-header policy | `clang-tidy --config-file=.clang-tidy <header> -- -x c++ -std=c++17 -Iinclude` over 46 public headers | 0 | 41.822s | Passed. The 47 naming diagnostics were 45 required trailing-underscore header guards and two `const` parameter classifications; there were zero other naming diagnostics. |
+| Doxygen/header-guard policy | Read-only scan of production headers and test Doxygen tags | 0 | 0.302s | 63/63 production headers had guards and Doxygen contracts; 0 test files contained imposed Doxygen tags. |
+| Header extension | `.hpp` scan under `include src tests` | 0 | 0.162s | 0 legacy `.hpp` files. |
+| Dependency precheck | Test declared GoogleTest/MKL/TBB/HDF5 paths | 0 | 0.165s | All paths found. |
+| Fresh configure | `cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 ...` | 0 | 6.204s | MSVC 19.51, VS18, build files generated. |
+| Build | `cmake --build .harness/build --config Debug --target fesa_tests` | 0 | 9.639s | Debug aggregate test target built. |
+| Discovery | `ctest --test-dir .harness/build -C Debug --show-only=json-v1` | 0 | 0.235s | 206 tests discovered. |
+| Full test | `ctest --test-dir .harness/build -C Debug --output-on-failure` | 0 | 9.908s | 206/206 passed. |
+| Artifact check | Exact declared artifact existence, SHA-256, line/row inventory | 0 | 0.218s | Six declared files present. |
+| Artifact no-change | `git diff --exit-code 1e5758f -- reference` | 0 | 0.165s | No reference diff. |
+| Artifact schema | Read-only type/header/key/finite precheck | 0 | 0.266s | B33/S4 type present; CSV headers/keys/finite checks passed. |
+| Compare | `ctest --test-dir .harness/build -C Debug -R "B33ReferenceComparison|Mitc4S4Reference" --output-on-failure` | 0 | 1.260s | 3/3 approved reference tests passed. |
+| Generated result check | Required `results.h5` existence and hashes | 0 | 0.218s | B33 and MITC4 comparison `results.h5` present. |
+| Post no-change | `git diff --exit-code 1e5758f -- reference` | 0 | 0.179s | No reference diff after compare. |
+| Whitespace check | `git diff --check` | 0 | 0.199s | No whitespace errors. |
+
+## Requirement traceability
+
+| Requirement | Final-gate evidence |
+| --- | --- |
+| `R-PRESERVE-001` | Full CTest 206/206, B33 comparison 176/176 rows passed, MITC4 S4 comparison 147/147 blocking U rows passed; no tolerance/comparator/reference diff. |
+| `R-STYLE-001` | `.h` header scan passed with zero `.hpp`; clang-format passed over 163 files; clang-tidy selected public-header check passed over 46 headers. |
+| `R-DOC-001` | Policy pytest passed; 63/63 production headers had header guards and Doxygen contracts, while 0 test files contained imposed Doxygen tags. Doxygen generation was intentionally not run. |
+| `R-DUP-001` | Step 7-9 summaries record the shared `Vector3`, element/result/I/O preservation tests, and duplicate-definition scans with zero remaining local definition families; full CTest 206/206 passed. |
+| `R-DUP-002` | Steps 10, 11, and 15 record the single dense-BLAS adapter, shared ASCII/source resolver, and DofManager-owned invariant logic with focused tests and helper/branch scans; full CTest 206/206 passed. |
+| `R-MODEL-001` | Steps 12-13 record polymorphic Material/ElementProperty/ElementDefinition ownership and stable Domain views; focused ownership tests and full CTest passed. |
+| `R-ELEMENT-001` | Step 14 records semantic/runtime separation, fail-closed factory compatibility checks, virtual destruction, owner-bounded views, and zero `dynamic_cast` uses; full CTest passed. |
+| `R-PIPELINE-001` | Steps 15-17 record generic DofManager, SparseAssembler, and ResultRecovery seams with fake runtime elements, deterministic/atomic tests, and zero concrete B33/MITC4 consumer branches; full CTest passed. |
+| `R-LOAD-001` | Step 18 records Domain-owned Load objects, ordered contributions, validation-before-candidate accumulation, and focused LoadAssembler tests; full CTest passed. |
+| `R-BC-001` | Step 19 records BoundaryCondition definitions and stable essential-constraint partition/reconstruction, including nonzero and `0 x 0 Kff` cases; full CTest passed. |
+| `R-ANALYSIS-001` | Step 20 records minimal base `Run()` dispatch and procedure-owned lifecycle tests, including factorize-before-load, exactly-one factorization, all-constrained solve, and writer suppression on recovery failure; full CTest passed. |
+| `R-MODULE-001` | Steps 21-23 record the approved Domain-mapper, result-recovery, and HDF5 private component splits with facade, diagnostic, identity, rollback, self-check, and atomic-finalization tests; full CTest passed. |
+| `R-AGENT-001` | Step 0 records the mandatory `docs/CODINGSTYLE.md` implementation-agent contract; the final clean-environment policy suite passed 21/21. |
+| `R-SCOPE-001` | No production behavior or future-feature changes in Step 24; reference tree diff against `1e5758f` is empty. |
+
+All acceptance traceability rows from the approved implementation plan are listed
+above. Their owning task rows `T00..T23` retain the recorded RED/GREEN/VERIFY
+summaries, and `T24` supplies the final style, build, CTest, artifact, HDF5,
+comparison, and no-change evidence.
+
+## Handoff
+
+Implementation gate verdict: `pass-for-physics-evaluation`.
+
+Open issues: none blocking. The inherited `FESA_HARNESS_CODEX_SANDBOX` override is an
+environment note only; the clean child-process policy command passed without code
+changes.

docs/cpp-object-oriented-modular-refactoring/reference-comparison.md

@@ -0,0 +1,165 @@
+# C++ Object-Oriented Modular Refactoring Reference Comparison Report
+
+## Metadata
+
+- owner_agent: `implementation-agent`
+- feature_id: `cpp-object-oriented-modular-refactoring`
+- report_status: `passed`
+- date: `2026-08-16`
+- reference_baseline: `1e5758f3e482fae4c3d58cac680abe0aac02e549`
+- command_order: `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT`
+- authoritative FESA output: generated `results.h5`
+- reference artifacts: read-only existing files under `reference/`
+
+## ARTIFACT CHECK
+
+Exact declared artifact inventory:
+
+| Artifact | Bytes | Lines | Data rows | SHA-256 |
+| --- | ---: | ---: | ---: | --- |
+| `reference/cantilever beam/cantilever beam.inp` | 2330 | 106 | N/A | `E406EA9560321B791DBDB829E03BD24593B9875E0195D35B86BD931EDA122EF3` |
+| `reference/cantilever beam/cantilever beam displacements.csv` | 1790 | 12 | 11 | `7B3312FBC8848E81D9A0FD4FF2B56BC1954636A2C14B5C1CBB269CB9477D3C31` |
+| `reference/cantilever beam/cantilever beam elemental forces.csv` | 1396 | 12 | 11 | `E5E77FEC0FA9482AE018DBF296E74D396335C7C711BD2E9AA2315247A34290BA` |
+| `reference/cantilever beam/cantilever beam reactions.csv` | 1780 | 12 | 11 | `BF30CDB0CD50106885DE14D63492737736C587426EBD787DE4F7EE6AA86DAA23` |
+| `reference/shell/shell.inp` | 4770 | 164 | N/A | `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF` |
+| `reference/shell/shell displacements.csv` | 5592 | 50 | 49 | `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B` |
+
+Read-only schema precheck:
+
+| Check | Result |
+| --- | --- |
+| B33 input contains `TYPE=B33` | Pass |
+| MITC4 S4 input contains `TYPE=S4` | Pass |
+| B33 displacement CSV header/key/finite precheck | Pass: 11 rows, 11 unique keys, 0 duplicate keys, 0 nonfinite values |
+| B33 reaction CSV header/key/finite precheck | Pass: 11 rows, 11 unique keys, 0 duplicate keys, 0 nonfinite values |
+| B33 elemental-force CSV header/key/finite precheck | Pass: 11 rows, 11 unique keys, 0 duplicate keys, 0 nonfinite values |
+| MITC4 S4 displacement CSV header/key/finite precheck | Pass: 49 rows, 49 unique keys, 0 duplicate keys, 0 nonfinite values |
+| Reference tree diff before compare | Pass: `git diff --exit-code 1e5758f -- reference`, exit 0 |
+
+## COMPARE
+
+Command:
+
+```powershell
+ctest --test-dir .harness/build -C Debug `
+  -R "B33ReferenceComparison|Mitc4S4Reference" --output-on-failure
+```
+
+Result: exit 0 in 1.260s, 3/3 tests passed.
+
+Generated artifacts:
+
+| Artifact | Bytes | SHA-256 |
+| --- | ---: | --- |
+| `.harness/build/reference/cantilever-beam-b33/results.h5` | 25336 | `58CD358F68D8094079E9E525C35EE88AE0575A2962D93472BF0678D78A785247` |
+| `.harness/build/reference/cantilever-beam-b33/comparison.json` | 128118 | `258347AEA791D981AEA9B2BCAD85DE5344D4859ECA3692DC5E7AA01A848F8E0D` |
+| `.harness/build/reference/mitc4-shell-s4-comparison/results.h5` | 95024 | `A8D2E12886E87BAA5D895B7E96278EA4B462718D13B985CA2688B505480F0195` |
+| `.harness/build/reference/mitc4-shell-s4-comparison/comparison.json` | 94349 | `8E8DEA51B6F7C663BACC41FDA6103A4596DB26E02F1EAD6069D458F51E0102E6` |
+| `.harness/build/reference/mitc4-shell-s4-metadata/results.h5` | 95024 | `8FD6609A2D3758365A2AC2E34692CC0EC982D8E67396BDFF03FAAF2539EF483C` |
+| `.harness/build/reference/mitc4-shell-s4-metadata/comparison.json` | 94349 | `8E8DEA51B6F7C663BACC41FDA6103A4596DB26E02F1EAD6069D458F51E0102E6` |
+
+The generated `comparison.json` files are the deterministic machine-readable
+per-row decision records. The summaries below preserve row counts, worst rows,
+precheck/tolerance decisions, and artifact hashes for audit.
+
+## HDF5-to-CSV projection and tolerance contracts
+
+### B33
+
+- Model: `cantilever-beam-b33`
+- HDF5 datasets:
+  - `/steps/Step-1/frames/0/nodal/displacement`
+  - `/steps/Step-1/frames/0/nodal/reaction`
+  - `/steps/Step-1/frames/0/element/section_resultant`
+- Row identity: model, `Step-1`, frame `0`, instance `PART-1_1-1`, source node label, quantity, component.
+- Components:
+  - displacement `UX/UY/UZ/URX/URY/URZ`
+  - reaction `RF1/RF2/RF3/RM1/RM2/RM3`
+  - section resultant `N/T/My/Mz`
+- Tolerance: `absolute_floor + 1.0e-6 * reference_scale`, with reference scale from read-only Abaqus rows only; displacement/rotation floor `1.0e-9`, force/moment floor `1.0e-3`.
+- Pre-tolerance policy: missing, extra, duplicate, nonfinite, schema-mismatched, or identity-mismatched rows fail before tolerance.
+
+### MITC4 S4
+
+- Case: `shell-s4`
+- Source element type: `S4`
+- Internal formulation: `FESA-MITC4`
+- Integration rule: `2x2x2-gauss; mitc4-edge-midpoint-shear`
+- HDF5 dataset: `/steps/Step-1/frames/0/nodal/displacement`
+- Row identity: case, instance, source node label, component.
+- Components: `U1/U2/U3` blocking; `UR1/UR2/UR3` warning-only.
+- Tolerance: fixed absolute `1.0e-5` for every U/UR row; no component scale, row denominator, zero clamp, omission, or averaging affects the decision.
+- Pre-tolerance policy: missing, extra, duplicate, nonfinite, header-mismatched, or identity-mismatched projected rows fail before tolerance.
+
+## CLASSIFY
+
+Blocking classification: `pass`.
+
+No missing, extra, duplicate, nonfinite, schema-mismatched, identity-mismatched, or
+tolerance-failed blocking row was reported by either generated comparison.
+
+### B33 row and metric decisions
+
+Overall: `passed=true`; row decisions: 176/176 passed; failed rows: 0; nonfinite row
+metrics: 0; stress comparison applicable: `false` with N/A reason
+`Abaqus beam stress comparison is N/A; analytical/unit and HDF5 schema tests provide stress evidence.`
+
+Physics evidence: endpoint consistency passed; free residual norm
+`9.356339321107032e-07`.
+
+| Quantity | Component | Rows | Reference scale | Max abs error | Max normalized error | RMS error | Norm error | Worst row decision |
+| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | --- |
+| displacement | UX | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `1e-09`, pass |
+| displacement | UY | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `1e-09`, pass |
+| displacement | UZ | 11 | 0.0190476272 | 5.33322917078971e-10 | 0.026602795021995 | 2.79095304685767e-10 | 9.25654406392607e-10 | node 11, FESA -0.019047626666677083, reference -0.0190476272, tol `2.00476272e-08`, pass |
+| displacement | URX | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `1e-09`, pass |
+| displacement | URY | 11 | 0.00285714399 | 1.00001394318094e-10 | 0.0259262798011579 | 5.80165854051178e-11 | 1.92419245406385e-10 | node 9, FESA 0.002742858240001394, reference 0.00274285814, tol `3.85714399e-09`, pass |
+| displacement | URZ | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `1e-09`, pass |
+| reaction | RF1 | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| reaction | RF2 | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| reaction | RF3 | 11 | 1000000 | 8.19563865661621e-07 | 8.1874512054108e-07 | 3.62393833938394e-07 | 1.20192437351202e-06 | node 1, FESA 1000000.0000008196, reference 1000000, tol `1.001`, pass |
+| reaction | RM1 | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| reaction | RM2 | 11 | 10000000 | 5.05149364471436e-06 | 5.05098854585977e-07 | 1.52613777609191e-06 | 5.06162638168429e-06 | node 1, FESA -10000000.000005051, reference -10000000, tol `10.001`, pass |
+| reaction | RM3 | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| section_resultant | N | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| section_resultant | T | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+| section_resultant | My | 11 | 10000000 | 0.0156002428611895 | 0.00155986829928902 | 0.00470365086790684 | 0.0156002450736765 | node 11, FESA 2.4286118949223834e-07, reference -0.0156, tol `10.001`, pass |
+| section_resultant | Mz | 11 | 0 | 0 | 0 | 0 | 0 | node 1, FESA 0, reference 0, tol `0.001`, pass |
+
+### MITC4 S4 row and metric decisions
+
+Overall: `passed=true`; rows: 294/294 within tolerance; blocking U rows: 147/147
+passed; warning-only UR rows: 147/147 within tolerance; warning count: 0; vector
+metrics: 49.
+
+| Component | Rows | Blocking rows | Reference scale | Tolerance | Max abs error | Max normalized error | RMS error | Vector norm error | Worst row decision |
+| --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- |
+| U1 | 49 | 49 | 3.11730945e-23 | 1e-05 | 3.11730945e-23 | 3.11730945e-18 | 1.13587076092006e-23 | 7.95109532644045e-23 | node 12, FESA 0, reference 3.11730945e-23, pass |
+| U2 | 49 | 49 | 3.11730945e-23 | 1e-05 | 3.11730945e-23 | 3.11730945e-18 | 1.13587076092006e-23 | 7.95109532644045e-23 | node 11, FESA 0, reference 3.11730945e-23, pass |
+| U3 | 49 | 49 | 2.37408203e-05 | 1e-05 | 1.90378534915144e-07 | 0.0190378534915144 | 4.24886125341252e-08 | 2.97420287738877e-07 | node 2, FESA -2.3550441765084857e-05, reference -2.37408203e-05, pass |
+| UR1 | 49 | 0 | 7.60725743e-06 | 1e-05 | 6.88285496274043e-08 | 0.00688285496274043 | 2.7737837921291e-08 | 1.94164865449037e-07 | node 44, FESA 1.3985063203725957e-06, reference 1.46733487e-06, pass warning-only |
+| UR2 | 49 | 0 | 7.60725743e-06 | 1e-05 | 6.8828549627399e-08 | 0.0068828549627399 | 2.77378379212909e-08 | 1.94164865449036e-07 | node 34, FESA 1.398506320372601e-06, reference 1.46733487e-06, pass warning-only |
+| UR3 | 49 | 0 | 5.27113701e-25 | 1e-05 | 5.27113701e-25 | 5.27113701e-20 | 2.12986098533393e-25 | 1.49090268973375e-24 | node 34, FESA 0, reference -5.27113701e-25, pass warning-only |
+
+Overall worst MITC4 row: node 2, component U3, FESA
+`-2.3550441765084857e-05`, reference `-2.37408203e-05`, absolute error
+`1.903785349151439e-07`, tolerance `1e-05`, normalized error
+`0.01903785349151439`, blocking pass.
+
+## Reference no-change assertion
+
+Post-compare command:
+
+```powershell
+git diff --exit-code 1e5758f -- reference
+```
+
+Result: exit 0, `<no reference diff>`.
+
+No reference input, CSV, tolerance, comparator contract, or generated reference
+artifact was modified. Generated FESA outputs are confined to `.harness/build/`.
+
+## Open issues
+
+None blocking. Passing comparison is only an implementation handoff to physics
+evaluation; it is not release readiness or physics approval.

docs/formulations/README.md

@@ -1,148 +0,0 @@
-# 정식화 문서 작성 가이드
-
-이 디렉터리는 Formulation Agent가 작성하거나 제안한 기능별 유한요소 정식화 문서를 보관하는 위치다.
-
-기본 파일명은 `docs/formulations/<feature-id>-formulation.md` 형식을 사용한다. 각 문서는 Requirement Agent의 요구조건과 Research Agent의 연구 브리프를 입력으로 받아 Implementation Planning Agent와 Implementation Agent가 사용할 수 있는 수학/알고리즘 계약을 제공해야 한다.
-
-## Formulation Agent 역할
-
-Formulation Agent는 구현 가능한 FEM 정식화 문서를 작성한다.
-
-수행한다:
-- strong form, weak form, discretization을 정리한다.
-- shape functions, `B` matrix 또는 kinematic operator를 정의한다.
-- constitutive contract, element residual/internal force, stiffness/tangent matrix를 정리한다.
-- numerical integration, mapping, Jacobian 규칙을 명시한다.
-- output recovery, algorithm pseudocode, edge case, numerical risk를 정리한다.
-- Numerical Review Agent가 검토할 handoff 항목을 남긴다.
-
-수행하지 않는다:
-- C++ 코드를 구현하지 않는다.
-- C++ API나 파일 구조를 설계하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- release readiness를 승인하지 않는다.
-- Numerical Review Agent 검토 전 정식화를 최종 승인하지 않는다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Formulation
-
-## Metadata
-- feature_id: <feature-id>
-- source_requirement: docs/requirements/<feature-id>.md
-- source_research: docs/research/<feature-id>-research.md
-- status: draft | needs-research | ready-for-numerical-review
-- owner_agent: formulation-agent
-- date: <YYYY-MM-DD>
-
-## Scope and Assumptions
-- analysis_type: linear static | nonlinear static | modal | dynamic | other
-- element_type: <target element>
-- deformation: small | large | TBD
-- linearity: linear | nonlinear | TBD
-- material_model_boundary: <scope>
-- coordinate_system: <global/local coordinates>
-- units: <unit system>
-
-## Primary Variables and DOFs
-- nodal_variables: <variables>
-- dof_ordering: <ordering>
-- sign_convention: <convention>
-- constrained_free_dof_assumptions: <assumptions>
-
-## Strong Form and Boundary Conditions
-- governing_equation: <equation>
-- dirichlet_boundary: <essential BC>
-- neumann_boundary: <natural BC>
-- natural_boundary_terms: <terms>
-
-## Weak or Variational Form
-- test_functions: <definition>
-- integration_by_parts: <steps>
-- internal_virtual_work: <expression>
-- external_virtual_work: <expression>
-
-## Discretization
-- interpolation: <field interpolation>
-- shape_functions: <shape functions>
-- partition_of_unity_check: <expected check>
-- kronecker_delta_check: <expected check>
-- nodal_layout: <layout>
-
-## Kinematics
-- strain_displacement_relation: <relation>
-- B_matrix_or_kinematic_operator: <definition>
-- deformation_gradient: <definition or N/A>
-- strain_measure: <definition or N/A>
-
-## Constitutive Contract
-- elasticity_matrix_or_stress_update: <contract>
-- material_state_variables: <variables or N/A>
-- assumptions_and_constraints: <assumptions>
-
-## Element Equations
-- internal_force_or_residual: <expression>
-- external_force: <expression>
-- stiffness_or_tangent_matrix: <expression>
-- mass_or_damping: <required expression or N/A>
-- vector_matrix_dimensions: <dimensions>
-
-## Mapping and Numerical Integration
-- reference_coordinates: <coordinates>
-- isoparametric_mapping: <mapping>
-- jacobian: <definition>
-- determinant_checks: <validity checks>
-- gauss_points_and_weights: <rule>
-- integration_policy: full | reduced | selective | analytical | TBD
-
-## Output Recovery
-- displacement: <location and convention>
-- reaction: <calculation>
-- element_force: <calculation>
-- strain: <location and recovery>
-- stress: <location and recovery>
-- nodal_extrapolation: <policy or N/A>
-
-## Algorithm Pseudocode
-```text
-math-level element routine and assembly flow only
-no C++ signatures, class names, or file paths
-```
-
-## Numerical Risks
-- rigid_body_modes: <risk/check>
-- patch_test: <expected test>
-- symmetry: <expected property>
-- positive_definiteness: <expected property>
-- hourglass: <risk or N/A>
-- shear_locking: <risk or N/A>
-- volumetric_locking: <risk or N/A>
-- distortion: <risk/check>
-- singular_jacobian: <risk/check>
-
-## Open Issues and Downstream Handoff
-
-### Numerical Review Agent
-- <derivations, assumptions, numerical risks, open issues>
-
-### I/O Definition Agent
-- <required inputs, outputs, units, coordinate conventions, output locations>
-
-### Reference Model Agent
-- <benchmarkable quantities, patch test needs, expected invariants, edge cases>
-
-### Implementation Planning Agent
-- <math-level pseudocode, acceptance-relevant quantities, tests to write first>
-```
-
-## 품질 기준
-
-- 요구조건과 연구 브리프에 없는 값을 임의로 확정하지 않는다.
-- strong form, weak form, discretization, kinematics, constitutive contract, element equations를 구분한다.
-- shape functions는 가능한 경우 partition of unity와 Kronecker delta 검사를 포함한다.
-- numerical integration은 integration point, weight, 적분 대상, full/reduced/selective 정책을 포함한다.
-- mapping은 reference coordinates, Jacobian, determinant validity check를 포함한다.
-- output recovery는 nodal, element-level, integration-point 위치를 구분한다.
-- numerical risk는 rigid body modes, patch test, hourglass, locking, Jacobian 문제를 명시한다.

docs/implementation-plans/README.md

@@ -1,192 +0,0 @@
-# Implementation Plan 문서 작성 가이드
-
-이 디렉터리는 Implementation Planning Agent가 작성하거나 제안한 기능별 구현계획 문서를 보관하는 위치다.
-
-Implementation Planning Agent는 승인된 요구조건, 연구 브리프, 정식화, 수치 리뷰, I/O
-정의와 lightweight reference-case inventory를 C++/MSVC 구현 전 TDD 작업계획으로
-변환한다. Project-local `$harness`를 사용해 multi-Step 초안을 먼저 제시하고 사용자가
-승인한 뒤에만 `phases/` planning files를 생성한다. 계획과 구현 전에
-`docs/HARNESS.md`와 `docs/HARNESS_WORKFLOW.md`를 읽는다.
-
-## Harness 실행 handoff
-
-Planning Agent는 draft -> explicit user approval -> planning files만 수행하며 Step을
-선택하거나 실행하지 않는다. 별도의 명시적 사용자 요청이 있어야 Executor
-(`scripts/execute.py`)가 branch, pending Step, retry, timestamps, commits, advancement와 phase
-status를 소유한다.
-
-Implementation Agent는 approved plan, materialized phase indexes, Executor-selected current
-`stepN.md` 하나만 사용한다. current Step에서 `RED -> observed failure -> minimal GREEN ->
-focused/full VERIFY`를 완료하고 다음 Step을 시작하지 않는다. Agent는 current Step의 `status`
-및 `summary`, `error_message`, `blocked_reason` payload만 쓴다. `.codex/hooks.json`의
-PreToolUse와 Stop은 자동으로 실행되며 hook entry point를 수동 실행해 대체하지 않는다.
-
-기본 파일명은 `docs/implementation-plans/<feature-id>-implementation-plan.md` 형식을 사용한다. 각 문서는 Implementation Agent가 먼저 작성해야 할 실패 테스트, 최소 구현 순서, CMake/CTest 등록 계획, acceptance traceability를 제공해야 한다.
-
-## Implementation Planning Agent 역할
-
-수행한다:
-- upstream 문서가 구현 계획에 충분한지 Readiness Check를 수행한다.
-- 요구조건과 정식화를 작은 Work Breakdown task로 나눈다.
-- unit, integration, parser/I/O, reference-comparison 테스트를 TDD 순서로 정렬한다.
-- CMake/CTest target, `add_test`, label, `ctest -C Debug` 검증 계획을 정의한다.
-- candidate source/header/test/CMake 파일과 ownership boundary를 제안한다.
-- requirement, task, test, reference model, acceptance criterion을 Acceptance Traceability Matrix로 연결한다.
-- `.harness/config.json` 또는 자동 감지 기본값에서 해석되는 MSVC build/test 명령과 feature-specific command를 명시한다.
-- 한 Step을 하나의 layer/module로 제한하고 prerequisite files, RED/GREEN/VERIFY, exact
-  acceptance commands와 구체적 금지사항을 포함한다.
-- 사용자 승인 전에는 `phases/` 파일을 만들지 않고, 승인 뒤에도 planning files만
-  materialize한다. 별도 요청 없이는 `scripts/execute.py`를 실행하지 않으며 Step을
-  선택하거나 실행하지 않는다.
-
-수행하지 않는다:
-- C++ 코드를 구현하지 않는다.
-- 테스트 파일을 작성하지 않는다.
-- CMake 파일을 수정하지 않는다.
-- CMake/CTest를 실행하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- solver 결과를 비교하지 않는다.
-- release readiness를 승인하지 않는다.
-- C++ API, class name, storage layout, file ownership을 확정하지 않는다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Implementation Plan
-
-## Metadata
-- feature_id: <feature-id>
-- source_requirement: docs/requirements/<feature-id>.md
-- source_research: docs/research/<feature-id>-research.md
-- source_formulation: docs/formulations/<feature-id>-formulation.md
-- source_numerical_review: docs/numerical-reviews/<feature-id>-review.md
-- source_io_definition: docs/io-definitions/<feature-id>-io.md
-- source_reference_models: docs/reference-models/<feature-id>-reference-models.md
-- status: draft | needs-upstream-decision | ready-for-implementation | blocked
-- owner_agent: implementation-planning-agent
-- date: <YYYY-MM-DD>
-
-## Readiness Check
-
-| input | required_status | observed_status | decision |
-| --- | --- | --- | --- |
-| requirement | approved or sufficient draft | <status> | proceed | needs-upstream-decision | blocked |
-| formulation | pass-for-implementation-planning or sufficient draft | <status> | proceed | needs-upstream-decision | blocked |
-| numerical_review | pass-for-implementation-planning | <status> | proceed | needs-upstream-decision | blocked |
-| io_definition | ready-for-implementation-planning or sufficient draft | <status> | proceed | needs-upstream-decision | blocked |
-| reference_models | ready-for-implementation-planning or planned artifacts | <status> | proceed | needs-upstream-decision | blocked |
-
-## Implementation Scope
-- included_behavior: <behavior to implement>
-- excluded_behavior: <behavior explicitly out of scope>
-- non_goals: <items not to design or implement in this phase>
-
-## Work Breakdown
-
-| task_id | order | purpose | upstream_trace | depends_on | expected_test_first |
-| --- | --- | --- | --- | --- | --- |
-| TASK-001 | 1 | <small implementation task> | <requirement/formulation/io/reference id> | none | TEST-001 |
-
-## TDD Test Plan
-
-| test_id | order | test_type | red_condition | green_condition | linked_task | command |
-| --- | --- | --- | --- | --- | --- | --- |
-| TEST-001 | 1 | unit | test fails because behavior is missing | test passes after minimal implementation | TASK-001 | ctest -C Debug -R <test-name> |
-| TEST-002 | 2 | integration | integrated path fails before implementation | integrated path passes | TASK-002 | ctest -C Debug -R <test-name> |
-| TEST-003 | 3 | parser/I/O | Abaqus .inp case is not accepted or mapped | input maps to expected semantic model | TASK-003 | ctest -C Debug -R <test-name> |
-| TEST-004 | 4 | reference-comparison | solver HDF5/CSV view comparison fails before implementation | comparison is within planned tolerance | TASK-004 | ctest -C Debug -R <test-name> |
-
-## CMake/CTest Plan
-- target_candidates: <library/test executable targets>
-- add_test_needs: <CTest registration needs>
-- labels: unit | integration | reference | parser | io
-- msvc_config: Debug
-- expected_feature_command: ctest --test-dir .harness/build -C Debug -R <feature-or-label> --output-on-failure
-- full_validation_source: .harness/config.json | Harness auto detection
-
-## Candidate Files and Ownership
-
-| file_candidate | purpose | owner_boundary | notes |
-| --- | --- | --- | --- |
-| include/fesa/<module>/<candidate>.hpp | <candidate public header role> | candidate only, not final API | <notes> |
-| src/<module>/<candidate>.cpp | <candidate implementation role> | candidate only, not final API | <notes> |
-| tests/<module>/<candidate>_test.cpp | <test role> | required before production change | <notes> |
-| CMakeLists.txt | <target/test registration role> | candidate only | <notes> |
-
-## Data Flow Contract
-1. Abaqus `.inp` input follows docs/io-definitions/<feature-id>-io.md.
-2. Parser/I/O path maps model data and history data into the internal semantic model.
-3. Solver path produces authoritative `results.h5` with displacement, reaction, internal force, stress, or feature-specific result datasets.
-4. Reference inputs and required CSV files use exact existing paths declared by the feature.
-5. Reference comparison tests compare only blocking/warning quantities by source ID/component.
-
-## Acceptance Traceability Matrix
-
-| requirement_id | task_id | test_id | reference_model_id | acceptance_criterion | status |
-| --- | --- | --- | --- | --- | --- |
-| <req-id> | TASK-001 | TEST-001 | <model-id or N/A> | <criterion> | draft |
-
-## Validation Commands
-```powershell
-cmake -S . -B .harness/build -A x64
-cmake --build .harness/build --config Debug
-ctest --test-dir .harness/build -C Debug -R <feature-or-label> --output-on-failure
-ctest --test-dir .harness/build -C Debug --output-on-failure
-```
-
-Preset 또는 직접 MSBuild 프로젝트는 `.harness/config.json`에 해석 가능한 명령을 적는다.
-Harness Python, Hook, agent config 변경이 계획 범위에 포함되면
-`uv run --with pytest python -m pytest -v -rs`도 추가한다. Stop 검증은 Step 종료 전 전체
-MSVC build/test를 다시 확인하며, 구현 보고서의 RED 실패 증거를 대체하지 않는다.
-
-## Risks and Downstream Handoff
-
-### Implementation Agent
-- <task order, tests to write first, candidate files, acceptance criteria>
-
-### Build/Test Executor Agent
-- <validation commands, expected CTest labels, feature-specific commands>
-
-### Correction Agent
-- <likely failure classifications and upstream rollback guidance>
-
-### Reference Verification Agent
-- <planned HDF5/CSV comparison tests, exact case paths, tolerance mapping, source-ID/component matching>
-
-## Harness Step Draft
-
-| step | name | owned layer/module | prerequisite files | RED/GREEN/VERIFY | acceptance commands | stop condition |
-| ---: | --- | --- | --- | --- | --- | --- |
-| 0 | <kebab-case> | <one scope> | <exact paths> | <test-first sequence> | <exact commands> | <condition> |
-
-User approval is required before materializing this draft under `phases/`.
-
-## Executor Handoff
-
-- Executor authorization: separate explicit user request for `scripts/execute.py`.
-- implementation input: approved plan, materialized phase indexes, and Executor-selected current
-  `stepN.md`.
-- implementation recipe: `RED -> observed failure -> minimal GREEN -> focused/full VERIFY`.
-- metadata ownership: Implementation Agent records only the current Step `status` plus `summary`,
-  `error_message`, or `blocked_reason`; Executor records branch, pending-Step selection, retry,
-  timestamps, commits, advancement, and phase status.
-- hooks: `.codex/hooks.json` automatically invokes PreToolUse and Stop; never manually run hook
-  entry points as substitutes.
-
-## Open Issues
-- <requirement, formulation, I/O, required comparison file/mapping, tolerance, or architecture issue>
-```
-
-## 품질 기준
-
-- 모든 `must` requirement는 최소 하나의 task와 test에 연결되어야 한다.
-- C++ production 변경마다 선행 테스트 파일 또는 테스트 추가 계획이 있어야 한다.
-- reference comparison이 필요한 기능은 exact existing input/required CSV path와 FESA
-  HDF5-to-reference-CSV source-ID/component mapping test 계획을 가져야 한다.
-- Implementation Planning Agent는 Harness Step 초안을 사용자에게 승인받은 뒤에만 phase
-  index와 step files를 생성하며 executor는 자동 실행하지 않는다.
-- CMake/CTest 계획은 MSVC x64 Debug 검증 경로와 호환되어야 한다.
-- 구현 계획은 테스트 작성, 실패 확인, 최소 구현, validation 순서를 명시해야 한다.
-- upstream 문서가 불완전하면 값을 임의로 채우지 않고 `needs-upstream-decision` 또는 `blocked`로 표시한다.
-- release 완료나 reference tolerance 통과 판정은 하지 않는다.

docs/io-definitions/README.md

@@ -1,179 +0,0 @@
-# I/O 정의 문서 작성 가이드
-
-이 디렉터리는 I/O Definition Agent가 작성하거나 제안한 기능별 입출력 정의 문서를 보관하는 위치다.
-
-FESA 솔버의 입력 파일은 Abaqus input file이다. 다만 초기 FESA는 Abaqus 전체 문법 호환을 목표로 하지 않고, 기능별로 지원할 Abaqus keyword subset과 내부 모델 매핑을 명확히 정의한다.
-
-기본 파일명은 `docs/io-definitions/<feature-id>-io.md` 형식을 사용한다. 각 문서는 Requirement Agent, Formulation Agent, Numerical Review Agent의 산출물을 입력으로 받아 Abaqus `.inp` 입력 계약, authoritative HDF5 result schema, FESA HDF5 to Reference CSV comparison schema를 정의해야 한다.
-
-## I/O Definition Agent 역할
-
-I/O Definition Agent는 Abaqus input file subset, 내부 solver model mapping, output request mapping, HDF5 result schema, reference CSV comparison row schema를 정의한다.
-
-수행한다:
-- 기능별 supported Abaqus keyword subset을 정의한다.
-- unsupported, ignored-with-warning, requires-user-decision keyword 정책을 정의한다.
-- model data와 history data를 내부 solver 개념으로 매핑한다.
-- node, element, set, material, section, boundary condition, load, step, output request의 의미 계약을 정의한다.
-- `results.h5`의 authoritative HDF5 schema를 정의한다.
-- 기능이 선언한 existing Abaqus CSV와 비교하기 위한 최소 source-ID/component mapping을 정의한다.
-
-수행하지 않는다:
-- parser를 구현하지 않는다.
-- C++ API나 파일 구조를 설계하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- solver 결과와 reference 결과를 비교하지 않는다.
-- release readiness를 승인하지 않는다.
-- 명시적으로 정의되지 않은 Abaqus full compatibility를 주장하지 않는다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> I/O Definition
-
-## Metadata
-- feature_id: <feature-id>
-- source_requirement: docs/requirements/<feature-id>.md
-- source_formulation: docs/formulations/<feature-id>-formulation.md
-- source_numerical_review: docs/numerical-reviews/<feature-id>-review.md
-- source_research: docs/research/<feature-id>-research.md
-- status: draft | needs-user-decision | ready-for-implementation-planning
-- owner_agent: io-definition-agent
-- date: <YYYY-MM-DD>
-
-## Abaqus Input Scope
-- input_format: Abaqus input file (`.inp`)
-- abaqus_documentation_source: <version/source URL>
-- compatibility_disclaimer: FESA supports only the keyword subset defined in this document.
-
-| keyword | support_status | level | required_parameters | mapped_internal_concept | notes |
-| --- | --- | --- | --- | --- | --- |
-| *HEADING | supported | model | N/A | model title | optional |
-| *INCLUDE | supported | model/history | INPUT | include file | path policy required |
-| *NODE | supported | model | N/A | node label and coordinates |  |
-| *NSET | supported | model | NSET | node set | sorted/unsorted policy required |
-| *ELEMENT | supported | model | TYPE | element label, type, connectivity |  |
-| *ELSET | supported | model | ELSET | element set |  |
-| *MATERIAL | supported | model | NAME | material |  |
-| *ELASTIC | supported | model | N/A or TYPE | elastic material data |  |
-| <section keyword> | supported | model | ELSET, MATERIAL | section assignment | e.g. *SOLID SECTION |
-| *BOUNDARY | supported | model/history | N/A | boundary condition |  |
-| *CLOAD | supported | history | N/A | concentrated load | feature-dependent |
-| *DLOAD | supported | history | N/A | distributed load | feature-dependent |
-| *STEP | supported | history | N/A | analysis step |  |
-| <procedure keyword> | supported | history | N/A | analysis procedure | e.g. *STATIC |
-| *OUTPUT | supported | history | N/A | output request root |  |
-| *NODE OUTPUT | supported | history | N/A | nodal output request |  |
-| *ELEMENT OUTPUT | supported | history | N/A | element output request |  |
-
-## Syntax Policy
-- case_insensitivity: <policy>
-- comma_separated_fields: <policy>
-- comment_lines: lines beginning with `**`
-- keyword_continuation: <policy>
-- include_files: <relative path policy>
-- labels: <case/quote/name policy>
-- line_length_limit: <policy>
-- ascii_assumption: <policy>
-- empty_data_fields: <policy>
-
-## Model Data Mapping
-- nodes: <node label, coordinate dimension, coordinate system>
-- elements: <element label, element type, connectivity>
-- node_sets: <set name, sorted/unsorted, expansion rules>
-- element_sets: <set name, expansion rules>
-- material: <material name and data>
-- section: <section assignment>
-- coordinates: <global/local coordinate conventions>
-- units: <unit system policy>
-
-## History Data Mapping
-- steps: <step naming and order>
-- procedure: <analysis procedure keyword mapping>
-- boundary_conditions: <boundary condition mapping>
-- loads: <load keyword mapping>
-- output_requests: <node/element output mapping>
-
-## Internal Model Contract
-- node_label: <semantic contract>
-- element_label: <semantic contract>
-- element_type: <semantic contract>
-- connectivity: <semantic contract>
-- set_membership: <semantic contract>
-- material: <semantic contract>
-- section: <semantic contract>
-- boundary_condition: <semantic contract>
-- load: <semantic contract>
-- step: <semantic contract>
-- output_request: <semantic contract>
-
-## Output HDF5 Schema
-
-- authoritative_file: `results.h5`
-- schema_version: <version>
-- root_attributes: units, coordinate_system, solver_version, feature_id, model_id
-- step_frame_identity: <step/frame naming and ordering>
-- row_identity_policy: <node_id | element_id | integration_point | component mapping>
-
-| quantity | dataset_path | shape | dtype | required_attributes | location | notes |
-| --- | --- | --- | --- | --- | --- | --- |
-| displacement | /steps/<step>/frames/<frame>/field_outputs/U | <nnode, ndim> | float64 | component_names, units | nodal |  |
-| reaction | /steps/<step>/frames/<frame>/field_outputs/RF | <nnode, ndim> | float64 | component_names, units | nodal |  |
-| element_force | /steps/<step>/frames/<frame>/field_outputs/element_forces | <nrow, ncomp> | float64 | component_names, element_ids, location | element | feature-dependent |
-| stress | /steps/<step>/frames/<frame>/field_outputs/S | <nrow, ncomp> | float64 | component_names, element_ids, integration_points | integration_point | feature-dependent |
-
-## FESA HDF5 to Reference CSV Comparison Mapping
-
-FESA solver output은 `results.h5`이다. 기능이 blocking 또는 warning-only로 선언한
-quantity에 대해서만 existing CSV path와 다음 mapping을 정의한다.
-
-| field | contract |
-| --- | --- |
-| reference_input | exact existing `.inp` path |
-| reference_csv | exact existing required CSV path |
-| hdf5_dataset | authoritative FESA dataset path |
-| source_identity | node/element source label used for exact matching |
-| components | required CSV columns and corresponding HDF5 components |
-| row_precheck | missing/extra/duplicate/nonfinite required rows fail before tolerance |
-| behavior | blocking or warning-only |
-| tolerance | upstream-approved formula |
-
-Row order alone으로 대응하지 않는다. Canonical filename, reference CSV schema version,
-README, metadata, provenance 또는 single-step/final-frame case의 duplicated unit/coordinate/
-step-frame columns를 요구하지 않는다.
-
-## Validation Rules
-- required_fields: <required input fields>
-- duplicate_labels: <policy>
-- missing_references: <policy>
-- unsupported_keywords: unsupported | ignored-with-warning | requires-user-decision
-- set_expansion: <policy>
-- coordinate_conventions: <policy>
-- output_quantity_availability: <policy>
-- hdf5_schema_validation: <required datasets and attributes>
-- reference_csv_validation: <stable ordering and required columns>
-
-## Open Issues and Downstream Handoff
-
-### Reference Model Agent
-- <Abaqus input examples and reference artifact schema needs>
-
-### Implementation Planning Agent
-- <parser acceptance cases, unsupported keyword diagnostics, HDF5 writer tests, comparison row mapping tests>
-
-### Reference Verification Agent
-- <HDF5 dataset paths, reference CSV row schemas, ID matching rules, units, coordinate conventions, tolerance-relevant fields>
-```
-
-## 품질 기준
-
-- 입력 파일은 Abaqus `.inp`임을 명시해야 한다.
-- Abaqus full compatibility를 주장하지 않고 기능별 supported keyword subset을 명시해야 한다.
-- model data와 history data의 매핑을 구분해야 한다.
-- unsupported keyword 처리 정책을 명확히 해야 한다.
-- 내부 모델 계약은 semantic fields로 작성하고 C++ class/function/API를 확정하지 않는다.
-- `results.h5`가 authoritative solver output임을 명시해야 한다.
-- HDF5 schema는 dataset path, dtype/shape, required attributes, ID field, component naming, coordinate system, units, step/frame identity, quantity location을 포함해야 한다.
-- Reference CSV comparison mapping은 exact path, source-ID column, required component columns,
-  HDF5 projection, row prechecks와 tolerance를 포함해야 한다.

docs/linear-static-3d-euler-beam/build-test.md

@@ -4,8 +4,8 @@
 
 - feature_id: `linear-static-3d-euler-beam`
 - source_commit: `400db191ce9f766ca6b34e5b609eaa13c54ccfa3`
-- source_implementation_report: `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`
-- source_implementation_plan: `docs/implementation-plans/linear-static-3d-euler-beam.md`
+- source_implementation_report: `docs/linear-static-3d-euler-beam/implementation-report.md`
+- source_implementation_plan: `docs/linear-static-3d-euler-beam/implementation-plan.md`
 - status: `pass-for-reference-verification`
 - owner_agent: `build-test-executor-agent`
 - date: `2026-08-09`

docs/linear-static-3d-euler-beam/coordination.md

@@ -0,0 +1,41 @@
+# Linear Static 3D Euler Beam Coordination Report
+
+## Metadata
+
+- feature_id: linear-static-3d-euler-beam
+- status: completed
+- owner_agent: coordinator-agent
+- date: 2026-08-15
+- migration_note: Existing evidence was relocated without rerunning workflow gates.
+
+## Worklist
+
+| stage | owner | status | evidence |
+| --- | --- | --- | --- |
+| 1. requirements | requirement-agent | passed | `requirements.md` |
+| 2. research | research-agent | passed | `research.md` |
+| 3. formulation | formulation-agent | passed | `formulation.md` |
+| 4. numerical and reference | numerical-review-agent | passed | `numerical-review.md`; `reference-model.md` |
+| 5. I/O | io-definition-agent | passed | `io.md` |
+| 6. implementation and verification | implementation-planning-agent; implementation-agent | passed | `implementation-plan.md`; `implementation-report.md`; `build-test.md`; `reference-comparison.md` |
+| 7. physics | physics-evaluation-agent | passed | `physics-evaluation.md` |
+| 8. release | release-agent | passed | `release.md` |
+
+## Current Progress
+
+- active_sub_agent: none
+- next_handoff: none
+- closure_evidence: `release.md` status `ready-for-release`
+- workflow_disposition: completed
+
+## Decision Log
+
+| date | decision | evidence | rationale |
+| --- | --- | --- | --- |
+| 2026-08-15 | Record migrated workflow as completed | `release.md` | Preserve the existing release verdict without retroactive gate execution. |
+
+## Blockers and Rework
+
+- active_blockers: none
+- correction_attempt_count: 0
+- needs_user_decision: false

docs/linear-static-3d-euler-beam/formulation.md

@@ -3,8 +3,8 @@
 ## Metadata
 
 - feature_id: `3d-isoparametric-euler-beam`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - status: `ready-for-numerical-review`
 - owner_agent: `formulation-agent`
@@ -1311,9 +1311,10 @@ $$
 
 가 일치해야 한다. 따라서 이 V0 조건의 equilibrium end action은 요소 끝에 작용하는
 outward action이고, endpoint section resultant는 동일 section을 양의 local $x$ 면으로
-자른 값이다. Abaqus node-station force row에는 equilibrium end action이 아니라 section
-resultant를 동일한 section-cut 부호로 정규화해 비교하며, interior node의 두 endpoint가
-승인 tolerance 안에서 먼저 일치해야 한다. 승인된 component mapping은
+자른 값이다. Abaqus element-endpoint force row에는 equilibrium end action이 아니라
+section resultant를 동일한 section-cut 부호로 정규화해 비교한다. Reference의 element
+label과 endpoint node label을 HDF5 row에 직접 대응하며 node station collapse, 대표 endpoint
+선택 또는 평균은 하지 않는다. 승인된 component mapping은
 `SF1 -> N`, `SM1 -> My`, `SM2 -> Mz`, `SM3 -> T`이고, CSV에 없는 transverse
 `SF2`, `SF3`는 reference comparison 대상이 아니다.
 
@@ -1529,19 +1530,25 @@ $$
 \sum\mathbf M_{\mathrm{reaction}}\right\|\le\epsilon_M.
 $$
 
-승인된 B33 reference row는 같은 model, step/frame, quantity, component별로 Abaqus
-reference 값만 사용해
+승인된 B33 reference row는 같은 model, step/frame, quantity, component로 family를 만들고
+Abaqus reference 값만 사용해
 
 $$
-\operatorname{reference\_scale}=\max_i|r_i|,\qquad
-\operatorname{row\_tolerance}=\operatorname{absolute\_floor}
-+10^{-6}\operatorname{reference\_scale}
+S=\max_i|r_i|
 $$
 
-를 적용한다. 승인된 SI bundle의 displacement/rotation floor는 $10^{-9}$,
-force/moment floor는 $10^{-3}$이다. Reference 값을 zero-clamp하거나 row를 제거하지
-않으며 missing, extra, duplicate, nonfinite, schema/identity mismatch는 수치 판정 전에
-실패한다. 이 reference policy는 위 formulation/analytical tolerance를 대체하지 않는다.
+를 계산한다. $|r_i|\le0.01S$인 near-zero row는 $|f_i-r_i|\le0.01S$로,
+나머지 row는 $|f_i-r_i|/|r_i|\le0.05$로 판정한다. 또한
+
+$$
+\frac{\sqrt{n^{-1}\sum_i(f_i-r_i)^2}}{S}\le0.01
+$$
+
+을 family 전체에서 만족해야 한다. $S=0$이면 모든 FESA 값이 정확히 0일 때만 통과하고,
+그렇지 않으면 zero-reference-scale-nonzero-error로 실패한다. 독립적인 절대오차 gate는
+사용하지 않는다. Reference 값을 zero-clamp하거나 row를 제거하지 않으며 missing, extra,
+duplicate, nonfinite, schema/identity mismatch는 수치 판정 전에 실패한다. 이 reference
+policy는 위 formulation/analytical tolerance를 대체하지 않는다.
 
 ## 19. Numerical Risks
 
@@ -1584,7 +1591,7 @@ deficiency는 별도의 under-integration 위험이다.
 ### 20.2 근거의 적용 경계
 
 Tier와 provenance가 정리된 research brief는
-`docs/research/linear-static-3d-euler-beam-research.md`에 있다. Wiki와 원출처는 핵심
+`docs/linear-static-3d-euler-beam/research.md`에 있다. Wiki와 원출처는 핵심
 beam 이론과 Abaqus component 의미를 제공하고, 이 문서의 DOF 순서,
 $\theta_y=-w'$ 부호, guide-vector 축, 12×12 행렬, line-load vector, 결과 위치와
 tolerance는 승인된 project requirement/design과 결합한 FESA 계약이다. 따라서 exact

docs/linear-static-3d-euler-beam/implementation-plan.md

@@ -11,12 +11,12 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-3d-euler-beam-review.md`
-- source_io_definition: `docs/io-definitions/linear-static-3d-euler-beam-io.md`
-- source_reference_models: `docs/reference-models/linear-static-3d-euler-beam-reference-models.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_numerical_review: `docs/linear-static-3d-euler-beam/numerical-review.md`
+- source_io_definition: `docs/linear-static-3d-euler-beam/io.md`
+- source_reference_models: `docs/linear-static-3d-euler-beam/reference-model.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - phase_steps: `phases/linear-static-3d-euler-beam/step7.md` through `step24.md`
 - status: `ready-for-implementation`
@@ -519,7 +519,7 @@ Every task below is one Harness Step and one independent review gate. Each task
 | create | `src/fesa/CMakeLists.txt` | `fesa_solver`, warning isolation, and later production source registration |
 | create | `tests/CMakeLists.txt` | GoogleTest targets/discovery, common feature label, test meta-target |
 | create | `tests/unit/build_info_test.cpp` | compile/runtime contract for solver version |
-| create | `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md` | cumulative Step 7-24 evidence ledger |
+| create | `docs/linear-static-3d-euler-beam/implementation-report.md` | cumulative Step 7-24 evidence ledger |
 
 - Consumes: repository toolchain and approved dependency locations; no solver module.
 - Produces: normalized dependency aliases `Fesa::MKL`, `Fesa::TBB`, and `Fesa::HDF5`; `fesa_solver`; `fesa_unit_tests`; custom target `fesa_tests`; common CTest label `linear-static-3d-euler-beam`; and `solverVersion()`.
@@ -1071,7 +1071,7 @@ if ($LASTEXITCODE -eq 0) { throw "HDF5 API leaked into public headers:`n$hdf5Lea
 
 ## Implementation Report Evidence Contract
 
-`docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md` is created in TASK-07 and appended in TASK-08 through TASK-24. It must use this fixed structure for every Step:
+`docs/linear-static-3d-euler-beam/implementation-report.md` is created in TASK-07 and appended in TASK-08 through TASK-24. It must use this fixed structure for every Step:
 
 ```markdown
 ## Step 16 — euler-beam-element

docs/linear-static-3d-euler-beam/implementation-report.md

@@ -3,7 +3,7 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_plan: `docs/implementation-plans/linear-static-3d-euler-beam.md`
+- source_plan: `docs/linear-static-3d-euler-beam/implementation-plan.md`
 - source_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - status: `in-progress`
 - owner_agent: `implementation-agent`
@@ -18,7 +18,7 @@
   `include/fesa/build_info.hpp`, `src/fesa/build_info.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
   `tests/unit/build_info_test.cpp`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`
+  `docs/linear-static-3d-euler-beam/implementation-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-030`, `FESA-REQ-LS3DEB-034`
 - test_ids: `T07-BUILD-001`, `T07-BUILD-002`
 
@@ -56,7 +56,7 @@
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
   `tests/unit/core/source_identity_test.cpp`,
   `tests/unit/core/diagnostic_test.cpp`, `tests/unit/core/status_test.cpp`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`
+  `docs/linear-static-3d-euler-beam/implementation-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-033`, `FESA-REQ-LS3DEB-034`
 - test_ids: `T08-CORE-001`, `T08-CORE-002`, `T08-CORE-003`
 
@@ -95,7 +95,7 @@
   `src/fesa/math/matrix.cpp`, `tests/unit/math/vector_test.cpp`,
   `tests/unit/math/matrix_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-025`, `FESA-REQ-LS3DEB-034`
 - test_ids: `T09-DENSE-001`, `T09-DENSE-002`
@@ -153,7 +153,7 @@
   `include/fesa/model/domain.hpp`, `src/fesa/model/domain.cpp`,
   `tests/unit/model/model_types_test.cpp`, `tests/unit/model/domain_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-10-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-015`, `FESA-REQ-LS3DEB-016`,
@@ -200,7 +200,7 @@
   `tests/unit/io/abaqus/input_syntax_test.cpp`,
   `tests/unit/io/abaqus/input_reader_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-11-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-010`, `FESA-REQ-LS3DEB-034`,
@@ -248,7 +248,7 @@
   `src/fesa/io/abaqus/domain_mapper.cpp`,
   `tests/unit/io/abaqus/domain_mapper_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-12-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-001`, `FESA-REQ-LS3DEB-002`,
@@ -368,7 +368,7 @@
   `src/fesa/analysis/analysis_model.cpp`,
   `tests/unit/analysis/analysis_model_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-13-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-001`, `FESA-REQ-LS3DEB-021`,
@@ -413,7 +413,7 @@
 - changed_files: `include/fesa/fem/dof_manager.hpp`,
   `src/fesa/fem/dof_manager.cpp`, `tests/unit/fem/dof_manager_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-14-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-003`, `FESA-REQ-LS3DEB-007`,
@@ -470,7 +470,7 @@
   `tests/unit/results/result_records_test.cpp`,
   `tests/unit/analysis/analysis_state_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-15-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-003`, `FESA-REQ-LS3DEB-023`,
@@ -522,7 +522,7 @@
   `src/fesa/elements/euler_beam_3d.cpp`,
   `tests/unit/elements/euler_beam_3d_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-16-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-002`, `FESA-REQ-LS3DEB-004`,
@@ -639,7 +639,7 @@
   `src/fesa/assembly/parallel_for.cpp`,
   `tests/unit/assembly/parallel_for_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`,
   `.superpowers/sdd/linear-static-3d-euler-beam/task-17-report.md`
 - requirement_ids: `FESA-REQ-LS3DEB-024`, `FESA-REQ-LS3DEB-025`,
@@ -728,7 +728,7 @@
   `tests/unit/math/sparse_matrix_test.cpp`,
   `tests/unit/assembly/sparse_assembler_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-024`, `FESA-REQ-LS3DEB-025`,
   `FESA-REQ-LS3DEB-034`, `FESA-REQ-LS3DEB-035`
@@ -818,7 +818,7 @@
   `src/fesa/constraints/essential_constraints.cpp`,
   `tests/unit/constraints/essential_constraints_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-007`, `FESA-REQ-LS3DEB-022`,
   `FESA-REQ-LS3DEB-027`, `FESA-REQ-LS3DEB-034`,
@@ -871,7 +871,7 @@
   `tests/unit/solvers/linear/linear_solver_test.cpp`,
   `tests/unit/solvers/linear/mkl_pardiso_solver_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-025`, `FESA-REQ-LS3DEB-026`,
   `FESA-REQ-LS3DEB-034`, `FESA-REQ-LS3DEB-035`
@@ -978,7 +978,7 @@
   `src/fesa/assembly/load_assembler.cpp`,
   `tests/unit/assembly/load_assembler_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-007`, `FESA-REQ-LS3DEB-011`,
   `FESA-REQ-LS3DEB-027`, `FESA-REQ-LS3DEB-034`
@@ -1030,7 +1030,7 @@
   `src/fesa/results/result_recovery.cpp`,
   `tests/unit/results/result_recovery_test.cpp`, `src/fesa/CMakeLists.txt`,
   `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-004`, `FESA-REQ-LS3DEB-027`,
   `FESA-REQ-LS3DEB-031`, `FESA-REQ-LS3DEB-032`,
@@ -1142,7 +1142,7 @@
   `tests/unit/io/hdf5/hdf5_results_writer_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
   `cmake/FesaDependencies.cmake`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-003`, `FESA-REQ-LS3DEB-015`,
   `FESA-REQ-LS3DEB-019`, `FESA-REQ-LS3DEB-020`,
@@ -1225,7 +1225,7 @@
   `tests/reference/reference_comparison_test.cpp`,
   `tests/reference/b33_reference_comparison_test.cpp`,
   `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt`,
-  `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`,
+  `docs/linear-static-3d-euler-beam/implementation-report.md`,
   `phases/linear-static-3d-euler-beam/index.json`
 - requirement_ids: `FESA-REQ-LS3DEB-001`, `FESA-REQ-LS3DEB-002`,
   `FESA-REQ-LS3DEB-005`, `FESA-REQ-LS3DEB-020`,

docs/linear-static-3d-euler-beam/io.md

@@ -3,16 +3,16 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-3d-euler-beam-review.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_numerical_review: `docs/linear-static-3d-euler-beam/numerical-review.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - status: `ready-for-implementation-planning`
 - owner_agent: `io-definition-agent`
 - date: `2026-08-09`
 - authoritative_output: `results.h5`
-- reference_baseline: `reference/cantilever beam/` at source commit `2b34d0b`
+- reference_baseline: exact read-only artifacts under `reference/cantilever beam/`
 
 이 문서는 승인된 V0의 semantic I/O contract만 정의한다. Parser, HDF5 writer,
 comparison tooling의 C++ API나 구현 구조는 정의하지 않으며 Abaqus full compatibility를
@@ -336,7 +336,7 @@ CSV는 `SF1/SM1/SM2/SM3`만 제공하고 section-cut result와 비교하므로 `
 0으로 만들거나 reference row로 합성하지 않는다. HDF5의 transverse end action은
 unit/analytical test와 physics sanity 대상이다.
 
-### Frame, instance, and node-station normalization
+### Frame, instance, and element-endpoint normalization
 
 - Legacy `Frame` value `Increment 1: Step Time = 1.000`은 canonical `(Step-1, frame 0)`으로
   변환한다. 승인 bundle의 다른 increment/time string은 `schema-mismatch`다.
@@ -344,29 +344,27 @@ unit/analytical test와 physics sanity 대상이다.
   `instance_name`과 case-insensitive lookup 후 raw identity 일치 여부를 확인한다.
 - `Node Label`은 instance 안의 preserved source node label로 resolve한다. Displacement와
   reaction key는 `(model_id,Step-1,0,instance_name,source_node_label,quantity,component)`다.
-- Elemental-force CSV는 element label이 없으므로 HDF5 endpoint section resultants를
-  source node station으로 project한다. Boundary station은 유일한 incident endpoint를 쓴다.
-- Interior station collapse는 정확히 두 incident B33 endpoints, 동일 section/local-axis
-  orientation, 일관된 chain connectivity, 해당 node의 concentrated force/moment가 없는
-  경우에만 허용한다. 두 positive-face section-cut 값이 아래 승인 component tolerance
-  안에서 먼저 일치해야 한다.
-- Interior 값이 일치하면 stable internal element ID가 작은 endpoint를 deterministic
-  representative로 선택한다. 두 값을 평균하지 않는다. 불일치는 `tolerance-failure`다.
-- Reversed connectivity, local-axis discontinuity, section jump, branch 또는 loaded interior
-  station은 element label 없는 legacy schema로 collapse할 수 없다. 이 approved bundle
-  밖에서는 element-aware reference row가 필요하며 legacy projection은 `schema-mismatch`로
-  중단한다.
+- Elemental-force CSV의 `Element Label`은 instance 안의 preserved source element label로,
+  `Node Label`은 해당 B33 connectivity의 정확한 endpoint source node로 resolve한다.
+- Section-resultant key는
+  `(model_id,Step-1,0,instance_name,source_element_label,source_node_label,quantity,component)`다.
+  한 element에는 connectivity와 일치하는 두 endpoint 행이 정확히 존재해야 한다.
+- Missing, extra, duplicate 또는 element-connectivity-mismatched endpoint row는
+  `schema-mismatch`로 tolerance 전에 실패한다. Source node station collapse,
+  deterministic representative 선택 또는 endpoint 평균은 사용하지 않는다.
 
 Projected canonical comparison row는 다음 fields를 가진다.
 
 ```text
-model_id, step_name, frame_index, instance_name, source_node_label,
-quantity, component, value, unit_dimension, coordinate_system, hdf5_dataset_path
+model_id, step_name, frame_index, instance_name,
+[source_element_label], source_node_label, quantity, component, value,
+unit_dimension, coordinate_system, hdf5_dataset_path
 ```
 
-Stable ordering은 quantity inventory order, instance declaration order, stable source node
-order, component order다. Approved model ID는 `cantilever-beam-b33`; nodal quantities는
-global Cartesian, section resultants는 beam local이다.
+`source_element_label`은 section-resultant row에 필수이고 nodal row에는 적용하지 않는다.
+Stable ordering은 quantity inventory order, instance declaration order, stable source
+node/element-endpoint order, component order다. Approved model ID는 `cantilever-beam-b33`;
+nodal quantities는 global Cartesian, section resultants는 beam local이다.
 
 ### Row-set precheck and tolerance
 
@@ -376,27 +374,32 @@ identity-mismatched row가 하나라도 있으면 tolerance 계산 전에 실패
 mandatory인 `SF2/SF3` 대응 end action, generalized results와 `S11`은 계약상 비교 대상이
 아니므로 extra reference row가 아니다.
 
-Matched rows는 같은 `model_id`, step/frame, quantity, component로 group한다.
+Matched rows는 같은 model/case, step/frame, logical quantity, unit dimension, coordinate
+system과 blocking behavior의 component family로 group한다. B33 family는 translation
+`UX/UY/UZ`, rotation `URX/URY/URZ`, reaction force `RF1/RF2/RF3`, reaction moment
+`RM1/RM2/RM3`, section force `N`, section moment `T/My/Mz`다.
 
 ```text
-reference_scale = max(abs(reference_value_i))
-row_tolerance = absolute_floor + 1e-6 * reference_scale
-row_pass = abs(fesa_value_i - reference_value_i) <= row_tolerance
+S = max(abs(reference_value_i))
+error_i = abs(fesa_value_i-reference_value_i)
+near_zero_band = 0.01*S
+
+if abs(reference_value_i) <= near_zero_band:
+    row_pass = error_i <= near_zero_band
+else:
+    row_pass = error_i/abs(reference_value_i) <= 0.05
+
+relative_rms = sqrt(mean(error_i^2))/S
+family_pass = all(row_pass) and relative_rms <= 0.01
 ```
 
-즉 exact policy는 `absolute_floor + 1e-6 * reference_scale`이다. `reference_scale`은
-read-only Abaqus values만 사용하고 FESA 값으로 조정하지 않는다. Scale이 zero면 relative
-term은 zero다. Reference value나 작은 residue를 zero-clamp하지 않고 모든 row를 판정한다.
-
-| approved SI component class | absolute floor |
-| --- | ---: |
-| displacement and rotation | `1e-9` |
-| force and moment | `1e-3` |
-
-Interior endpoint consistency도 해당 CSV quantity/component의 same Abaqus-only scale과
-floor를 사용한다. Verification report는 모든 row pass/fail과 quantity별 max absolute
-error, component-scale normalized error, RMS error, norm error, worst row/component를
-기록한다. 이 reference tolerance는 analytical/formulation tolerance를 대체하지 않는다.
+`S`는 read-only Abaqus values만 사용하고 FESA 값으로 조정하지 않는다. 독립적인
+absolute-error gate와 zero clamp는 사용하지 않는다. `S=0`이면 모든 FESA 값도 exact
+zero일 때 relative RMS를 0으로 기록하고 통과하며, 하나라도 nonzero이면 NaN/Inf 대신
+`zero-reference-scale-nonzero-error`로 실패한다. Verification report는 모든 row의 적용
+branch와 pass/fail, family identity/components, scale, near-zero band/count, max absolute
+error, scale-relative RMS와 worst row/component를 기록한다. 이 reference tolerance는
+analytical/formulation tolerance를 대체하지 않는다.
 
 ## CLI and Diagnostics Contract
 
@@ -445,17 +448,15 @@ release approval을 이 status가 의미하지 않는다.
 
 ### Resolved numerical-review handoff
 
-`NR-O03-STATION-NORMALIZATION`은 approved legacy bundle에 대해 unloaded, consistently
-oriented two-endpoint interior station만 collapse하고, tolerance check 후 smaller stable
-element ID를 선택하는 규칙으로 구체화했다. Reversed/branched/loaded/jumped station은
-element-aware reference가 없는 한 비교하지 않는다.
+`NR-O03-STATION-NORMALIZATION`은 2026-08-18 regenerated elemental-force CSV의
+`Element Label`로 해소됐다. Comparator는 `(instance, element label, endpoint node label,
+component)`를 직접 대응하며 station collapse, 대표 endpoint 선택 또는 평균을 하지 않는다.
 
 ### Reference Model Agent
 
-- Exact legacy inventory, generator `Abaqus/CAE Learning Edition 2024`, source commit
-  `2b34d0b`, external SI provenance와 stress comparison N/A를 계약에 고정한다.
-- 추가 reference model은 canonical filenames와 metadata를 사용하며 이 legacy file을
-  변경하지 않는다.
+- Exact legacy inventory, current row schema와 stress comparison N/A를 계약에 고정한다.
+- 추가 reference case의 exact paths와 수치 비교 계약은 해당 feature requirement가 정하며,
+  canonical naming이나 optional metadata를 readiness 조건으로 추가하지 않는다.
 
 ### Implementation Planning Agent
 
@@ -467,7 +468,7 @@ element-aware reference가 없는 한 비교하지 않는다.
 
 ### Reference Verification Agent
 
-- Artifact precheck 뒤 HDF5-to-legacy projection, node-station eligibility, row-set equality,
-  component-scale comparison 순서를 유지한다.
+- Artifact precheck 뒤 HDF5-to-CSV source identity projection, direct element-endpoint row-set
+  equality, common family-scale comparison 순서를 유지한다.
 - Missing/extra/nonfinite row를 무시하거나 `SF2/SF3`/stress reference row를 합성하지 않는다.
 

docs/linear-static-3d-euler-beam/numerical-review.md

@@ -3,9 +3,9 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_requirements: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_requirements: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - status: `pass-for-implementation-planning`
 - owner_agent: `numerical-review-agent`
@@ -191,8 +191,13 @@ Confirmed defects, risks, and open issues are separated.
 2. `NR-O02-DETERMINISTIC-REDUCTION`: stable COO sort and duplicate-summation rules are project policy and must be made explicit before NR-T11.
 3. `NR-O03-STATION-NORMALIZATION`: reversed connectivity/local-axis orientation and legitimate jumps at loaded interior nodes need an explicit downstream row-normalization/eligibility rule. The legacy baseline may use its documented stable orientation and unloaded interior stations, but mismatch must never be averaged. NR-T07 covers element signs.
 
+2026-08-18 amendment: `NR-O03-STATION-NORMALIZATION` is resolved and superseded for the
+approved case. The regenerated elemental-force CSV supplies `Element Label`, so comparison uses
+direct `(instance, element label, endpoint node label, component)` identity and never collapses,
+selects or averages adjacent endpoints.
+
 No open issue requires formulation revision. NR-O01/NR-O02 are implementation-planning
-handoffs; NR-O03 belongs to I/O and reference-model contracts.
+handoffs; the resolved NR-O03 identity is fixed by the I/O and reference-model contracts.
 
 ## Required Revisions
 
@@ -206,7 +211,7 @@ handoffs; NR-O03 belongs to I/O and reference-model contracts.
 
 ### Reference Model Agent
 
-- Make NR-O03 orientation and unloaded-interior assumptions explicit without modifying the approved legacy artifacts.
+- Enforce the resolved NR-O03 direct element-endpoint identity without modifying reference artifacts.
 
 ## Downstream Handoff
 

docs/linear-static-3d-euler-beam/physics-evaluation.md

@@ -5,12 +5,12 @@
 - feature_id: `linear-static-3d-euler-beam`
 - model_id: `cantilever-beam-b33`
 - evaluated_head: `d76d052456ec134a98bcd5aa3b3c18a6b0ad6ba4`
-- source_reference_verification_report: `docs/reference-verifications/linear-static-3d-euler-beam-reference-verification.md`
-- source_reference_model: `docs/reference-models/linear-static-3d-euler-beam-reference-models.md`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-3d-euler-beam-review.md`
-- source_io_definition: `docs/io-definitions/linear-static-3d-euler-beam-io.md`
+- source_reference_verification_report: `docs/linear-static-3d-euler-beam/reference-comparison.md`
+- source_reference_model: `docs/linear-static-3d-euler-beam/reference-model.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_numerical_review: `docs/linear-static-3d-euler-beam/numerical-review.md`
+- source_io_definition: `docs/linear-static-3d-euler-beam/io.md`
 - status: `pass-for-release-agent`
 - owner_agent: `physics-evaluation-agent`
 - date: `2026-08-09`
@@ -26,7 +26,7 @@ artifact identity was reproduced before the physics checks and again after the t
 
 | evidence | exact path or identity | status | notes |
 | --- | --- | --- | --- |
-| reference verification | `docs/reference-verifications/linear-static-3d-euler-beam-reference-verification.md` | pass-for-physics-evaluation | Required gate satisfied. |
+| reference verification | `docs/linear-static-3d-euler-beam/reference-comparison.md` | pass-for-physics-evaluation | Required gate satisfied. |
 | solver HDF5 | `.harness/build/reference/cantilever-beam-b33/results.h5` | present and readable | 25,336 bytes; post-acceptance-run SHA-256 `13ECCF68262C14BBDE0F63667C0F10896ACD40EFEC56E8C9121C298333FD9B6D`. |
 | comparison evidence | `.harness/build/reference/cantilever-beam-b33/comparison.json` | present and passing | 128,118 bytes; SHA-256 `258347AEA791D981AEA9B2BCAD85DE5344D4859ECA3692DC5E7AA01A848F8E0D`; `passed=true`, 176 rows, 16 metrics. |
 | reference input | `reference/cantilever beam/cantilever beam.inp` | exact read-only artifact | SHA-256 `E406EA9560321B791DB829E03BD24593B9875E0195D35B86BD931EDA122EF3`; `TYPE=B33`. |

docs/linear-static-3d-euler-beam/reference-comparison.md

@@ -5,16 +5,50 @@
 - feature_id: `linear-static-3d-euler-beam`
 - model_id: `cantilever-beam-b33`
 - source_head: `451d9077ea70e3087454db3760e677da0095d27f`
-- source_build_test_report: `docs/build-test-reports/linear-static-3d-euler-beam.md`
-- source_reference_models: `docs/reference-models/linear-static-3d-euler-beam-reference-models.md`
-- source_io_definition: `docs/io-definitions/linear-static-3d-euler-beam-io.md`
-- source_implementation_plan: `docs/implementation-plans/linear-static-3d-euler-beam.md`
-- source_implementation_report: `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`
-- status: `pass-for-physics-evaluation`
+- source_build_test_report: `docs/linear-static-3d-euler-beam/build-test.md`
+- source_reference_models: `docs/linear-static-3d-euler-beam/reference-model.md`
+- source_io_definition: `docs/linear-static-3d-euler-beam/io.md`
+- source_implementation_plan: `docs/linear-static-3d-euler-beam/implementation-plan.md`
+- source_implementation_report: `docs/linear-static-3d-euler-beam/implementation-report.md`
+- historical_status: `pass-for-physics-evaluation`
+- current_status: `pass-for-physics-evaluation`
+- superseded_on: `2026-08-18`
+- revalidated_on: `2026-08-18`
 - owner_agent: `reference-verification-agent`
 - date: `2026-08-09`
 - review_fix_date: `2026-08-10`
 
+The 2026-08-09 evidence below is preserved as a historical record of the former 11-station,
+component-scale comparison. ADR-022 and the regenerated 20-row element-endpoint CSV supersede
+that identity and tolerance. The following revalidation is the current reference-gate evidence.
+
+## 2026-08-18 Common-policy Revalidation
+
+`cmake --build .harness/build --config Debug` passed, followed by
+`ctest --test-dir .harness/build -C Debug --output-on-failure`: 214/214 tests passed.
+The B33 reference test generated
+`.harness/build/reference/cantilever-beam-b33/comparison.json` from the exact read-only
+reference paths.
+
+- exact canonical rows: 212 = 66 displacement + 66 reaction + 80 direct element-endpoint
+  section-resultant rows
+- exact families: 6; all row gates and family RMS gates passed
+- direct section identity: 20 endpoint rows, 10 B33 elements x 2 connectivity-matched endpoints
+- warnings, identity/schema/nonfinite failures: 0
+- overall verdict: `passed=true`
+
+| family | rows | scale | near-zero rows | max absolute error | scale-relative RMS |
+| --- | ---: | ---: | ---: | ---: | ---: |
+| translation `UX/UY/UZ` | 33 | `1.90476272e-2` | 23 | `5.333229170789711e-10` | `8.459623217000381e-9` |
+| rotation `URX/URY/URZ` | 33 | `2.85714399e-3` | 23 | `1.000013943180944e-10` | `1.1723557271997219e-8` |
+| reaction force `RF1/RF2/RF3` | 33 | `1.0e6` | 32 | `8.195638656616211e-7` | `2.0922817757699256e-13` |
+| reaction moment `RM1/RM2/RM3` | 33 | `1.0e7` | 32 | `5.0514936447143555e-6` | `8.811160558471221e-14` |
+| section force `N` | 20 | `0` | 20 | `0` | `0` |
+| section moment `T/My/Mz` | 60 | `1.0e7` | 41 | `1.2499958951957524e-1` | `2.2910664449287597e-9` |
+
+All values above come from the generated JSON ledger. The historical report body starts below
+and must not be read as the current identity/tolerance evidence.
+
 The prerequisite build/test report has status
 `pass-for-reference-verification`. This report applies only the approved Abaqus
 B33 reference tolerance. It does not approve physics sanity or release readiness.

docs/linear-static-3d-euler-beam/reference-model.md

@@ -3,18 +3,19 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-3d-euler-beam-review.md`
-- source_io_definition: `docs/io-definitions/linear-static-3d-euler-beam-io.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_numerical_review: `docs/linear-static-3d-euler-beam/numerical-review.md`
+- source_io_definition: `docs/linear-static-3d-euler-beam/io.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - status: `ready-for-implementation-planning`
 - owner_agent: `reference-model-agent`
 - date: `2026-08-09`
+- amended_on: `2026-08-18`
 - approved_reference_model: `cantilever-beam-b33`
-- approved_reference_schema: `abaqus-cae-report-csv-v0`
-- reference_baseline: `reference/cantilever beam/` at source commit `2b34d0b`
+- approved_reference_identity: direct source-node and element-endpoint rows
+- reference_baseline: exact current files under `reference/cantilever beam/`
 
 이 문서는 구현 전에 필요한 code verification, analytical solution verification 및
 approved B33 reference comparison의 모델·artifact 계약을 정의한다. 이 status는 모델과
@@ -45,7 +46,7 @@ test fixtures이며 reference artifact bundle로 가장하지 않는다.
 논리 모델 `cantilever-beam-b33`의 exact read-only legacy bundle만 사용하여 FESA
 `results.h5`의 displacement, reaction 및 endpoint section resultant를 Abaqus/CAE report
 CSV row와 비교한다. Artifact precheck와 exact row-set matching이 먼저 통과해야 하며,
-수치 비교는 component-scale mixed tolerance를 사용한다. Axial `S11` output은 필수지만
+수치 비교는 ADR-022의 common family-scale tolerance를 사용한다. Axial `S11` output은 필수지만
 Abaqus beam stress comparison은 명시적 N/A다.
 
 ### Excluded validation scope
@@ -136,8 +137,8 @@ single-step deck and changes only the named condition.
 - boundary_conditions: source node 1, DOFs 1 through 6 fixed
 - load: source node 11, global DOF 3, magnitude `-1e6` N
 - model_id: `cantilever-beam-b33`
-- logical_schema: `abaqus-cae-report-csv-v0`
-- source_commit: `2b34d0b`
+- historical_schema_record: `abaqus-cae-report-csv-v0` for the 2026-08-09 inventory only
+- historical_source_commit: `2b34d0b` for the 2026-08-09 inventory only
 - generator: `Abaqus/CAE Learning Edition 2024`
 - units: SI
 - nodal_coordinate_system: global Cartesian
@@ -145,7 +146,7 @@ single-step deck and changes only the named condition.
 - step_name: `Step-1`
 - increment: `1`
 - step_time: `1.0`
-- artifact_status: all four exact paths present; structural precheck observed; FESA comparison not run
+- artifact_status: all four exact paths present; 20-row element-endpoint comparison passed on 2026-08-18
 - stress: N/A for Abaqus reference comparison; mandatory FESA `S11` remains covered by unit/analytical and HDF5 schema tests
 
 The input and CSV numeric reference values are not recalculated, repaired, rounded, clamped or
@@ -193,93 +194,44 @@ For this approved legacy bundle:
 - `README.md`: N/A
 - stress CSV: N/A because beam stress reference comparison is outside the approved V0 scope
 
-The approved design and this contract record model ID, provenance, generator, source commit,
-units, coordinate systems, step/frame identity, logical CSV schema, exact inventory, tolerance
-policy and the stress N/A reason. The optional metadata file's absence and the approved legacy
-README/stress exclusions therefore do not change the record to `needs-reference-artifacts`.
+This contract records the exact required paths, row identities, comparison quantities, tolerance
+policy and the stress N/A reason. Historical schema/provenance fields are retained as dated
+inventory only and are not readiness gates. The optional metadata file's absence and the approved
+legacy README/stress exclusions do not change the record to `needs-reference-artifacts`.
 
-### Future reference bundles
+### Later reference cases and optional metadata
 
-Every later reference model shall use this structure unless its approved requirement explicitly
-marks a quantity N/A:
-
-```text
-reference/
-  <model-id>/
-    model.inp
-    metadata.json                     # optional
-    <model-id>_displacements.csv
-    <model-id>_reactions.csv
-    <model-id>_internalforces.csv
-    <model-id>_stresses.csv
-    README.md
-```
-
-CSV names are canonical `<model-id>_*.csv` names. `README.md` is mandatory for later bundles;
-`metadata.json` is optional. A quantity CSV may be omitted only when the upstream acceptance
-contract explicitly records N/A and gives its verification replacement. Missing required files
-or required Reference Model Contract provenance keep that model at `needs-reference-artifacts`.
-
-## Reference Metadata Contract
-
-This document is the required source of truth for the following metadata. A later bundle may
-optionally duplicate it in `metadata.json` using at least this schema:
-
-```json
-{
-  "feature_id": "linear-static-3d-euler-beam",
-  "model_id": "<model-id>",
-  "artifact_status": "needs-reference-artifacts | ready-for-verification",
-  "input_file": "model.inp",
-  "abaqus_version": "<exact generator/version>",
-  "generation_owner": "<person or approved procedure>",
-  "generation_date": "<YYYY-MM-DD>",
-  "source_commit": "<commit>",
-  "units": "<consistent unit system>",
-  "coordinate_system": "<nodal and element result systems>",
-  "analysis_type": "single linear static",
-  "element_types": ["B33"],
-  "step_name": "Step-1",
-  "increment": 1,
-  "step_time": 1.0,
-  "output_requests": ["U", "RF", "SF"],
-  "reference_csv_schema_version": "<approved schema>",
-  "reference_csv_files": ["<canonical filenames>"],
-  "tolerance_policy": "<approved quantity/component policy>",
-  "limitations": ["<known limitations and explicit N/A quantities>"]
-}
-```
-
-No agent may invent unknown provenance fields or mark a bundle ready merely because filenames
-exist. An absent `metadata.json` is allowed. If the file exists, inventory it read-only and report
-any disagreement with this contract or stored artifacts as an upstream contract/provenance issue.
+Each later feature requirement names its exact input and comparison CSV paths, required row
+identity/components and N/A quantities. Canonical filenames, `README.md`, `metadata.json`,
+generator/version, provenance, duplicated unit/schema fields and a portfolio-wide directory layout
+are not readiness gates unless that feature explicitly makes one part of its numerical comparison
+contract. No agent may invent unknown provenance. If optional metadata exists, inspect it read-only
+and report disagreement without rewriting the reference artifacts.
 
 ## Abaqus Reference CSV Requirements
 
 Header comparison trims whitespace around each comma-separated field but does not rename fields.
 For every file, `Frame` must normalize exactly from
 `Increment 1: Step Time = 1.000` to `(Step-1, frame 0)`, `Part Instance Name` must resolve to the
-preserved instance identity, `Node Label` must be a unique source-node station, and all projected
+preserved instance identity, every declared source identity must be unique, and all projected
 numeric values must be finite.
 
 | exact legacy path | expected trimmed header | unique row key | observed inventory |
 | --- | --- | --- | --- |
 | `reference/cantilever beam/cantilever beam displacements.csv` | `Frame, Part Instance Name, Node Label, U-U1, U-U2, U-U3, UR-UR1, UR-UR2, UR-UR3` | `(Frame, Part Instance Name, Node Label)` | 11 rows; header/key/finite/arity checks observed |
 | `reference/cantilever beam/cantilever beam reactions.csv` | `Frame, Part Instance Name, Node Label, RF-RF1, RF-RF2, RF-RF3, RM-RM1, RM-RM2, RM-RM3` | `(Frame, Part Instance Name, Node Label)` | 11 rows; header/key/finite/arity checks observed |
-| `reference/cantilever beam/cantilever beam elemental forces.csv` | `Frame, Part Instance Name, Node Label, SF-SF1, SM-SM1, SM-SM2, SM-SM3` | `(Frame, Part Instance Name, Node Label)` | 11 rows; header/key/finite/arity checks observed |
+| `reference/cantilever beam/cantilever beam elemental forces.csv` | `Frame, Part Instance Name, Element Label, Node Label, SF-SF1, SM-SM1, SM-SM2, SM-SM3` | `(Frame, Part Instance Name, Element Label, Node Label)` | 20 rows; 10 B33 elements x 2 connectivity-matched endpoints; header/key/finite/arity checks observed |
 
 The wide-row key becomes unique canonical component rows after adding `quantity` and `component`.
 Missing, extra, duplicate, nonfinite, header/schema or identity mismatch stops comparison as
 `needs-reference-artifacts` or `schema-mismatch`. No bad or near-zero row may be silently dropped.
 
-For the elemental-force CSV, source node station is not an element-end identity. A boundary
-station uses its only incident endpoint. An interior station may collapse exactly two endpoints
-only when chain connectivity, section and local axes are consistent and the station has no
-concentrated force/moment. The two positive-local-x section-cut values must first agree within the
-approved component tolerance. If they agree, choose the endpoint with smaller stable internal
-element ID; never average. Reversed orientation, branch, section jump, local-axis discontinuity or
-loaded interior station requires an element-aware future schema and is a `schema-mismatch` under
-this legacy schema.
+For the elemental-force CSV, `(Part Instance Name, Element Label, Node Label)` is the source
+element-endpoint identity. Each B33 element must have exactly two rows whose node labels match its
+ordered input connectivity. Every row maps directly to the corresponding HDF5
+`[element,endpoint,N/T/My/Mz]` row. Missing, extra, duplicate or connectivity-mismatched endpoint
+identity fails as `schema-mismatch` before tolerance. The comparator does not collapse endpoints to
+a node station, choose a representative or average values.
 
 ## Coverage Matrix
 
@@ -288,20 +240,20 @@ the canonical V0 step identity.
 
 | verification quantity | requirement ids | model_id | FESA HDF5 dataset | legacy CSV and components | row identity/location | tolerance | verification method | status |
 | --- | --- | --- | --- | --- | --- | --- | --- | --- |
-| nodal displacement/rotation | 003, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/displacement` | `reference/cantilever beam/cantilever beam displacements.csv`: `U-U1/U-U2/U-U3 -> UX/UY/UZ`, `UR-UR1/UR-UR2/UR-UR3 -> URX/URY/URZ` | preserved instance + source node; global nodal | `1e-9 + 1e-6*reference_scale` per displacement/rotation component | HDF5-to-read-only CSV after artifact/row-set precheck | ready for implementation planning; comparison not run |
-| nodal reaction force | 007, 027, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/reaction` | `reference/cantilever beam/cantilever beam reactions.csv`: `RF-RF1/RF-RF2/RF-RF3 -> RF1/RF2/RF3` | preserved instance + source node; global nodal | `1e-3 + 1e-6*reference_scale` per force component | HDF5-to-read-only CSV plus global equilibrium | ready for implementation planning; comparison not run |
-| nodal reaction moment | 007, 027, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/reaction` | `reference/cantilever beam/cantilever beam reactions.csv`: `RM-RM1/RM-RM2/RM-RM3 -> RM1/RM2/RM3` | preserved instance + source node; global nodal | `1e-3 + 1e-6*reference_scale` per moment component | HDF5-to-read-only CSV plus moment equilibrium | ready for implementation planning; comparison not run |
-| section axial force | 031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/element/section_resultant` | `reference/cantilever beam/cantilever beam elemental forces.csv`: `SF-SF1 -> N` | positive-local-x endpoint projected to eligible source node station | `1e-3 + 1e-6*reference_scale` for `N` | endpoint consistency, deterministic station selection, HDF5-to-CSV | ready for implementation planning; comparison not run |
-| section moments/torsion | 031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/element/section_resultant` | `reference/cantilever beam/cantilever beam elemental forces.csv`: `SM-SM1 -> My`, `SM-SM2 -> Mz`, `SM-SM3 -> T` | positive-local-x endpoint projected to eligible source node station; beam local | `1e-3 + 1e-6*reference_scale` separately for `My`, `Mz`, `T` | endpoint consistency, deterministic station selection, HDF5-to-CSV | ready for implementation planning; comparison not run |
+| nodal displacement/rotation | 003, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/displacement` | `reference/cantilever beam/cantilever beam displacements.csv`: `U-U1/U-U2/U-U3 -> UX/UY/UZ`, `UR-UR1/UR-UR2/UR-UR3 -> URX/URY/URZ` | preserved instance + source node; global nodal | common translation/rotation family policy | HDF5-to-read-only CSV after artifact/row-set precheck | passed 2026-08-18 |
+| nodal reaction force | 007, 027, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/reaction` | `reference/cantilever beam/cantilever beam reactions.csv`: `RF-RF1/RF-RF2/RF-RF3 -> RF1/RF2/RF3` | preserved instance + source node; global nodal | common reaction-force family policy | HDF5-to-read-only CSV plus global equilibrium | passed 2026-08-18 |
+| nodal reaction moment | 007, 027, 029-031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/nodal/reaction` | `reference/cantilever beam/cantilever beam reactions.csv`: `RM-RM1/RM-RM2/RM-RM3 -> RM1/RM2/RM3` | preserved instance + source node; global nodal | common reaction-moment family policy | HDF5-to-read-only CSV plus moment equilibrium | passed 2026-08-18 |
+| section axial force | 031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/element/section_resultant` | `reference/cantilever beam/cantilever beam elemental forces.csv`: `SF-SF1 -> N` | direct instance + source element + endpoint node; beam local | common section-force family policy | direct HDF5-to-endpoint CSV comparison | passed 2026-08-18 |
+| section moments/torsion | 031, 036-042 | `cantilever-beam-b33` | `/steps/Step-1/frames/0/element/section_resultant` | `reference/cantilever beam/cantilever beam elemental forces.csv`: `SM-SM1 -> My`, `SM-SM2 -> Mz`, `SM-SM3 -> T` | direct instance + source element + endpoint node; beam local | common section-moment family policy | direct HDF5-to-endpoint CSV comparison | passed 2026-08-18 |
 | equilibrium end action | 031, 035, 043 | analytical models and physics portfolio | `/steps/Step-1/frames/0/element/end_force_local` | Abaqus CSV N/A for direct outward-action comparison | element endpoint `xi=-1,+1`; local outward action `[FX,FY,FZ,MX,MY,MZ]` | analytical normalized `1e-12`; residual `1e-10` | unit/analytical end-sign tests and later physics sanity | planned |
 | generalized strain/resultant | 029, 031, 035 | code and analytical models | `/steps/Step-1/frames/0/element/generalized_strain` and `/steps/Step-1/frames/0/element/generalized_resultant` | Abaqus CSV N/A | two Gauss points; beam local | matrix/formulation normalized `1e-12`, analytical relative `1e-9` | formulation/unit/HDF5 schema tests | planned |
 | axial stress | 029, 032, 035 | axial/local-z analytical models | `/steps/Step-1/frames/0/element/stress_s11` | stress CSV N/A; Abaqus beam stress reference comparison N/A | element, Gauss point, input section point or `fesa-default` centroid | analytical relative `1e-9`; exact unit/row schema | unit/analytical recovery and HDF5 schema tests | planned; reference N/A |
 
-For every matched reference group,
-`reference_scale=max(abs(Abaqus reference rows))` for the same model, step/frame, quantity and
-component, and `row_tolerance=absolute_floor+1e-6*reference_scale`. Abaqus values alone set the
-scale. A zero scale uses only the floor. Every row decision and max absolute, component-scale
-normalized, RMS, norm and worst-row/component metrics must be reported.
+For every matched family, `S=max(abs(Abaqus reference rows))` uses only read-only reference values.
+Rows with `abs(reference)<=0.01*S` use `error<=0.01*S`; other rows use relative error `<=0.05`;
+the family also requires `RMS(error)/S<=0.01`. A zero-scale family requires exact-zero FESA values.
+Every row branch/decision and family scale, near-zero band/count, max absolute error,
+scale-relative RMS, worst row/component and zero-scale diagnostic must be reported.
 
 ### Complete must-requirement coverage
 
@@ -346,19 +298,20 @@ it does not waive the requirement.
 | `FESA-REQ-LS3DEB-033` | `smoke-b33-cli`, diagnostic negative cases | CLI exit-code/field/order integration tests; CSV N/A |
 | `FESA-REQ-LS3DEB-034` | every implementation model/test | per-step RED/GREEN/VERIFY and full MSVC/CTest evidence; model CSV N/A |
 | `FESA-REQ-LS3DEB-035` | `NR-T01` through `NR-T11` and analytical inventory | exact numerical criteria in model records |
-| `FESA-REQ-LS3DEB-036` | `cantilever-beam-b33` and comparison-policy unit fixtures | Abaqus-only component scale and exact formula tests |
-| `FESA-REQ-LS3DEB-037` | same comparison fixtures, including zero-scale groups | exact SI floors and zero-scale tests |
+| `FESA-REQ-LS3DEB-036` | `cantilever-beam-b33` and comparison-policy unit fixtures | exact family membership and Abaqus-only scale tests |
+| `FESA-REQ-LS3DEB-037` | same comparison fixtures, including boundary and zero-scale families | near-zero, relative-row, scale-relative RMS and zero-scale tests |
 | `FESA-REQ-LS3DEB-038` | malformed reference cases in `neg-b33-input-contract` | pre-tolerance fail-fast and no-clamp/no-drop tests |
-| `FESA-REQ-LS3DEB-039` | verification-report schema fixture | every row decision and aggregate/worst metrics test |
+| `FESA-REQ-LS3DEB-039` | verification-report schema fixture | every row branch/decision and family aggregate/worst metrics test |
 | `FESA-REQ-LS3DEB-040` | exact legacy Artifact Bundle Contract | inventory plus `git diff --exit-code -- reference/` process check |
 | `FESA-REQ-LS3DEB-041` | approved artifact precheck | four files, B33, exact headers, unique keys, finite values |
-| `FESA-REQ-LS3DEB-042` | three comparison quantities and station normalization | exact component mapping, endpoint consistency and no-average tests |
+| `FESA-REQ-LS3DEB-042` | three comparison quantities and direct endpoint projection | exact component mapping plus missing/extra/duplicate/connectivity-mismatch endpoint tests |
 | `FESA-REQ-LS3DEB-043` | gate audit; later physics portfolio | reference execution N/A at this step; enforce build/test -> comparison -> physics -> release order |
 | `FESA-REQ-LS3DEB-044` | process/Git diff audit and limitations review | no reference execution/mutation and no out-of-scope support claim |
 
 ## Artifact Acceptance Checklist
 
-Read-only inventory inspection on `2026-08-09` established the following pre-implementation facts:
+The following is historical 2026-08-09 pre-implementation inventory evidence. Its 11-row
+elemental-force observation is superseded by the 2026-08-18 regenerated 20-row endpoint file:
 
 - all four exact legacy paths exist;
 - the input declares `TYPE=B33`;
@@ -370,6 +323,11 @@ Read-only inventory inspection on `2026-08-09` established the following pre-imp
 - absent `metadata.json` is allowed by project-wide policy; legacy `README.md` and stress CSV are accepted N/A exceptions;
 - no reference value was recalculated and no comparison was performed.
 
+Read-only inspection on `2026-08-18` confirms that the elemental-force CSV has 20 rows,
+exactly two connectivity-matched endpoint rows for each of the 10 B33 elements. Displacement and
+reaction CSVs remain 11-row source-node tables. The fresh comparison passed on 2026-08-18;
+current metrics are recorded in `reference-comparison.md`.
+
 Before an actual comparison, tooling must repeat all artifact checks, verify the exact Frame and
 instance identities, compare the complete projected row sets, and stop on any failure. The Step AC
 must also show no working-tree diff under `reference/`. Passing this checklist is not a reference
@@ -383,8 +341,8 @@ comparison pass.
   model.
 - `NR-O01` (official oneMKL PARDISO contract) and `NR-O02` (deterministic duplicate-reduction
   algorithm) remain implementation-planning inputs, not reference artifact defects.
-- Future reversed, branched, loaded-interior or section-jump reference models require an
-  element-aware canonical CSV schema; the legacy node-station schema must not be generalized.
+- Future reversed, branched, loaded-interior or section-jump reference models use the same
+  element-endpoint identity and require their own approved reference coverage.
 
 ### Implementation Planning Agent
 
@@ -403,8 +361,8 @@ artifacts for code/analytical fixtures.
 
 ### Reference Verification Agent
 
-Run `ARTIFACT CHECK -> HDF5 ROW PROJECTION -> EXACT ROW-SET CHECK -> ENDPOINT CONSISTENCY ->
-COMPONENT-SCALE COMPARE -> REPORT`. Use only the exact legacy files and the HDF5 paths/component
+Run `ARTIFACT CHECK -> HDF5 ROW PROJECTION -> EXACT ROW-SET CHECK -> DIRECT ENDPOINT MATCH ->
+COMMON FAMILY-SCALE COMPARE -> REPORT`. Use only the exact legacy files and the HDF5 paths/component
 mappings in the Coverage Matrix. Do not synthesize `SF2/SF3` or stress rows, clamp values, omit
 rows, average interior endpoints, or change the approved tolerance.
 

docs/linear-static-3d-euler-beam/release.md

@@ -4,26 +4,32 @@
 
 - feature_id: `linear-static-3d-euler-beam`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md` (`status: approved`; approved 2026-08-08, amended 2026-08-09)
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
-- source_research: `docs/research/linear-static-3d-euler-beam-research.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-3d-euler-beam-review.md`
-- source_io_definition: `docs/io-definitions/linear-static-3d-euler-beam-io.md`
-- source_reference_model: `docs/reference-models/linear-static-3d-euler-beam-reference-models.md`
-- source_implementation_plan: `docs/implementation-plans/linear-static-3d-euler-beam.md`
-- source_implementation_report: `docs/implementation-plans/linear-static-3d-euler-beam-implementation-report.md`
-- source_build_test_report: `docs/build-test-reports/linear-static-3d-euler-beam.md`
-- source_reference_verification_report: `docs/reference-verifications/linear-static-3d-euler-beam-reference-verification.md`
-- source_physics_evaluation_report: `docs/physics-evaluations/linear-static-3d-euler-beam-physics-evaluation.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
+- source_research: `docs/linear-static-3d-euler-beam/research.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- source_numerical_review: `docs/linear-static-3d-euler-beam/numerical-review.md`
+- source_io_definition: `docs/linear-static-3d-euler-beam/io.md`
+- source_reference_model: `docs/linear-static-3d-euler-beam/reference-model.md`
+- source_implementation_plan: `docs/linear-static-3d-euler-beam/implementation-plan.md`
+- source_implementation_report: `docs/linear-static-3d-euler-beam/implementation-report.md`
+- source_build_test_report: `docs/linear-static-3d-euler-beam/build-test.md`
+- source_reference_verification_report: `docs/linear-static-3d-euler-beam/reference-comparison.md`
+- source_physics_evaluation_report: `docs/linear-static-3d-euler-beam/physics-evaluation.md`
 - audited_head: `822b06be3d2128d5dfdc5e394078abbb9dcd5a50`
 - reference_model_id: `cantilever-beam-b33`
 - reference_schema: `abaqus-cae-report-csv-v0`
 - reference_baseline: `reference/cantilever beam/` at source commit `2b34d0b`
-- status: `ready-for-release`
+- historical_status: `ready-for-release`
+- current_status: `superseded-pending-revalidation`
+- superseded_on: `2026-08-18`
 - owner_agent: `release-agent`
 - date: `2026-08-10`
 - final_review_source_head: `b7a1258ce0f36a85b888e23470cf9d936a7595cd`
 
+The readiness evidence below is historical. The 2026-08-18 build/reference revalidation passed,
+but fresh physics and release audits are still required before a current `ready-for-release`
+verdict may be issued.
+
 This is an internal feature-readiness verdict. It authorizes no publish, deploy, package, tag,
 push, external release, or reference-artifact operation.
 

docs/linear-static-3d-euler-beam/requirements.md

@@ -9,8 +9,8 @@
 - date: `2026-08-09`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
 - approval_basis: design `status: approved`, user approval on `2026-08-08`, and amendment on `2026-08-09`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- reference_baseline: `reference/cantilever beam/` at source commit `2b34d0b`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- reference_baseline: exact read-only artifacts under `reference/cantilever beam/`
 
 ## Purpose
 
@@ -113,7 +113,7 @@ compatibility, 새로운 수학 계약 또는 새로운 구현 정책을 추가
 - nodal_displacement: required, global six components, HDF5-to-Abaqus CSV comparison
 - reaction: required, global six components plus global force/moment equilibrium
 - equilibrium_end_action: required, local endpoint six components; unit/analytical and physics tests
-- section_resultant: required, endpoint `[N,T,My,Mz]`, node-station-normalized reference comparison
+- section_resultant: required, endpoint `[N,T,My,Mz]`, element-endpoint reference comparison
 - generalized_strain_and_resultant: required at two Gauss points; formulation and schema tests
 - stress: axial `S11` required; Abaqus reference comparison N/A
 - residual: required, free-DOF and normalized global equilibrium checks
@@ -122,21 +122,22 @@ compatibility, 새로운 수학 계약 또는 새로운 구현 정책을 추가
 
 ## Tolerance Policy
 
-- **FESA-REQ-LS3DEB-036** — The approved B33 reference comparison shall group rows by the same model, step/frame, quantity, and component, compute `reference_scale` only from read-only Abaqus values, and apply `absolute_floor + 1e-6 * reference_scale` to every matched row.
-- **FESA-REQ-LS3DEB-037** — For the approved SI bundle, displacement and rotation shall use `absolute_floor=1e-9`, force and moment shall use `absolute_floor=1e-3`, and a zero component scale shall use the applicable absolute floor alone.
+- **FESA-REQ-LS3DEB-036** — The approved B33 reference comparison shall group rows into translation, rotation, reaction-force, reaction-moment, section-force and section-moment families and compute each family scale `S=max(abs(reference))` only from read-only Abaqus values.
+- **FESA-REQ-LS3DEB-037** — A matched row with `abs(reference)<=0.01*S` shall pass when `abs(fesa-reference)<=0.01*S`; every other row shall pass when its relative error is at most `0.05`; each family shall also satisfy `RMS(error)/S<=0.01`. No independent absolute-error gate is used, and a zero-scale family passes only when every FESA value is exactly zero.
 - **FESA-REQ-LS3DEB-038** — Reference values shall not be zero-clamped and rows shall not be dropped; missing, extra, duplicate, nonfinite, schema-mismatched, or identity-mismatched rows shall fail before tolerance evaluation.
-- **FESA-REQ-LS3DEB-039** — The verification report shall record every row decision and maximum absolute error, component-scale normalized error, RMS error, norm error, and worst row/component for each compared quantity.
+- **FESA-REQ-LS3DEB-039** — The verification report shall record every row decision and comparison branch, family identity/components, reference scale, near-zero band/count, maximum absolute error, scale-relative RMS, worst row/component and zero-scale diagnostic for each compared family.
 
 ## Reference Artifact Requirements
 
-The approved logical model is `cantilever-beam-b33`, schema is
-`abaqus-cae-report-csv-v0`, source commit is `2b34d0b`, generator is
+The approved logical model is `cantilever-beam-b33`. The input header records
 `Abaqus/CAE Learning Edition 2024`, and the coordinate/output contract is global Cartesian
 nodal output plus beam-local section-force output at `Step-1`, increment 1, step time 1.0.
+The historical `abaqus-cae-report-csv-v0` and source commit `2b34d0b` inventory is not a
+readiness requirement for the regenerated elemental-force CSV.
 
 - **FESA-REQ-LS3DEB-040** — The V0 reference baseline shall use the exact read-only files `reference/cantilever beam/cantilever beam.inp`, `reference/cantilever beam/cantilever beam displacements.csv`, `reference/cantilever beam/cantilever beam reactions.csv`, and `reference/cantilever beam/cantilever beam elemental forces.csv` without rename, rewrite, correction, or restoration; absent `metadata.json` is allowed by project-wide policy and `README.md` is N/A for this approved legacy bundle.
 - **FESA-REQ-LS3DEB-041** — Before comparison, artifact validation shall confirm all four files, `TYPE=B33`, expected CAE report headers, unique row keys, and finite values; failure shall be classified as `needs-reference-artifacts` or `schema-mismatch` and comparison shall not start.
-- **FESA-REQ-LS3DEB-042** — Reference verification shall compare displacement by source-node identity (`U1/U2/U3/UR1/UR2/UR3`), reaction by source-node identity (`RF1/RF2/RF3/RM1/RM2/RM3`), and node-station-normalized section resultant by `SF1->N`, `SM1->My`, `SM2->Mz`, `SM3->T`; adjacent interior endpoints shall first agree within approved tolerance and shall not be averaged to hide a mismatch.
+- **FESA-REQ-LS3DEB-042** — Reference verification shall compare displacement by source-node identity (`U1/U2/U3/UR1/UR2/UR3`), reaction by source-node identity (`RF1/RF2/RF3/RM1/RM2/RM3`), and section resultants by direct `(instance, element label, endpoint node label, component)` identity using `SF1->N`, `SM1->My`, `SM2->Mz`, `SM3->T`; missing, extra, duplicate or connectivity-mismatched endpoint rows shall fail before tolerance and shall not be collapsed or averaged.
 - **FESA-REQ-LS3DEB-043** — Reference comparison shall run only after build/test passes; physics sanity shall run only after reference comparison passes and shall check global force/moment equilibrium, reaction sign, displacement direction, symmetry, element section-force consistency, and normalized residual; release readiness shall require all prior gate evidence and known limitations.
 - **FESA-REQ-LS3DEB-044** — FESA agents and Harness shall not execute Abaqus, Nastran, or another reference solver and shall not create, modify, rename, or restore reference artifacts; release documentation shall not claim support for any Out Of Scope behavior.
 
@@ -180,13 +181,13 @@ nodal output plus beam-local section-force output at `Step-1`, increment 1, step
 | FESA-REQ-LS3DEB-033 | The CLI shall support `fesa.exe <model.inp> --output <results.h5>`, default output to the current directory's `results.h5`, use exit codes `0=success`, `2=usage`, `3=input`, `4=model`, `5=solver`, `6=HDF5`, and emit `severity`, `code`, `file`, `line`, `keyword`, `entity_identity`, and `message` diagnostics to stderr in deterministic order. | output | Stabilize automation and failure classification. | Approved design §10 | must | CLI integration and diagnostic ordering tests | Default/explicit output works and every failure class returns its exact code and complete ordered fields. | Exact codes, fields, and order | io-definition-agent; implementation-planning-agent | approved |
 | FESA-REQ-LS3DEB-034 | Every production C++ behavior shall be developed in one step as GoogleTest `RED -> GREEN -> VERIFY`, with a related C++ test file, focused CTest evidence, full MSVC x64 Debug build/CTest evidence, at least one discovered test, and no new warning under the FESA target's `/W4 /WX` policy. | verification | Enforce project TDD and warning policy. | Approved design §§11.1, 11.4; ADR-012 | must | Implementation report; build/CTest logs | The related test fails first, then focused/full tests pass, discovery finds tests, and FESA emits no warning. | Zero test failures and new warnings | implementation-planning-agent; implementation-agent; build-test-executor-agent | approved |
 | FESA-REQ-LS3DEB-035 | Numerical tests shall satisfy normalized `1e-12` for stiffness symmetry and two-point-Gauss/closed-form agreement, normalized `1e-10` for rigid-mode and linear-system residual, and relative `1e-9` for analytical solutions, while checking six rigid modes, rank 6, positive deformation energy, transformation orthogonality/energy invariance, prescribed-displacement recovery, and axial/torsion/two-plane bending benchmarks. | verification | Detect sign, integration, rank, and transform defects. | Approved design §§11.2, 11.3; formulation §18 | must | Unit, analytical, and orchestration tests | Every listed invariant and analytical case passes at its stated threshold. | `1e-12` matrix; `1e-10` residual; `1e-9` analytical | formulation-agent; numerical-review-agent; implementation-planning-agent | approved |
-| FESA-REQ-LS3DEB-036 | The approved B33 reference comparison shall group rows by the same model, step/frame, quantity, and component, compute `reference_scale` only from read-only Abaqus values, and apply `absolute_floor + 1e-6 * reference_scale` to every matched row. | tolerance | Give zero and nonzero rows one deterministic rule. | Approved design §11.3; ADR-014 | must | Comparison unit/integration test; report review | Every group uses the Abaqus-only maximum absolute scale and every matched row uses the exact formula. | Relative coefficient `1e-6` | reference-model-agent; reference-verification-agent | approved |
-| FESA-REQ-LS3DEB-037 | For the approved SI bundle, displacement and rotation shall use `absolute_floor=1e-9`, force and moment shall use `absolute_floor=1e-3`, and a zero component scale shall use the applicable absolute floor alone. | tolerance | Preserve dimensional meaning near zero. | Approved design §11.3; ADR-014 | must | Comparison tests with zero/near-zero rows | Each quantity uses its exact SI floor and zero-scale groups use no relative contribution. | SI `1e-9` displacement/rotation; `1e-3` force/moment | reference-model-agent; reference-verification-agent | approved |
+| FESA-REQ-LS3DEB-036 | The approved B33 reference comparison shall group rows into translation, rotation, reaction-force, reaction-moment, section-force and section-moment families and compute `S=max(abs(reference))` from read-only Abaqus values. | tolerance | Give zero-like components a dimensionally compatible reference scale. | Common tolerance design; ADR-022 | must | Comparison unit/integration test; report review | Every family uses the exact declared components and Abaqus-only maximum scale. | Family scale | reference-model-agent; reference-verification-agent | approved |
+| FESA-REQ-LS3DEB-037 | Near-zero rows shall use the `0.01*S` fallback, other rows relative error `0.05`, and every family scale-relative RMS `0.01`; zero-scale families require exact-zero FESA values and no independent absolute gate is used. | tolerance | Stabilize zero-like rows while retaining row and aggregate checks. | Common tolerance design; ADR-022 | must | Boundary, zero-scale and RMS comparison tests | Every row and family uses the exact common constants and branch rules without NaN/Inf. | `0.01`, `0.05`, `0.01` | reference-model-agent; reference-verification-agent | approved |
 | FESA-REQ-LS3DEB-038 | Reference values shall not be zero-clamped and rows shall not be dropped; missing, extra, duplicate, nonfinite, schema-mismatched, or identity-mismatched rows shall fail before tolerance evaluation. | tolerance | Prevent false passes through omission or clamping. | Approved design §§11.3, 12; ADR-014 | must | Negative comparison tests | Every listed invalid case fails before numeric comparison and zero values remain unchanged. | No ignored invalid rows | reference-verification-agent | approved |
-| FESA-REQ-LS3DEB-039 | The verification report shall record every row decision and maximum absolute error, component-scale normalized error, RMS error, norm error, and worst row/component for each compared quantity. | tolerance | Make the pass/fail decision auditable. | Approved design §11.3; ADR-014 | must | Verification report schema/review | Per-row decisions and all required aggregate/worst metrics are present for every quantity. | Report completeness | reference-verification-agent; release-agent | approved |
+| FESA-REQ-LS3DEB-039 | The verification report shall record every row branch/decision, family identity/components, scale, near-zero band/count, maximum absolute error, scale-relative RMS, worst row/component and zero-scale diagnostic. | tolerance | Make the pass/fail decision auditable. | Common tolerance design; ADR-022 | must | Verification report schema/review | Per-row decisions and all required family metrics are present without nonfinite report values. | Report completeness | reference-verification-agent; release-agent | approved |
 | FESA-REQ-LS3DEB-040 | The V0 reference baseline shall use the exact read-only files `reference/cantilever beam/cantilever beam.inp`, `reference/cantilever beam/cantilever beam displacements.csv`, `reference/cantilever beam/cantilever beam reactions.csv`, and `reference/cantilever beam/cantilever beam elemental forces.csv` without rename, rewrite, correction, or restoration; absent `metadata.json` is allowed by project-wide policy and `README.md` is N/A for this approved legacy bundle. | reference | Protect the approved correctness baseline. | Approved design §12; ADR-010, ADR-014 | must | Artifact inventory; Git diff review | Exact filenames exist and no reference file is added, removed, renamed, or content-modified. | Exact path/content identity | reference-model-agent; reference-verification-agent; release-agent | approved |
 | FESA-REQ-LS3DEB-041 | Before comparison, artifact validation shall confirm all four files, `TYPE=B33`, expected CAE report headers, unique row keys, and finite values; failure shall be classified as `needs-reference-artifacts` or `schema-mismatch` and comparison shall not start. | reference | Detect stale B31 or malformed evidence. | Approved design §§6.3, 12 | must | Artifact-check integration test | All checks pass before comparison and every failure uses an approved classification. | Exact inventory and schema | reference-model-agent; reference-verification-agent | approved |
-| FESA-REQ-LS3DEB-042 | Reference verification shall compare displacement by source-node identity (`U1/U2/U3/UR1/UR2/UR3`), reaction by source-node identity (`RF1/RF2/RF3/RM1/RM2/RM3`), and node-station-normalized section resultant by `SF1->N`, `SM1->My`, `SM2->Mz`, `SM3->T`; adjacent interior endpoints shall first agree within approved tolerance and shall not be averaged to hide a mismatch. | reference | Compare equivalent quantities despite legacy station rows. | Approved design §§8.2, 12 | must | Reference comparison test/report | All components match by source identity and interior endpoints pass before deterministic representative selection. | Requirements 036 and 037 policy | io-definition-agent; reference-model-agent; reference-verification-agent | approved |
+| FESA-REQ-LS3DEB-042 | Reference verification shall compare displacement and reaction by source-node identity and section resultants by direct `(instance, element label, endpoint node label, component)` identity using `SF1->N`, `SM1->My`, `SM2->Mz`, `SM3->T`; endpoint rows shall not be collapsed or averaged. | reference | Preserve the source identity supplied by the element-endpoint CSV and HDF5. | Common tolerance design; ADR-022 | must | Reference comparison test/report | All 20 B33 endpoint rows map one-to-one and malformed endpoint identity fails before tolerance. | Requirements 036 and 037 policy | io-definition-agent; reference-model-agent; reference-verification-agent | approved |
 | FESA-REQ-LS3DEB-043 | Reference comparison shall run only after build/test passes; physics sanity shall run only after reference comparison passes and shall check global force/moment equilibrium, reaction sign, displacement direction, symmetry, element section-force consistency, and normalized residual; release readiness shall require all prior gate evidence and known limitations. | governance | Keep numerical similarity distinct from physical/release approval. | Approved design §§11, 12, 13 | must | Gate evidence audit | Each downstream report cites the preceding pass and physics evidence covers all six checks before release review. | Relevant upstream tolerances | coordinator-agent; physics-evaluation-agent; release-agent | approved |
 | FESA-REQ-LS3DEB-044 | FESA agents and Harness shall not execute Abaqus, Nastran, or another reference solver and shall not create, modify, rename, or restore reference artifacts; release documentation shall not claim support for any Out Of Scope behavior. | governance | Prevent baseline contamination and scope inflation. | Approved design §§2.2, 12; ADR-010 | must | Process audit; Git diff; release documentation review | No reference execution or artifact mutation occurs and every exclusion is recorded without a support claim. | N/A | coordinator-agent; reference-model-agent; release-agent | approved |
 

docs/linear-static-3d-euler-beam/research.md

@@ -3,9 +3,9 @@
 ## Metadata
 
 - feature_id: `linear-static-3d-euler-beam`
-- source_requirement: `docs/requirements/linear-static-3d-euler-beam.md`
+- source_requirement: `docs/linear-static-3d-euler-beam/requirements.md`
 - approved_design: `docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
 - status: `ready-for-formulation`
 - owner_agent: `research-agent`
 - date: `2026-08-09`
@@ -32,7 +32,7 @@ This brief supplies evidence for formulation alignment and later verification pl
 | S4 | Isoparametric mapping and quadrature synthesis | FEM wiki, with Bathe, Logan, Abaqus, and local source provenance | [[Isoparametric Finite Elements]] | *Finite Element Procedures*; Logan textbook; *Abaqus Theory Manual*; *Abaqus Analysis User's Guide, Volume IV* | Tier 2 synthesis; Tier 1 where Abaqus-specific | natural coordinates, Jacobian mapping, quadrature choice, under-integration risk |
 | S5 | FEM program verification workflow | FEM wiki, with textbook and official-manual provenance | [[Finite Element Program Implementation]]; [[Finite Element Modeling and Convergence Checks]] | *Finite Element Procedures*; Logan textbook; Abaqus user guides; production solver manuals | Tier 2 synthesis; Tier 1 for cited official-manual behavior | element-local/global separation, sparse assembly context, constraint/solve/recovery workflow, benchmark and equilibrium checks |
 | P1 | Approved FESA feature contract | FESA project | N/A | approved requirements and design named in Metadata | Project contract, not external evidence | exact V0 scope, fixed tolerance policy, read-only reference inventory, required output and orchestration |
-| P2 | Existing candidate formulation | FESA project | cites [[Beam and Frame Finite Elements]], [[Isoparametric Finite Elements]], and S5 pages | `docs/formulations/3d-isoparametric-euler-beam-formulation.md` | Candidate derivation, not an approved source | equations and sign choices that downstream Formulation and Numerical Review agents must check |
+| P2 | Existing candidate formulation | FESA project | cites [[Beam and Frame Finite Elements]], [[Isoparametric Finite Elements]], and S5 pages | `docs/linear-static-3d-euler-beam/formulation.md` | Candidate derivation, not an approved source | equations and sign choices that downstream Formulation and Numerical Review agents must check |
 
 The wiki source records identify S1 as high-confidence manual provenance, S2 as a high-confidence textbook source, and S3 as a current textbook source. P1 and P2 are intentionally not assigned an external reliability tier.
 

docs/linear-static-mitc4-shell/build-test.md

@@ -0,0 +1,193 @@
+# Linear Static MITC4 Shell Build/Test Report
+
+## Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- source_commit: `820ba30c717b3d0e113775608e20dfd5fbc05d53`
+- source_implementation_report: `N/A`; Harness completion evidence is recorded in
+  `phases/linear-static-mitc4-shell/index.json`
+- source_implementation_plan:
+  `docs/linear-static-mitc4-shell/implementation-plan.md`
+- status: `pass-for-reference-verification`
+- owner_agent: `build-test-executor-agent`
+- date: `2026-08-13`
+
+## Verdict
+
+The formal build/test gate passes. A fresh Visual Studio 18 2026/MSVC x64 Debug
+configuration, clean full rebuild, three implementation-plan-focused CTest runs,
+nonzero JSON discovery, full CTest regression, and the required Harness Python
+self-test all completed successfully.
+
+- clean full rebuild: exit `0`; compiler/linker warning scan `0`; error scan `0`
+- focused kernel/model/assembly/recovery regex: `87/87` passed
+- focused linear-static lifecycle regex: `10/10` passed
+- focused MITC4 reference-test regex: `8/8` passed
+- CTest discovery: `144` tests, including `32` names containing `Mitc4`
+- full CTest after clean rebuild: `144/144` passed
+- Harness Python self-test: `7/7` passed
+- failed tests: `0`
+- required S4 reference hash mismatches or Git changes: `0`
+
+This verdict authorizes handoff to Reference Verification only. It does not approve
+reference tolerance results, physics sanity, or release readiness.
+
+## Execution Environment
+
+- os: `Microsoft Windows 11 Home 10.0.26200 build 26200`
+- branch: `feat-linear-static-mitc4-shell`
+- source_head: `820ba30c717b3d0e113775608e20dfd5fbc05d53`
+- generator: `Visual Studio 18 2026`
+- MSBuild: `18.8.2+ce25c0108`
+- compiler: `MSVC 19.51.36252.0`
+- compiler_architecture: `x64`
+- Windows SDK: `10.0.26100.0`
+- language/config: `C++17 / Debug`
+- CMake: `4.4.0`
+- build_dir: `.harness/build`
+- harness_config: absent
+- project_selection: auto CMake from root `CMakeLists.txt`
+- command_discovery_path: default CMake/MSVC x64 Debug with implementation-plan
+  generator and dependency paths
+- resolved dependencies: GoogleTest `C:/git/googletest`; oneMKL `2026.1.0`;
+  oneTBB package path `2023.1`; HDF5 package path `2.1.1`
+
+The generated compiler metadata at
+`.harness/build/CMakeFiles/4.4.0/CMakeCXXCompiler.cmake` independently confirms
+`CMAKE_CXX_COMPILER_ID=MSVC` and version `19.51.36252.0`.
+
+## Command Log Summary
+
+All durations below are wall-clock durations measured for the command. The initial
+green chain was retained as corroborating evidence; the later clean-rebuild chain is
+the canonical acceptance evidence.
+
+| order | command | exit_code | duration | relevant stdout/stderr tail |
+| ---: | --- | ---: | ---: | --- |
+| 1 | `git status --short --branch; git rev-parse HEAD` | 0 | `0.8 s` | Clean `feat-linear-static-mitc4-shell`; HEAD matched `820ba30...`. |
+| 2 | Required dependency `Test-Path` precheck from the implementation plan | 0 | `0.215 s` | All four GoogleTest/MKL/TBB/HDF5 paths returned `True`. |
+| 3 | `cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 "-DFESA_GTEST_SOURCE_DIR=C:/git/googletest" "-DMKL_DIR=C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl" "-DTBB_DIR=C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb" "-DHDF5_DIR=C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"` | 0 | `8.684 s` | MSVC `19.51.36252.0`, MKL `2026.1.0`; `Configuring done`, `Generating done`. The failed pthread probes were normal CMake feature probes; `Found Threads: TRUE`. |
+| 4 | `cmake --build .harness/build --config Debug` | 0 | `4.382 s` | Initial incremental build produced `fesa.exe` and all three test executables; warning/error scan `0/0`. |
+| 5 | `ctest --test-dir .harness/build -C Debug -R "DomainModel|InpDomainMapping|Mitc4Geometry|Mitc4Shell|DofManager|EssentialConstraints|SparseAssembly|LoadAssembly|ResultRecovery|Hdf5ResultsWriter" --output-on-failure` | 0 | `5.743 s` | `100% tests passed out of 87`; failed `0`. |
+| 6 | `ctest --test-dir .harness/build -C Debug -R "LinearStaticCli|Mitc4ShellCli" --output-on-failure` | 0 | `1.442 s` | `100% tests passed out of 10`; failed `0`. |
+| 7 | `ctest --test-dir .harness/build -C Debug -R "Mitc4ReferenceComparison|Mitc4S4Reference" --output-on-failure` | 0 | `1.377 s` | `100% tests passed out of 8`; failed `0`. |
+| 8 | `ctest --test-dir .harness/build -C Debug --show-only=json-v1` | 0 | `0.208 s` | Discovered `144` tests; `32` test names contained `Mitc4`. |
+| 9 | `ctest --test-dir .harness/build -C Debug --output-on-failure` | 0 | `9.789 s` | Initial full regression: `144/144` passed; failed `0`. |
+| 10 | First environment/reference audit helper | 1 | `0.9 s` | Non-gating report-generation error: it hardcoded nonexistent CMake internal directory `CMakeFiles/4.3.0`; both S4 hashes had already matched before the helper stopped. No build or test failed. |
+| 11 | Corrected dynamic environment/reference audit helper | 0 | `0.583 s` | Located `CMakeFiles/4.4.0/CMakeCXXCompiler.cmake`; confirmed MSVC/version, both S4 hashes, no S4 diff, and clean worktree. |
+| 12 | `cmake --build .harness/build --config Debug --target clean` | 0 | `1.251 s` | Generated Visual Studio clean target completed. |
+| 13 | `cmake --build .harness/build --config Debug` | 0 | `94.616 s` | Canonical clean rebuild compiled/linked the solver, CLI, and all test executables; `89` output lines, warning scan `0`, error scan `0`. |
+| 14 | `cmake --build .harness/build --config Debug --target fesa_unit_tests` | 0 | `1.943 s` | `fesa_solver`, GoogleTest, and `fesa_unit_tests.exe` confirmed. |
+| 15 | Focused kernel/model/assembly/recovery regex from order 5 | 0 | `4.863 s` | Canonical focused run: `87/87` passed; failed `0`. |
+| 16 | `cmake --build .harness/build --config Debug --target fesa_integration_tests` | 0 | `1.712 s` | `fesa_solver`, GoogleTest, and `fesa_integration_tests.exe` confirmed. |
+| 17 | Focused lifecycle regex from order 6 | 0 | `1.531 s` | Canonical focused run: `10/10` passed; failed `0`. |
+| 18 | `cmake --build .harness/build --config Debug --target fesa_reference_tests` | 0 | `1.769 s` | `fesa_solver`, GoogleTest, and `fesa_reference_tests.exe` confirmed. |
+| 19 | Focused reference regex from order 7 | 0 | `1.511 s` | Canonical focused run: `8/8` passed; failed `0`; S4 end-to-end tests were included. |
+| 20 | `ctest --test-dir .harness/build -C Debug --show-only=json-v1` | 0 | `0.080 s` | Canonical discovery: `144` total, `32` MITC4-named; unit/integration/reference labels `121/10/13`. |
+| 21 | `ctest --test-dir .harness/build -C Debug --output-on-failure` | 0 | `9.669 s` | Canonical full regression: `100% tests passed out of 144`; failed `0`; CTest real time `9.62 s`. |
+| 22 | Full `reference/` Git scope and worktree audit | 0 | `0.362 s` | `reference_diff_count=0`, `reference_status_count=0`; branch and HEAD unchanged. |
+| 23 | `git merge-base dev HEAD` plus changed-file scope audit | 0 | `0.321 s` | Merge base `a058ef7...`; detected `scripts/execute.py`, so Harness Python self-test was required. |
+| 24 | `uv run --with pytest python -m pytest -v -rs` | 0 | `4.651 s` | Python `3.11.15`, pytest `9.1.1`; `7 passed in 0.11s`; failed `0`. |
+
+## Validation Results
+
+| validation_stage | result | evidence |
+| --- | --- | --- |
+| clean baseline | pass | Branch and HEAD matched the requested baseline; worktree was clean before report creation. |
+| dependency precheck | pass | All four implementation-plan dependency paths existed. |
+| fresh configure | pass | VS18/x64 configure exit `0`; compiler metadata is MSVC x64. |
+| clean full build | pass | Exit `0`; solver, CLI, unit, integration, and reference targets compiled/linked. |
+| compiler warning policy | pass | Clean rebuild emitted zero warning lines and zero compiler/linker error lines; product CMake uses warnings-as-errors. |
+| feature-specific tests | pass | Approved focused regexes passed `87/87`, `10/10`, and `8/8`. |
+| CTest discovery | pass | Nonzero inventory: `144` tests, including `32` MITC4-named tests. |
+| full CTest | pass | `144/144` passed after the clean rebuild. |
+| Harness self-test | pass | Required because `scripts/execute.py` differs from the `dev` merge base; `7/7` passed. |
+| reference immutability | pass | Required S4 SHA-256 values match; entire `reference/` tree has no Git diff or status entry. |
+
+## CTest Inventory
+
+| inventory | count |
+| --- | ---: |
+| all discovered tests | 144 |
+| unit label | 121 |
+| integration label | 10 |
+| reference label | 13 |
+| test names containing `Mitc4` | 32 |
+| failed tests | 0 |
+
+The `32` MITC4-named tests comprise `30` tests in `Mitc4*` suites plus
+`DomainModel.Mitc4ShellRecordsPreserveSourceAndInternalIdentity` and
+`InpDomainMapping.MapsS4AndS4rThroughOneMitc4Identity`. The broader focused regex
+also exercises shared Domain, DOF, constraints, assembly, recovery, and HDF5 seams.
+
+CTest currently applies the target-wide legacy label
+`linear-static-3d-euler-beam` to all `144` tests and has no separate MITC4 label.
+The approved implementation plan explicitly permits exact suite regex filtering,
+which was used here, so this labeling limitation is nonblocking.
+
+## Warning and Failure Audit
+
+- Clean rebuild warning lines: `0`.
+- Clean rebuild compiler/linker error lines: `0`.
+- Focused CTest failures: `0`.
+- Full CTest failures: `0`.
+- Harness Python failures: `0`.
+- Configure probes for POSIX pthread symbols reported expected negative probe
+  results on MSVC before CMake resolved `Threads: TRUE`; this is not a configure
+  warning or failure.
+- One report-generation helper failed because it assumed a CMake internal versioned
+  directory. The corrected dynamic lookup passed. This incident did not affect
+  configure, compile, link, test, reference artifacts, or the gate verdict.
+
+## Reference Read-Only Audit
+
+| exact approved path | observed SHA-256 | expected match |
+| --- | --- | --- |
+| `reference/shell/shell.inp` | `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF` | yes |
+| `reference/shell/shell displacements.csv` | `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B` | yes |
+
+The post-test audit found zero changed paths anywhere under `reference/`. No Abaqus
+or other reference solver was run.
+
+## Failure Classification
+
+- classification: `N/A`
+- primary_failure: `N/A`
+- first_failed_build_or_test_command: `N/A`
+- failed_target_or_test: `N/A`
+- correction_handoff: `N/A`
+- non_gating_incident: report-only environment metadata lookup used the wrong CMake
+  internal directory and was rerun successfully; no product correction is required
+
+## Failed Test Inventory
+
+None. All focused, full CTest, and Harness Python tests passed.
+
+## Handoff Recommendation
+
+| target_agent | reason | required_input |
+| --- | --- | --- |
+| Reference Verification Agent | Build/test status is `pass-for-reference-verification`. | This report, source HEAD `820ba30...`, `.harness/build/tests/Debug/fesa_reference_tests.exe`, and the unchanged declared S4 input/CSV pair. |
+
+The focused reference CTest generated/used build-local comparison evidence, but the
+Build/Test Executor does not interpret that evidence as the formal reference
+tolerance verdict. Reference Verification must independently inspect and report it.
+
+## No-Change Assertion
+
+- source_files_modified: `false`
+- test_files_modified: `false`
+- cmake_files_modified: `false`
+- requirements_formulation_io_numerical_reference_documents_modified: `false`
+- phase_files_modified: `false`
+- tolerance_policies_modified: `false`
+- reference_artifacts_modified: `false`
+- owned_report_modified: `true`
+- notes: The worktree was clean at the requested source HEAD before this report was
+  created. Build products remained under the ignored `.harness/build/` directory.
+
+## Open Issues
+
+- None blocking Reference Verification.
+- The target-wide legacy CTest label prevents a precise MITC4 label count; exact
+  implementation-plan suite regexes provide the approved focused test entry point.

docs/linear-static-mitc4-shell/coordination.md

@@ -0,0 +1,42 @@
+# Linear Static MITC4 Shell Coordination Report
+
+## Metadata
+
+- feature_id: linear-static-mitc4-shell
+- status: completed
+- owner_agent: coordinator-agent
+- date: 2026-08-15
+- migration_note: Existing evidence was relocated without rerunning workflow gates.
+
+## Worklist
+
+| stage | owner | status | evidence |
+| --- | --- | --- | --- |
+| 1. requirements | requirement-agent | passed | `requirements.md` |
+| 2. research | research-agent | passed | `research.md` |
+| 3. formulation | formulation-agent | passed | `formulation.md` |
+| 4. numerical and reference | numerical-review-agent | passed | `numerical-review.md`; `reference-model.md` |
+| 5. I/O | io-definition-agent | passed | `io.md` |
+| 6. implementation and verification | implementation-planning-agent; implementation-agent | passed | `implementation-plan.md`; `build-test.md`; `reference-comparison.md` |
+| 7. physics | physics-evaluation-agent | passed | `physics-evaluation.md` |
+| 8. release | release-agent | passed | `release.md` |
+
+## Current Progress
+
+- active_sub_agent: none
+- next_handoff: none
+- closure_evidence: `release.md` status `ready-for-release`
+- workflow_disposition: completed
+- historical_gap: No standalone `implementation-report.md` existed before migration; no replacement was fabricated.
+
+## Decision Log
+
+| date | decision | evidence | rationale |
+| --- | --- | --- | --- |
+| 2026-08-15 | Record migrated workflow as completed | `release.md` | Preserve the existing release verdict without retroactive gate execution. |
+
+## Blockers and Rework
+
+- active_blockers: none
+- correction_attempt_count: 0
+- needs_user_decision: false

docs/linear-static-mitc4-shell/formulation.md

@@ -3,12 +3,12 @@
 ## Metadata
 
 - feature_id: `linear-static-mitc4-shell`
-- source_requirement: `docs/requirements/linear-static-mitc4-shell.md`
-- source_research: `docs/research/linear-static-mitc4-shell-research.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-mitc4-shell-review.md`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_research: `docs/linear-static-mitc4-shell/research.md`
+- source_numerical_review: `docs/linear-static-mitc4-shell/numerical-review.md`
 - status: `approved-for-implementation-planning`
 - owner_agent: `formulation-agent`
-- date: `2026-08-12`
+- date: `2026-08-13`
 - revision_basis: approved independent-reference policy, fixed drilling rule, and
   removal of calibration gates `NR-O01` through `NR-O04`
 - revision_state: `numerical-review-passed`
@@ -1540,13 +1540,15 @@ or drilling-energy warning is part of this check.
 
 ### 17.5 Reference-comparison boundary
 
-Abaqus comparisons block only on matched global `U1/U2/U3` rows under
-`tolerance_c=1e-9+1e-6*reference_scale_c`, where `reference_scale_c` is the maximum
-absolute finite Abaqus value for the same component. `UR1/UR2/UR3` uses the same
-formula but an exceedance emits only a deterministic nonblocking warning.
+Abaqus comparison uses only the declared full-integration S4 case. Matched global
+`U1/U2/U3` families are blocking and `UR1/UR2/UR3` families are warning-only. Both use
+the common family-scale row rule (`S=max(abs(reference))`, near-zero ratio `0.01`, relative
+tolerance `0.05`) and scale-relative RMS tolerance `0.01`. `S=0` requires exact-zero FESA
+values and there is no independent absolute-error gate.
 FESA `S4` and `S4R` inputs must produce the same internal numerical rows for identical
-supported models while preserving distinct source metadata. Abaqus S4 and S4R are
-not expected to be numerically identical on finite meshes.
+supported models while preserving distinct source metadata. This common-path property
+is verified without consuming an S4R Abaqus artifact; Abaqus S4R is not an acceptance
+reference for this full-integration FESA formulation.
 
 ### 17.6 Future nonlinear tangent verification
 
@@ -1584,7 +1586,7 @@ procedure.
 
 The detailed source tiers, extracted facts, page references, benchmark provenance,
 and evidence limits are owned by
-`docs/research/linear-static-mitc4-shell-research.md`. The primary local source set
+`docs/linear-static-mitc4-shell/research.md`. The primary local source set
 under `docs/reference-papers/MITC4/` includes:
 
 - `AContinuumMechanicsBasedFourNodeShell_001.md` and `_002.md`;
@@ -1628,9 +1630,9 @@ research brief remain the project source of truth.
 | `017-020` | global 24-DOF virtual work; `rho_M <= 1e-12`; distributed loads excluded | I/O diagnostic/schema |
 | `024-029` | deterministic element buffers, partitioned linear lifecycle, full-residual reaction | planning |
 | `039-048` | nodal/global and shell/local recovery inventory, units, identities, physical shell energy | I/O schema |
-| `049-057` | normalized invariants, patches, fixed drilling, declared references and equilibrium | Numerical Review/reference/physics |
-| `058-064` | exact B33 mixed tolerance; U blocking and UR warning-only | reference verification |
-| `065-072` | exact existing S4/S4R paths, immutability and displacement-only boundary | reference model |
+| `049-057` | normalized invariants, patches, fixed drilling, declared S4 reference and equilibrium | Numerical Review/reference/physics |
+| `058-064` | common family-scale row/RMS policy; U blocking and UR warning-only | reference verification |
+| `065-072` | exact existing S4 paths, S4R reference non-consumption, immutability and displacement-only boundary | reference model |
 
 ### 20.1 Numerical Review revision traceability
 
@@ -1678,10 +1680,13 @@ items or an expanded reference portfolio as missing evidence.
 
 ### 21.3 Reference Model handoff
 
-- Record the existing `reference/shell/` S4 and `reference/shellR/` S4R input and
-  displacement CSV paths, but compare only formulation-independent global displacement
-  evidence rather than claiming element equivalence.
-- Use the exact B33 mixed tolerance; do not add administrative metadata or portfolio gates.
+- Record only the existing `reference/shell/` S4 input and displacement CSV as the
+  acceptance pair and compare formulation-independent global displacement evidence
+  without claiming element equivalence.
+- Do not consume `reference/shellR/` in acceptance comparison; preserve S4R support
+  through source-mapping/common-kernel/metadata tests.
+- Use the project-wide common family-scale row/RMS tolerance; do not add an independent
+  absolute-error gate, administrative metadata or portfolio gates.
 - Do not create, repair, rename, or run reference artifacts during this formulation
   gate.
 

docs/linear-static-mitc4-shell/implementation-plan.md

@@ -0,0 +1,921 @@
+# Linear Static MITC4 Shell Implementation Plan
+
+> **For agentic implementation workers:** 이 문서는 사용자 승인 전 `draft`다. 승인 후
+> Implementation Planning Agent가 같은 내용으로 Harness phase planning files만
+> materialize한다. Implementation Agent는 승인된 phase의 pending Step을
+> `docs/HARNESS.md`와 `docs/HARNESS_WORKFLOW.md`에 따라 하나씩 실행해야 하며,
+> Harness hook/runner lifecycle을 우회해서는 안 된다.
+
+## 1. Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- document_type: `implementation-plan`
+- status: `ready-for-implementation`
+- approval_state: `harness-step-draft-approved-2026-08-12`
+- owner_agent: `implementation-planning-agent`
+- date: `2026-08-13`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_research: `docs/linear-static-mitc4-shell/research.md`
+- source_formulation: `docs/linear-static-mitc4-shell/formulation.md`
+- source_numerical_review:
+  `docs/linear-static-mitc4-shell/numerical-review.md`
+- source_io_definition:
+  `docs/linear-static-mitc4-shell/io.md`
+- source_reference_models:
+  `docs/linear-static-mitc4-shell/reference-model.md`
+- target_platform: `Windows x64 / MSVC / C++17`
+- build_system: `CMake + CTest`
+- execution_infrastructure: `Python Harness`
+- current_numerical_verdict: `pass-for-implementation-planning`
+- harness_task_name_candidate: `linear-static-mitc4-shell`
+- implementation_authorized: `true-user-requested-2026-08-12`
+- harness_execution_authorized: `true-user-requested-2026-08-12`
+- phase_files_authorized: `true-user-approved-2026-08-12`
+
+### Goal
+
+승인된 Abaqus `.inp` subset의 `S4` 및 `S4R` source element를 하나의 FESA
+`FESA-MITC4` 선형 정적 shell formulation으로 해석하고, 물리적으로 분리된 fixed
+drilling stabilization, deterministic sparse assembly, mandatory HDF5 shell results,
+sole declared S4 displacement reference case를 end-to-end로 연결한다.
+
+### Architecture
+
+기존 B33 경로를 파괴하거나 speculative common element hierarchy를 만들지 않는다.
+Domain에 shell semantic records를 추가하고, 별도 concrete `Mitc4Shell` kernel과
+shell-specific recovery를 만든다. 기존 `Analysis` lifecycle, six-DOF nodal
+load/constraint, sparse COO reduction, `LinearSolver`, generic `ResultsWriter` 경계,
+HDF5 temporary/self-check/atomic replacement는 재사용한다.
+
+### Tech Stack
+
+- C++17, MSVC `/W4 /WX`
+- CMake 3.25+, Visual Studio 18 2026 generator, x64 Debug
+- GoogleTest/CTest
+- Intel oneAPI MKL and TBB
+- HDF5 schema version 0
+- Python 3 Harness runner and repository hooks
+
+### Global Constraints
+
+1. 모든 production behavior는 같은 Step 안에서 `RED -> observed failure ->
+   minimal GREEN -> focused VERIFY -> full VERIFY`를 닫는다.
+2. production C++ 파일 변경은 관련 C++ test와 함께 수행한다.
+3. public solver headers에 MKL, TBB, HDF5, Win32/vendor type을 노출하지 않는다.
+4. Node/Element에 equation ID를 저장하지 않고 `DofManager`만 numbering, scatter,
+   constrained/free mapping 및 sparse pattern을 소유한다.
+5. `S4`와 `S4R`은 같은 FESA numerical path를 사용하며 source metadata만 다르다.
+6. physical 20-DOF kernel과 four numerical drilling coordinates를 recovery/output에서
+   분리한다.
+7. reference files를 생성, 수정, rename, repair, normalize하지 않으며 Abaqus를
+   실행하지 않는다.
+8. future geometrically nonlinear residual/tangent, distributed loads, mixed beam-shell
+   models, reduced integration/hourglass, drilling calibration/energy output,
+   `NR-O03`/`NR-O04`는 구현 범위 밖이다.
+9. 이 `draft`의 승인은 phase planning files 작성만 허용한다. Harness 실행은 별도의
+   명시적 사용자 요청이 있어야 한다.
+
+## 2. Readiness Check
+
+| gate | evidence | status | planning consequence |
+| --- | --- | --- | --- |
+| Requirements | `docs/linear-static-mitc4-shell/requirements.md`, 001-072 approved | pass | 모든 must를 task/test에 추적 |
+| Research | `docs/linear-static-mitc4-shell/research.md` | pass | source-backed MITC4 tying/director/drilling 경계 유지 |
+| Formulation | `docs/linear-static-mitc4-shell/formulation.md`, `approved-for-implementation-planning` | pass | linear sections만 구현; Section 15 future nonlinear 제외 |
+| Numerical Review | `docs/linear-static-mitc4-shell/numerical-review.md` | pass | critical blocker 없음; planning authorized |
+| I/O | `docs/linear-static-mitc4-shell/io.md`, `approved-for-implementation-planning` | pass | keyword, diagnostic, HDF5 schema를 그대로 구현 |
+| Reference Model | `docs/linear-static-mitc4-shell/reference-model.md` | pass | sole S4 input/CSV pair만 read-only acceptance input으로 사용 |
+| Repository seams | parser/model, element/analysis, result/reference 영역 read-only 조사 | pass | candidate files와 current signatures 확인 |
+| Toolchain paths | GoogleTest/MKL/TBB/HDF5 config directories 존재 | pass | Section 10의 exact configure command 사용 가능 |
+
+Toolchain path evidence was resolved in this workspace on `2026-08-12`with
+`Test-Path -LiteralPath`returning `True`for all four literal directories used in
+Sections 10 and 12. They are environment-resolved paths, not placeholders. Each Step
+copies the fail-fast path precheck before CMake so a changed machine configuration
+stops before RED/GREEN evidence is recorded.
+
+Declared read-only reference inventory:
+
+| case | source | displacement CSV | SHA-256 |
+| --- | --- | --- | --- |
+| S4 | `reference/shell/shell.inp` | `reference/shell/shell displacements.csv` | input `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF`; CSV `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B` |
+
+README, `metadata.json`, Abaqus provenance/version, canonical naming 및 추가 portfolio는
+readiness gate가 아니다. 현재 열린 product/numerical 결정은 없다. 남은 승인 항목은
+Section 12의 multi-Step Harness draft뿐이다.
+
+## 3. Implementation Scope
+
+### Included behavior
+
+- exactly one `*STEP, *STATIC`, small displacement/small rotation
+- four-node `S4`/`S4R` parsing and one internal `FESA-MITC4` identity
+- stable source instance/label/type and four-node source ordering
+- homogeneous isotropic `E,nu` and centered single-layer constant `t`
+- deterministic positive-thickness nodal directors and right-handed frames
+- six global nodal DOFs; 24-entry shell scatter
+- physical membrane, bending, transverse shear with MITC edge-midpoint tying
+- common `2x2x2` Gauss integration for S4 and S4R
+- fixed `k_d=10^-3 min(R+)` drilling stabilization from the eight physical
+  tangent-rotation diagonals only
+- nodal BC and global CLOAD; aggregate director-parallel moment rejection
+- deterministic global stiffness, constrained partition, factorize-before-load,
+  effective RHS, solve, full residual/reaction
+- physical-only shell recovery, energy/equilibrium/verification metrics
+- exact mandatory HDF5 model/result schema and failure-atomic finalization
+- direct HDF5-to-Abaqus-displacement comparison: U blocking, UR warning-only
+
+### Non-goals
+
+- Abaqus `S4`/`S4R` algorithm equivalence
+- B31 mapping, mixed B33/MITC4 model execution or common public element hierarchy
+- `*DLOAD`, pressure, gravity, body/edge/follower load
+- composite, offsets, variable thickness, explicit normal/orientation
+- reduced integration, hourglass control, MITC4+
+- nonlinear state, finite director update, nonlinear residual/tangent execution
+- drilling physical strain/load/result/stress/energy, coefficient sweep or calibration
+- `NR-O03` smooth-angle and `NR-O04` warp/distortion threshold sweeps
+- extra shell benchmarks as a completion gate
+- Abaqus execution or reference artifact mutation
+
+## 4. Work Breakdown
+
+Tasks are ordered by implementation dependency. Harness executes Steps 0-13 in strict
+order even when the minimal dependency set below is smaller.
+
+| task | depends on |
+| --- | --- |
+| TASK-00 | none |
+| TASK-01 | TASK-00 |
+| TASK-02 | TASK-01 |
+| TASK-03 | TASK-02 |
+| TASK-04 | TASK-03 |
+| TASK-05 | TASK-04 |
+| TASK-06 | TASK-00; executed after TASK-05 |
+| TASK-07 | TASK-04, TASK-06 |
+| TASK-08 | TASK-02, TASK-07 |
+| TASK-09 | TASK-05, TASK-08 |
+| TASK-10 | TASK-05, TASK-09 |
+| TASK-11 | TASK-09, TASK-10 |
+| TASK-12 | TASK-01, TASK-07, TASK-08, TASK-10, TASK-11 |
+| TASK-13 | TASK-11, TASK-12 |
+| TASK-14 | TASK-13 |
+| TASK-15 | TASK-01, TASK-07, TASK-11 |
+| TASK-16 | TASK-00 through TASK-15 |
+
+### TASK-00: `shell-semantic-model`
+
+- Own only model semantic records and immutable Domain access.
+- Candidate additions: `Mitc4ShellDefinition`, `ShellSection`, source type enum/value,
+  internal formulation identity, four-node indices, material/section indices, node
+  director/frame storage.
+- Preserve current B33 records and stable source/internal identity; reject mixed models
+  at mapping rather than designing a heterogeneous solver hierarchy.
+- Tests: `MITC4-MODEL-001`, `MITC4-MODEL-002`.
+
+### TASK-01: `shell-domain-mapping`
+
+- Extend `AbaqusDomainMapper::map(const ParsedInput&)`; keep `AbaqusInputReader`
+  syntax-only.
+- Accept exact S4/S4R and single-layer SHELL SECTION grammar, resolve ELSET/material,
+  identity wrappers and six-DOF BC/CLOAD.
+- Fail closed with exact I/O diagnostic classes for connectivity, assignment,
+  unsupported option/procedure/mixed model/distributed load.
+- Tests: `MITC4-MAP-001` through `MITC4-MAP-004`.
+
+### TASK-02: `shell-director-geometry`
+
+- Add deterministic geometry preprocessing after instance expansion: element normal
+  candidates, area-weighted nodal director, pairwise orientation validation and nodal
+  frame tie-break.
+- Validate center, eight stiffness points, four tying points and committed recovery
+  points for finite bases, nonzero surface measure and positive finite Jacobian.
+- Do not introduce calibrated angle, warp or distortion thresholds.
+- Tests: `MITC4-GEO-001` through `MITC4-GEO-004`.
+
+### TASK-03: `mitc4-kinematics-constitutive`
+
+- Create concrete candidate `Mitc4Shell` without a public base hierarchy.
+- Implement shape identities, local frames, `T_p`/`T_d` channel maps, direct
+  membrane/bending terms, all four covariant MITC tying samples/interpolation,
+  engineering-shear factors, `C_ps/C_5/A/D/A_s` and fixed point order.
+- Keep all intermediate physical coordinates at 20 DOFs.
+- Tests: `MITC4-KIN-001` through `MITC4-KIN-005`.
+
+### TASK-04: `mitc4-stiffness-drilling`
+
+- Form physical `K20` with common `2x2x2` integration and embed by congruence.
+- Build `R+` only from the eight physical tangent-rotation diagonals, use exact
+  `k_d=10^-3 min(R+)` and `T_d^T(k_d I4)T_d`; fail empty/nonfinite `R+`.
+- Tests: `MITC4-KERNEL-001` through `MITC4-KERNEL-006`.
+
+### TASK-05: `mitc4-physical-recovery`
+
+- Add only the element-kernel physical recovery seam for generalized strain/resultant,
+  bottom/middle/top stress and physical element energy.
+- Prove pure numerical drill action contributes zero to every physical recovery value.
+- Tests: `MITC4-KERNEL-007`, `MITC4-PHYSREC-001`.
+
+### TASK-06: `shell-dof-scatter`
+
+- Retain six per-node full DOFs and add typed 24-entry shell scatter/pattern access;
+  preserve existing 12-entry B33 behavior.
+- Preserve stable full/free/constrained numbering and all-constrained `0x0 Kff`.
+- Tests: `MITC4-DOF-001` through `MITC4-DOF-003`.
+
+### TASK-07: `shell-sparse-assembly`
+
+- Dispatch shell topology to `Mitc4Shell::globalStiffness()`.
+- Emit 576 element-local COO entries with stable `elementOrder`/`localOrder` into
+  worker-owned buffers; preserve canonical fixed reduction and diagonal slots.
+- Tests: `MITC4-ASM-001` through `MITC4-ASM-003`.
+
+### TASK-08: `shell-load-validation`
+
+- Reuse global six-DOF `LoadAssembler` and existing constraint partition.
+- Aggregate CLOAD deterministically before testing exact-zero or
+  `rho_M=abs(dot(d,M))/norm(M)<=1e-12`; rejected moments never reach stabilization.
+- Preserve force/moment units and factorization-before-load lifecycle.
+- Tests: `MITC4-LOAD-001` through `MITC4-LOAD-004`.
+
+### TASK-09: `shell-analysis-state`
+
+- Add shell result rows to `AnalysisState` and `result_records.hpp`: four fixed
+  midsurface locations, local frames, eight generalized strains/resultants,
+  bottom/middle/top stress, physical energy, equilibrium and verification metrics.
+- Preserve stable row order and expose candidate-owned containers only; perform no
+  element calculation in this task.
+- Tests: `MITC4-STATE-001` through `MITC4-STATE-003`.
+
+### TASK-10: `shell-result-recovery`
+
+- Recover full residual `K*d-F` for nodal reaction/equilibrium evidence but call the
+  kernel physical-only recovery for shell quantities.
+- Validate candidates fully before committing state; keep stable row order.
+- Tests: `MITC4-REC-001` through `MITC4-REC-005`.
+
+### TASK-11: `shell-hdf5-output`
+
+- Extend `Hdf5ResultsWriter` behind the unchanged generic `ResultsWriter::write`
+  boundary.
+- Add exact schema paths/shapes/component attributes from the I/O contract without
+  changing B33 dataset meaning.
+- Reuse temporary write, finite/schema self-check, close/reopen and atomic replace.
+- Tests: `MITC4-H5-001` through `MITC4-H5-004`.
+
+### TASK-12: `shell-linear-static-flow`
+
+- Route a shell Domain through the existing eight-hook `Analysis::run()` lifecycle.
+- Preserve assemble/partition, factorize, load/effective RHS, substitute,
+  reconstruct, full residual/recover, atomic write order and single factorization.
+- Validate nonzero prescribed values, singular support and valid all-constrained solve.
+- Tests: `MITC4-FLOW-001` through `MITC4-FLOW-004`.
+
+### TASK-13: `shell-reference-comparison`
+
+- Add a test-only MITC4 comparator rather than widening beam-specific assumptions.
+- Consume exactly one declared input and displacement CSV per case; map instance,
+  source node and six displacement components directly to HDF5.
+- Precheck header/row set/duplicates/nonfinite/schema/identity before numeric work.
+- Apply fixed absolute `1.0e-5` to every U/UR row without a scale-dependent decision
+  term; U blocks, UR only emits deterministic warning; record all required metrics
+  and worst row.
+- Tests: `MITC4-REF-001` through `MITC4-REF-006`.
+
+### TASK-14: `shell-s4-end-to-end`
+
+- This is declared-case sub-work owned by TASK-13/Step 13's test-only reference module,
+  not an independent production-module Step.
+- Run the declared S4 deck through CLI, authoritative HDF5, schema checks and
+  read-only comparator.
+- Assert source type `S4`, internal `FESA-MITC4` and U pass/UR report behavior.
+- Tests: `MITC4-E2E-S4-001`, `MITC4-E2E-S4-002`.
+
+### TASK-15: `shell-s4r-common-path-verification`
+
+- This is non-reference coverage distributed across the existing domain-mapping,
+  sparse-assembly and HDF5 tests; Step 13 does not consume an S4R Abaqus artifact.
+- Assert S4R selects the same FESA-MITC4 integration/kernel/assembly path as an
+  otherwise identical S4 semantic fixture while preserving only source metadata.
+- Tests: `InpDomainMapping.MapsS4AndS4rThroughOneMitc4Identity`,
+  `SparseAssembly.S4AndS4rSemanticFixturesAssembleIdenticalStiffness`, and the
+  existing shell HDF5 source-type metadata coverage.
+
+### TASK-16: `mitc4-full-verification`
+
+- Run MSVC x64 Debug full build, CTest discovery, focused feature suites and full CTest.
+- Audit warnings, deterministic repeat evidence, required HDF5 inventory, reference
+  immutability and git diff scope.
+- Produce only downstream implementation evidence; do not claim reference verification,
+  physics sanity or release readiness.
+- Tests/audits: `MITC4-VERIFY-001` through `MITC4-VERIFY-004`.
+
+## 5. TDD Test Plan
+
+| test id | initial RED assertion | minimal GREEN behavior | focused verification |
+| --- | --- | --- | --- |
+| MITC4-MODEL-001 | four-node S4/S4R records/director/source identity do not exist | immutable shell records preserve source and internal identity | `DomainModel.*` |
+| MITC4-MODEL-002 | shell section/material/thickness ownership unavailable | one resolved centered homogeneous assignment per element | `DomainModel.*` |
+| MITC4-MAP-001 | valid S4/S4R and SHELL SECTION reject | both source types map to FESA-MITC4 | `InpDomainMapping.*` |
+| MITC4-MAP-002 | duplicate/unresolved/conflicting section cases lack exact errors | exact assignment/material validation | `InpDomainMapping.*` |
+| MITC4-MAP-003 | excluded shell options/mixed models may be ignored | exact fail-closed diagnostics | `InpDomainMapping.*` |
+| MITC4-MAP-004 | second/nonstatic/NLGEOM/DLOAD cases lack shell boundary tests | exact procedure/load rejection; no-op output unchanged | `InpDomainMapping.*` |
+| MITC4-GEO-001 | planar/rotated/warped directors unavailable | deterministic unit director and right-handed frame | `Mitc4Geometry.*` |
+| MITC4-GEO-002 | incident orientation/tie-break is untested | stable area-weighted result across repeated order | `Mitc4Geometry.*` |
+| MITC4-GEO-003 | duplicate/bow-tie/zero/reversed/nonfinite geometry may pass | exact geometry failures | `Mitc4Geometry.*` |
+| MITC4-GEO-004 | required-point Jacobian inventory absent | every required point finite and positive | `Mitc4Geometry.*` |
+| MITC4-KIN-001 | shape/derivative identities unavailable | partition/unity/derivative identities | `Mitc4ShellKinematics.*` |
+| MITC4-KIN-002 | frame/T transforms unavailable | orthonormal/right-handed frames and channel selection | `Mitc4ShellKinematics.*` |
+| MITC4-KIN-003 | tying values/weights unavailable | four hand-computed covariant tying values | `Mitc4ShellKinematics.*` |
+| MITC4-KIN-004 | constitutive/section matrices unavailable | exact coefficients, symmetry, positivity, unit rescaling | `Mitc4ShellConstitutive.*` |
+| MITC4-KIN-005 | quadrature order not fixed | exact common 2x2x2 points/weights | `Mitc4ShellKinematics.*` |
+| MITC4-KERNEL-001 | no 24x24 shell stiffness | finite symmetric 24x24 stiffness | `Mitc4ShellKernel.*` |
+| MITC4-KERNEL-002 | energy congruence unavailable | 20/24 virtual work and energy equal within 1e-12 | `Mitc4ShellKernel.*` |
+| MITC4-KERNEL-003 | rigid/rank checks fail | six physical rigid modes, physical rank 14, stabilized rank 18 | `Mitc4ShellKernel.*` |
+| MITC4-KERNEL-004 | deformation energy/patch fields unavailable | positive membrane/bending/shear/twist modes | `Mitc4ShellPatch.*` |
+| MITC4-KERNEL-005 | R+ membership/coefficient not implemented | exact rotational-only min and 1e-3 factor | `Mitc4ShellDrilling.*` |
+| MITC4-KERNEL-006 | empty R+ and pure drill behavior unspecified | deterministic failure; pure drill stabilized | `Mitc4ShellDrilling.*` |
+| MITC4-KERNEL-007 | drilling contaminates recovery/energy | physical recovery/energy exact zero for pure drill | `Mitc4ShellDrilling.*` |
+| MITC4-PHYSREC-001 | kernel recovery contract is not independently callable | exact physical strain/resultant/stress/energy for hand field | `Mitc4ShellPhysicalRecovery.*` |
+| MITC4-DOF-001 | shell scatter fixed at 12 | stable 24-entry scatter in node/component order | `DofManager.*` |
+| MITC4-DOF-002 | shell pattern unavailable | sorted unique pattern with all diagonals | `DofManager.*` |
+| MITC4-DOF-003 | constraint roundtrip lacks shell-sized system | no/mixed/all constraints and nonzero values roundtrip | `DofManager.*; EssentialConstraints.*` |
+| MITC4-ASM-001 | assembler rejects four-node element | 576 stable local contributions assemble | `SparseAssembly.*` |
+| MITC4-ASM-002 | thread/repetition may reorder | serial/TBB/repeated CSR bytes and values match | `SparseAssembly.*` |
+| MITC4-ASM-003 | S4/S4R could branch numerically | identical semantic fixtures produce identical K | `SparseAssembly.*` |
+| MITC4-LOAD-001 | shell force/moment fixture unavailable | six global components aggregate stably | `LoadAssembly.*` |
+| MITC4-LOAD-002 | director-parallel exact-zero branch untested | zero aggregate moment accepted | `LoadAssembly.*` |
+| MITC4-LOAD-003 | rho boundary/rejection untested | <=1e-12 accepted; >1e-12 exact diagnostic | `LoadAssembly.*` |
+| MITC4-LOAD-004 | rejected moment might enter drill channel | failure occurs before RHS/substitution | `LoadAssembly.*` |
+| MITC4-STATE-001 | AnalysisState has no shell row containers | exact shell row/component/location types are owned | `AnalysisState.*` |
+| MITC4-STATE-002 | physical energy/equilibrium/metrics are absent | finite candidate global evidence is stored | `AnalysisState.*` |
+| MITC4-STATE-003 | shell candidate rollback is unavailable | invalid candidate leaves prior state unchanged | `AnalysisState.*` |
+| MITC4-REC-001 | shell row types absent | four ordered location/frame/strain/resultant rows | `ResultRecovery.*` |
+| MITC4-REC-002 | bottom/middle/top stress absent | direct S11/S22/S12 recovery in fixed order | `ResultRecovery.*` |
+| MITC4-REC-003 | energy may include drilling | physical strain energy excludes stabilization | `ResultRecovery.*` |
+| MITC4-REC-004 | reaction/equilibrium semantics may diverge | full residual, force/moment balance, normalized metrics | `ResultRecovery.*` |
+| MITC4-REC-005 | invalid partial rows may commit | nonfinite/inventory failure preserves old state | `ResultRecovery.*` |
+| MITC4-H5-001 | shell metadata/model schema missing | exact metadata, elements, director/frame, material/section | `Hdf5ResultsWriter.*` |
+| MITC4-H5-002 | mandatory shell result paths missing | exact displacement/reaction/frame/strain/resultant/stress/global rows | `Hdf5ResultsWriter.*` |
+| MITC4-H5-003 | output request could filter or drilling paths appear | inventory unconditional; forbidden paths absent | `Hdf5ResultsWriter.*` |
+| MITC4-H5-004 | invalid shell candidate may replace final file | self-check failure preserves prior final | `Hdf5ResultsWriter.*` |
+| MITC4-FLOW-001 | shell cannot traverse Analysis lifecycle | exact hook order and one factorization | `LinearStaticCli.*` |
+| MITC4-FLOW-002 | prescribed shell RHS/reconstruction absent | Ff-Kfc*dc and full displacement correct | `LinearStaticCli.*` |
+| MITC4-FLOW-003 | singular/all-constrained behavior unproven | singular fails; 0x0 Kff succeeds | `LinearStaticCli.*` |
+| MITC4-FLOW-004 | failed output may commit state | candidate state/output commits only after validation | `LinearStaticCli.*` |
+| MITC4-REF-001 | six-column CSV header unsupported | exact MITC4 header maps instance/node/U/UR | `Mitc4ReferenceComparison.*` |
+| MITC4-REF-002 | invalid row inventory may be ignored | missing/extra/duplicate/nonfinite/schema mismatch fails first | `Mitc4ReferenceComparison.*` |
+| MITC4-REF-003 | tolerance could remain scale-dependent or clamp/row-normalize | exact fixed absolute `1.0e-5` for every U/UR row | `Mitc4ReferenceComparison.*` |
+| MITC4-REF-004 | UR may block verdict | U blocks; UR only warns | `Mitc4ReferenceComparison.*` |
+| MITC4-REF-005 | report metrics/order incomplete | row decisions, max, normalized, RMS, vector, worst row deterministic | `Mitc4ReferenceComparison.*` |
+| MITC4-REF-006 | comparator may require administrative files | only declared input/CSV/HDF5 required | `Mitc4ReferenceComparison.*` |
+| MITC4-E2E-S4-001 | S4 deck cannot produce valid shell HDF5 | CLI succeeds with exact S4 metadata/schema | `Mitc4S4Reference.*` |
+| MITC4-E2E-S4-002 | S4 U comparison unavailable | all U rows pass; UR fully reported | `Mitc4S4Reference.*` |
+| MITC4-VERIFY-001 | feature tests not discoverable | all planned suites discovered/labeled | CTest JSON inventory |
+| MITC4-VERIFY-002 | warnings/regressions unknown | MSVC Debug full build passes /W4 /WX | full build |
+| MITC4-VERIFY-003 | nondeterminism unknown | repeated focused/full tests pass | full CTest |
+| MITC4-VERIFY-004 | artifact governance unknown | reference hashes/path and git diff unchanged | read-only audit |
+
+Numerical thresholds are fixed: frame/symmetry/transformation-energy `<=1e-12`;
+rigid action, linear residual and global equilibrium `<=1e-10`. Patch tests compare
+against independently hand-computed analytical fields and signs, not the production
+routine itself.
+
+## 6. CMake/CTest Plan
+
+### Existing topology to preserve
+
+- production library: `fesa_solver`
+- CLI: `fesa_cli`
+- unit: `fesa_unit_tests`
+- integration: `fesa_integration_tests`
+- reference: `fesa_reference_tests`
+- aggregate: `fesa_tests`
+
+No new executable target is required. New kernel/test/comparator source files are
+explicitly registered in existing source lists. Add `linear-static-mitc4-shell` as an
+additive CTest label without removing the B33 label. Because
+`gtest_discover_tests()` currently applies target-wide labels, focused Step
+verification uses exact suite regexes; label-based MITC4 filtering is enabled only
+if per-test labeling can be added without relabeling unrelated B33 tests.
+
+Candidate registrations:
+
+- `src/fesa/CMakeLists.txt`: candidate `model/shell_geometry.cpp`,
+  `elements/mitc4_shell.cpp` and any separate shell recovery implementation.
+- `tests/CMakeLists.txt`: candidate `shell_geometry_test.cpp`,
+  `mitc4_shell_test.cpp`, optional focused shell recovery/HDF5 tests, and MITC4
+  reference comparator/test sources.
+- Preserve existing runtime staging for MKL/TBB/HDF5 on all three test executables.
+
+CTest discovery must show every suite named in Section 5 before implementation is
+considered verified.
+
+## 7. Candidate Files and Ownership
+
+All signatures in this section are candidates for the Implementation Agent to confirm
+against the approved contracts during the owning Step; they are not new approved APIs.
+
+| owner/module | candidate files | candidate interface/direction |
+| --- | --- | --- |
+| model | `include/fesa/model/model_types.hpp`, `include/fesa/model/domain.hpp`, `src/fesa/model/domain.cpp` | separate shell records/accessors; no equation IDs |
+| shell semantic geometry | candidate new `include/fesa/model/shell_geometry.hpp`, `src/fesa/model/shell_geometry.cpp` | director/frame and required-point geometry validation without parser/kernel dependency |
+| Abaqus semantic map | `include/fesa/io/abaqus/domain_mapper.hpp`, `src/fesa/io/abaqus/domain_mapper.cpp` | keep `map(const ParsedInput&) -> Result<Domain>` |
+| analysis view | `include/fesa/analysis/analysis_model.hpp`, `src/fesa/analysis/analysis_model.cpp` | non-owning active shell view only if record representation requires |
+| shell kernel | new `include/fesa/elements/mitc4_shell.hpp`, `src/fesa/elements/mitc4_shell.cpp` | concrete create/stiffness/physical-recover API |
+| DOF | `include/fesa/fem/dof_manager.hpp`, `src/fesa/fem/dof_manager.cpp` | typed `array<size_t,24>` shell scatter beside B33 scatter |
+| sparse assembly | `src/fesa/assembly/sparse_assembler.cpp` | topology dispatch; unchanged public assemble signature |
+| load | `src/fesa/assembly/load_assembler.cpp` only if aggregate director projection cannot live in semantic validation | reuse full six-DOF assembly |
+| state/results | `include/fesa/analysis/analysis_state.hpp`, `src/fesa/analysis/analysis_state.cpp`, `include/fesa/results/result_records.hpp` | additive shell rows/global metrics |
+| recovery | `include/fesa/results/result_recovery.hpp`, `src/fesa/results/result_recovery.cpp` or new shell-specific cpp | candidate-then-commit; physical/drill split |
+| HDF5 | `src/fesa/io/hdf5/hdf5_results_writer.cpp` | generic public writer unchanged |
+| lifecycle | `src/fesa/analysis/linear_static_analysis.cpp` | preserve existing hook order |
+| app | `tests/integration/app/fesa_application_test.cpp`; production app only if feature dispatch needs it | CLI syntax unchanged |
+| reference tests | new `tests/reference/mitc4_reference_comparison.hpp/.cpp` and tests | test-only direct HDF5/CSV comparator |
+| build graph | `src/fesa/CMakeLists.txt`, `tests/CMakeLists.txt` | explicit additive source/test registration |
+
+Candidate kernel seam:
+
+```cpp
+class Mitc4Shell {
+public:
+    static Result<Mitc4Shell> create(
+        std::array<const Node*, 4> nodes,
+        std::array<Vector3, 4> initialDirectors,
+        const ShellSection& section,
+        const LinearElasticMaterial& material);
+
+    [[nodiscard]] Matrix physicalLocalStiffness20() const;
+    [[nodiscard]] Matrix globalStiffness24() const;
+    [[nodiscard]] Result<ShellRecovery> recoverPhysical(
+        const Vector& globalElementDisplacement24) const;
+};
+```
+
+This candidate explicitly avoids returning drill energy/results and avoids a common
+public `Element` hierarchy. Exact value/reference ownership and math types are decided
+inside TASK-03 after compiling the first RED test.
+
+## 8. Data Flow Contract
+
+```text
+.inp bytes
+  -> AbaqusInputReader (syntax/source locations only)
+  -> AbaqusDomainMapper (approved semantics, S4/S4R provenance, section/material)
+  -> geometry/director preprocessing
+  -> immutable Domain
+  -> non-owning AnalysisModel
+  -> DofManager (6 DOF/node, 24-entry shell scatter, free/constrained maps)
+  -> Mitc4Shell physical K20 + fixed drill embedding -> global K24
+  -> worker-local COO -> stable SparseMatrix reduction
+  -> Kff/Kfc/Kcf/Kcc partition -> Kff factorize
+  -> aggregate CLOAD/director check -> Ff-Kfc*dc
+  -> substitute -> full displacement
+  -> full residual K*d-F
+  -> physical-only shell recovery + global evidence
+  -> validated candidate AnalysisState
+  -> temporary results.h5 -> schema self-check -> atomic finalization
+
+authoritative results.h5
+  -> test-only MITC4 HDF5 projection
+  -> exact source instance/node/component row-set precheck
+  -> read-only declared Abaqus displacement CSV
+  -> U blocking / UR warning-only deterministic report
+```
+
+Invariants:
+
+- Domain outlives AnalysisModel; no Domain copying.
+- Source labels/instances/types never become equation indices.
+- `S4`/`S4R` source type is metadata; both call the same kernel/quadrature.
+- full residual uses stabilized global K; shell strain/resultant/stress/physical energy
+  use physical recovery only.
+- HDF5 `nodal/reaction` is full residual; constraint mask decides reaction versus free
+  residual evidence.
+- result rows use fixed GP1..GP4 and BOTTOM/MIDDLE/TOP identity; no averaging or Abaqus
+  integration-point relabeling.
+- state/output mutation occurs only after complete validation.
+
+## 9. Acceptance Traceability Matrix
+
+The inclusive ranges below cover every must requirement `001` through `072` exactly
+once without gap or overlapping requirement range.
+
+| requirement range | owning tasks | test/evidence id | reference model id | acceptance |
+| --- | --- | --- | --- | --- |
+| 001 | TASK-01, TASK-12 | MAP-004, FLOW-001 | N/A | one static step only; deterministic rejection otherwise |
+| 002-004 | TASK-00, TASK-01, TASK-11, TASK-14, TASK-15 | MODEL-001, MAP-001, H5-001, S4 E2E plus S4R common-path tests | `shell-s4` for reference; S4R N/A | S4/S4R one kernel; distinct stable source identity |
+| 005 | TASK-00, TASK-06, TASK-11 | MODEL-001, DOF-001/003, H5-001/002 | N/A | exact six-component order; no distributed equation ownership |
+| 006-010 | TASK-00, TASK-01 | MODEL-002, MAP-002/003 | N/A | finite E,nu,t; one assignment; unsupported meanings fail |
+| 011-016 | TASK-02, TASK-03 | GEO-001..004, KIN-002 | N/A | deterministic unit directors; exact geometry predicates |
+| 017-020 | TASK-01, TASK-08 | MAP-004, LOAD-001..004 | N/A | six-DOF BC/CLOAD; drilling/distributed load fails |
+| 021-023 | TASK-01 | MAP-001..004 and existing no-op regression | N/A | approved parser subset/identity wrappers/no-op policy |
+| 024-030 | TASK-00, TASK-06, TASK-07, TASK-09, TASK-10, TASK-12, TASK-16 | MODEL, DOF, ASM, STATE, REC, FLOW, VERIFY suites | N/A | ownership, deterministic assembly, lifecycle, residual reaction |
+| 031-038 | TASK-03, TASK-04, TASK-05, TASK-07, TASK-10 | KIN, KERNEL, PHYSREC, ASM-003, REC-003 | N/A | physical 5-DOF embedding, exact drilling, same quadrature, invariants |
+| 039-048 | TASK-09, TASK-10, TASK-11, TASK-12 | STATE-001..003, REC-001..005, H5-001..004, FLOW-004 | N/A | mandatory finite schema/results and atomic commit |
+| 049 | TASK-00 through TASK-16 | every production Task records RED/failure/GREEN/focused/full VERIFY; VERIFY-001..003 | `shell-s4` where reference-dependent | TDD evidence, related C++ tests, MSVC Debug no-warning build and full CTest |
+| 050 | TASK-02 through TASK-05, TASK-07 | GEO-001/002, KIN-001/002, KERNEL-001..004, ASM-002 | N/A | frames, Jacobian, symmetry, transformation, six modes, positivity and repeatability |
+| 051 | TASK-02 through TASK-05, TASK-10, TASK-12 | GEO-001, KERNEL-001..003, REC-004, FLOW-002 | N/A | exact normalized 1e-12 and 1e-10 algebraic thresholds |
+| 052 | TASK-03 through TASK-05, TASK-10 | KIN-003/004, KERNEL-004, PHYSREC-001, REC-001/002 | N/A | independent membrane, bending, shear, twist fields and recovery signs/order |
+| 053 | TASK-04, TASK-05, TASK-14 | KERNEL-001..007, PHYSREC-001, declared S4 E2E suite | `shell-s4` | formulation invariants/patches and the S4 reference case pass |
+| 054 | TASK-02 | GEO-001..004 | N/A | exact valid/rejected geometry inventory; no NR-O03/O04 |
+| 055 | TASK-16 | VERIFY scope audit | N/A | extra benchmark portfolio is explicitly nonblocking and absent from completion gate |
+| 056 | TASK-04, TASK-05, TASK-07, TASK-10 | KERNEL-005..007, PHYSREC-001, ASM-001/002, REC-003 | N/A | exact fixed drilling, deterministic rank/action and physical-output exclusion |
+| 057 | TASK-10, TASK-12, downstream Physics Evaluation | REC-002..004, FLOW-002, later physics report | `shell-s4` | implementation exposes equilibrium/sign/energy evidence; physical plausibility verdict is downstream |
+| 058-060 | TASK-13, TASK-14 | REF-003 and S4 E2E comparison | `shell-s4` | exact fixed absolute U tolerance `1.0e-5` without scale/clamp |
+| 061-062 | TASK-13, TASK-14 | REF-004/005 and S4 E2E test | `shell-s4` | UR fixed absolute `1.0e-5`, deterministic warning only |
+| 063-064 | TASK-13 | REF-001/002/005 | `shell-s4` | schema/row failure before numeric comparison; full metrics |
+| 065-068 | TASK-13, TASK-14, TASK-15 | REF-001/002/006, S4 case precheck and S4R non-consumption/common-path evidence | `shell-s4` | exact two S4 paths; unique finite mapped rows; no S4R artifact consumption |
+| 069-071 | TASK-14, TASK-15 | S4 E2E suite and S4R common-path tests | `shell-s4` | S4 U blocks/UR warns; S4R mapping has no Abaqus equality gate |
+| 072 | TASK-13 through TASK-16 | REF-006, VERIFY-004 and git/hash audit | `shell-s4` | no reference solver run or artifact mutation |
+
+## 10. Validation Commands
+
+These commands are the environment-resolved Windows x64 Debug baseline. Planning does
+not execute build/tests. Each approved Harness Step first runs this literal path
+precheck and then repeats the relevant build/test subset after its RED and GREEN
+changes.
+
+```powershell
+$requiredBuildPaths = @(
+  "C:/git/googletest",
+  "C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl",
+  "C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb",
+  "C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+)
+foreach ($requiredBuildPath in $requiredBuildPaths) {
+  if (-not (Test-Path -LiteralPath $requiredBuildPath)) {
+    throw "Required configured build path is absent: $requiredBuildPath"
+  }
+}
+
+cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 `
+  "-DFESA_GTEST_SOURCE_DIR=C:/git/googletest" `
+  "-DMKL_DIR=C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl" `
+  "-DTBB_DIR=C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb" `
+  "-DHDF5_DIR=C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+
+cmake --build .harness/build --config Debug
+ctest --test-dir .harness/build -C Debug --show-only=json-v1
+ctest --test-dir .harness/build -C Debug --output-on-failure
+```
+
+Focused commands:
+
+```powershell
+cmake --build .harness/build --config Debug --target fesa_unit_tests
+ctest --test-dir .harness/build -C Debug -R "DomainModel|InpDomainMapping|Mitc4Geometry|Mitc4Shell|DofManager|EssentialConstraints|SparseAssembly|LoadAssembly|ResultRecovery|Hdf5ResultsWriter" --output-on-failure
+
+cmake --build .harness/build --config Debug --target fesa_integration_tests
+ctest --test-dir .harness/build -C Debug -R "LinearStaticCli|Mitc4ShellCli" --output-on-failure
+
+cmake --build .harness/build --config Debug --target fesa_reference_tests
+ctest --test-dir .harness/build -C Debug -R "Mitc4ReferenceComparison|Mitc4S4Reference" --output-on-failure
+```
+
+Planning-document verification:
+
+```powershell
+git diff --check -- docs/linear-static-mitc4-shell/implementation-plan.md
+git status --short
+git diff --name-only
+```
+
+Reference immutability audit is read-only and fails on any mismatch:
+
+```powershell
+$expectedReferenceHashes = [ordered]@{
+  "reference/shell/shell.inp" = "4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF"
+  "reference/shell/shell displacements.csv" = "C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B"
+}
+foreach ($referencePath in $expectedReferenceHashes.Keys) {
+  $actualHash = (Get-FileHash -Algorithm SHA256 -LiteralPath $referencePath).Hash
+  if ($actualHash -ne $expectedReferenceHashes[$referencePath]) {
+    throw "Reference artifact changed: $referencePath"
+  }
+}
+```
+
+Do not invoke `scripts/hooks/*.py` manually. Harness runner installs/uses the configured
+PreToolUse and Stop hooks automatically. Do not invoke `scripts/execute.py` until the
+user separately authorizes execution.
+
+## 11. Risks and Downstream Handoff
+
+| risk | controlling plan decision | stop condition |
+| --- | --- | --- |
+| beam-only ModelDefinition leaks through every layer | add shell semantic records first; preserve B33 access | stop if a Step requires unapproved mixed-model hierarchy |
+| fixed 12-DOF scatter breaks B33 | typed beam/shell scatter paths and B33 regressions | stop on any B33 focused/full regression |
+| drilling contaminates physical output | separate physical recovery from stabilized full K | stop if pure-drill recovery/energy is nonzero |
+| director/frame sign nondeterminism | source-order, area-weight and tie-break tests | stop on repeat/order/thread difference |
+| S4R accidentally selects reduced integration | one internal formulation and ASM/E2E equality tests | stop if source type reaches quadrature dispatch |
+| full residual confused with element resultants | keep reaction from Kd-F, physical recovery separately | stop if reaction is reconstructed from shell resultants |
+| writer expansion changes B33 schema | additive shell branches plus full B33 CTest | stop on B33 schema regression |
+| comparator inherits B33 four-file/Frame assumptions | dedicated test-only MITC4 comparator | stop if administrative files become required |
+| reference artifacts staged by Harness | clean isolated worktree, hash/diff audit | stop immediately on any reference path change |
+| future nonlinear equations enter production | scope and code review prohibition | stop if nonlinear state/tangent is allocated or called |
+
+### Implementation Agent
+
+- Read the approved implementation plan, `docs/HARNESS.md` and
+  `docs/HARNESS_WORKFLOW.md` before every Step.
+- Execute only the Harness-selected pending Step and use the repository hooks/scripts
+  through the documented runner lifecycle.
+- Never choose the next Step, edit executor-owned timestamps/statuses, batch multiple
+  Steps, or implement outside the Step's file ownership.
+- Record the RED command and observed failure before GREEN.
+
+### Build/Test Executor Agent
+
+- Use Section 10 exact CMake/MSVC/CTest baseline after implementation Steps finish.
+- Report new warnings, discovery gaps and exact failing test names without changing
+  upstream contracts.
+
+### Correction Agent
+
+- Apply only minimal failure-driven corrections inside approved file ownership.
+- Do not reinterpret MITC4 mathematics, I/O schema or tolerance.
+
+### Reference Verification Agent
+
+- Consume authoritative HDF5 and declared read-only CSV directly.
+- Create the separate reference-verification report; only U affects pass/fail and every
+  UR warning remains visible.
+
+### Physics Evaluation and Release Agents
+
+- Physics evaluates equilibrium, directions, symmetry, energy and result signs after
+  reference verification.
+- Release readiness remains blocked until implementation, build/test, reference and
+  physics gates all provide evidence.
+
+## 12. Harness Step Draft
+
+This approved draft is materialized under `phases/linear-static-mitc4-shell/`.
+Every Step is zero-based, kebab-case and closes RED/GREEN/VERIFY before the
+Executor advances it. TASK-14 is declared-case sub-work inside Step 13's single
+test-only reference module; TASK-15 is non-reference coverage already owned by
+Steps 1/7/11; TASK-16 is downstream Build/Test Executor evidence,
+not an Implementation Agent phase Step. During materialization, the exact required-reading
+paths, configure block, focused command and full VERIFY block shown here are copied
+verbatim into every `stepN.md`; a materialized Step never refers back to this draft or
+to an external conversation.
+
+Every materialized Step copies this mandatory reading inventory before its
+Step-specific paths:
+
+- `AGENTS.md`
+- `docs/PRD.md`, `docs/ARCHITECTURE.md`, `docs/ADR.md`
+- `docs/HARNESS.md`, `docs/HARNESS_WORKFLOW.md`
+- this approved implementation plan and all source contracts listed in Metadata
+- `.codex/hooks.json`
+- `phases/index.json`, `phases/linear-static-mitc4-shell/index.json` and its own
+  `phases/linear-static-mitc4-shell/stepN.md`
+- every path created/modified by prerequisite Steps and their step output/status
+
+If any mandatory or Step-specific path is missing or contradicts the approved plan,
+the Implementation Agent records `blocked` for the Executor-selected current Step and
+stops; it does not invent the missing contract.
+
+Step-specific path ledger copied into the corresponding materialized Step:
+
+| Step | exact source/prerequisite paths |
+| ---: | --- |
+| 0 | `include/fesa/model/model_types.hpp`; `include/fesa/model/domain.hpp`; `src/fesa/model/domain.cpp`; `tests/unit/model/model_types_test.cpp`; `tests/unit/model/domain_test.cpp` |
+| 1 | Step 0 paths; `include/fesa/io/abaqus/domain_mapper.hpp`; `src/fesa/io/abaqus/domain_mapper.cpp`; `tests/unit/io/abaqus/domain_mapper_test.cpp` |
+| 2 | Step 0-1 paths; candidate new `include/fesa/model/shell_geometry.hpp`; `src/fesa/model/shell_geometry.cpp`; `tests/unit/model/shell_geometry_test.cpp`; `src/fesa/CMakeLists.txt`; `tests/CMakeLists.txt` |
+| 3 | Step 2 paths; candidate new `include/fesa/elements/mitc4_shell.hpp`; `src/fesa/elements/mitc4_shell.cpp`; `tests/unit/elements/mitc4_shell_test.cpp`; `src/fesa/CMakeLists.txt`; `tests/CMakeLists.txt` |
+| 4 | Step 3 MITC4 kernel/header/test paths |
+| 5 | Step 3-4 MITC4 kernel/header/test paths |
+| 6 | Step 0 paths; `include/fesa/fem/dof_manager.hpp`; `src/fesa/fem/dof_manager.cpp`; `tests/unit/fem/dof_manager_test.cpp`; `tests/unit/constraints/essential_constraints_test.cpp` |
+| 7 | Steps 4/6 paths; `include/fesa/assembly/sparse_assembler.hpp`; `src/fesa/assembly/sparse_assembler.cpp`; `tests/unit/assembly/sparse_assembler_test.cpp` |
+| 8 | Steps 2/7 paths; `include/fesa/assembly/load_assembler.hpp`; `src/fesa/assembly/load_assembler.cpp`; `tests/unit/assembly/load_assembler_test.cpp` |
+| 9 | Step 5 paths; `include/fesa/results/result_records.hpp`; `include/fesa/analysis/analysis_state.hpp`; `src/fesa/analysis/analysis_state.cpp`; `tests/unit/results/result_records_test.cpp` |
+| 10 | Steps 5/9 paths; `include/fesa/results/result_recovery.hpp`; `src/fesa/results/result_recovery.cpp`; `tests/unit/results/result_recovery_test.cpp` |
+| 11 | Steps 9/10 paths; `include/fesa/io/hdf5/hdf5_results_writer.hpp`; `src/fesa/io/hdf5/hdf5_results_writer.cpp`; `tests/unit/io/hdf5/hdf5_results_writer_test.cpp` |
+| 12 | Steps 7-11 paths; `include/fesa/analysis/linear_static_analysis.hpp`; `src/fesa/analysis/linear_static_analysis.cpp`; `tests/integration/analysis/linear_static_analysis_test.cpp`; `tests/integration/app/fesa_application_test.cpp` |
+| 13 | Steps 11/12 paths; candidate new `tests/reference/mitc4_reference_comparison.hpp`, `tests/reference/mitc4_reference_comparison.cpp`, `tests/reference/mitc4_reference_comparison_test.cpp`, `tests/reference/mitc4_reference_cases_test.cpp`; `tests/CMakeLists.txt`; the two exact read-only S4 reference paths in Section 2 |
+
+The configure command copied before each Step's RED build is:
+
+```powershell
+$requiredBuildPaths = @(
+  "C:/git/googletest",
+  "C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl",
+  "C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb",
+  "C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+)
+foreach ($requiredBuildPath in $requiredBuildPaths) {
+  if (-not (Test-Path -LiteralPath $requiredBuildPath)) {
+    throw "Required configured build path is absent: $requiredBuildPath"
+  }
+}
+
+cmake --fresh -S . -B .harness/build -G "Visual Studio 18 2026" -A x64 `
+  "-DFESA_GTEST_SOURCE_DIR=C:/git/googletest" `
+  "-DMKL_DIR=C:/Program Files (x86)/Intel/oneAPI/mkl/2026.1/lib/cmake/mkl" `
+  "-DTBB_DIR=C:/Program Files (x86)/Intel/oneAPI/tbb/2023.1/lib/cmake/tbb" `
+  "-DHDF5_DIR=C:/Program Files/HDF_Group/HDF5/2.1.1/cmake"
+```
+
+The full VERIFY block copied at the end of every Step is:
+
+```powershell
+cmake --build .harness/build --config Debug
+ctest --test-dir .harness/build -C Debug --show-only=json-v1
+ctest --test-dir .harness/build -C Debug --output-on-failure
+```
+
+### Step 0 — `shell-semantic-model`
+
+- Required reading: requirements 002-016/024, I/O Sections 3/5, architecture model
+  ownership, current `model_types.hpp`/`domain.*` and their tests.
+- Prerequisite: approved plan only; no production dependency on a prior Step.
+- RED: add MODEL-001/002 tests and CMake registration if needed; run:
+  `cmake --build .harness/build --config Debug --target fesa_unit_tests` then
+  `ctest --test-dir .harness/build -C Debug -R "DomainModel" --output-on-failure`.
+  Record the missing shell type/accessor failure. Unexpected pass is a stop.
+- GREEN: minimally add shell semantic records/Domain const access; no parser/kernel.
+- VERIFY: rerun the focused commands and common full VERIFY.
+- Prohibitions: no equation IDs, element hierarchy, mixed-model execution, parser edits.
+
+### Step 1 — `shell-domain-mapping`
+
+- Required reading: Step 0 outputs, requirements 001-010/017-023, I/O Sections 1-3/8-9,
+  current `domain_mapper.*` and parser/model tests.
+- Prerequisite: Step 0 `completed`.
+- RED: add MAP-001..004; build `fesa_unit_tests` and run
+  `ctest --test-dir .harness/build -C Debug -R "InpDomainMapping" --output-on-failure`.
+  Record valid S4/S4R rejection and exact negative diagnostic failures.
+- GREEN: minimally extend semantic mapping; keep `input_reader` syntax-only.
+- VERIFY: focused parser/model tests, then common full VERIFY.
+- Prohibitions: no element math, no silent unsupported keyword, no reference edits.
+
+### Step 2 — `shell-director-geometry`
+
+- Required reading: requirements 011-016, formulation Sections 4/9/17, Numerical Review
+  5.2/5.5/6.1, I/O 3.3/9, completed semantic model/mapping.
+- Prerequisite: Step 1 `completed`.
+- RED: add GEO-001..004; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "Mitc4Geometry" --output-on-failure`.
+- GREEN: implement deterministic director/frame and exact geometry inventory validation.
+- VERIFY: focused geometry/mapping tests, then common full VERIFY.
+- Prohibitions: no calibrated angle/warp/distortion threshold; no NR-O03/O04.
+
+### Step 3 — `mitc4-kinematics-constitutive`
+
+- Required reading: formulation Sections 3-11/17, Numerical Review 5.1-5.6/6.1,
+  current math adapters and Euler kernel style.
+- Prerequisite: Step 2 `completed`.
+- RED: add KIN-001..005 in new `mitc4_shell_test.cpp`, register it, build unit target,
+  run `ctest --test-dir .harness/build -C Debug -R "Mitc4ShellKinematics|Mitc4ShellConstitutive" --output-on-failure`.
+- GREEN: minimally add concrete shell kernel kinematics/constitutive seams; no global
+  assembly or drilling.
+- VERIFY: focused kernel tests, then common full VERIFY.
+- Prohibitions: no public base hierarchy, no S4R branch, no nonlinear tangent code.
+
+### Step 4 — `mitc4-stiffness-drilling`
+
+- Required reading: formulation Sections 10-14/17, requirements 031-038/050-056,
+  Numerical Review 5.7/6.1-6.2, completed Step 3 kernel.
+- Prerequisite: Step 3 `completed`.
+- RED: add KERNEL-001..006; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "Mitc4ShellKernel|Mitc4ShellPatch|Mitc4ShellDrilling" --output-on-failure`.
+- GREEN: implement K20, 20-to-24 congruence and exact fixed drilling only.
+- VERIFY: focused kernel suite, then common full VERIFY.
+- Prohibitions: translations in R+, coefficient sweep, recovery/result output, future nonlinear.
+
+### Step 5 — `mitc4-physical-recovery`
+
+- Required reading: formulation Sections 14/16, requirements 035/042-046/052/056,
+  I/O result component/location order, completed Step 4 kernel.
+- Prerequisite: Step 4 `completed`.
+- RED: add KERNEL-007 and PHYSREC-001; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "Mitc4ShellDrilling|Mitc4ShellPhysicalRecovery" --output-on-failure`.
+- GREEN: minimally add kernel-local physical recovery and physical energy API only.
+- VERIFY: focused physical recovery/drilling tests, then common full VERIFY.
+- Prohibitions: no AnalysisState/global result rows, no drilling contribution or HDF5.
+
+### Step 6 — `shell-dof-scatter`
+
+- Required reading: requirements 005/025, architecture DofManager ownership, current
+  `dof_manager.*` and constraint tests, completed shell model.
+- Prerequisite: Step 5 `completed`.
+- RED: add DOF-001..003; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "DofManager|EssentialConstraints" --output-on-failure`.
+- GREEN: minimally add typed 24-entry scatter/pattern while preserving 12-entry B33.
+- VERIFY: focused DOF/constraint tests, then common full VERIFY.
+- Prohibitions: no equation IDs in model, no constraint/load ownership move.
+
+### Step 7 — `shell-sparse-assembly`
+
+- Required reading: requirements 027/030/037, architecture deterministic COO rule,
+  current `sparse_assembler.cpp`/`sparse_matrix.cpp` tests, completed kernel/scatter.
+- Prerequisite: Step 6 `completed`.
+- RED: add ASM-001..003; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "SparseAssembly" --output-on-failure`.
+- GREEN: minimally dispatch shell and emit stable 576-entry local buffers.
+- VERIFY: focused assembly tests, then common full VERIFY.
+- Prohibitions: no worker global CSR mutation, unordered reduction or S4R integration branch.
+
+### Step 8 — `shell-load-validation`
+
+- Required reading: requirements 017-020/028, formulation 8/13, I/O 4.2-4.3,
+  current `load_assembler.*` and constraint lifecycle.
+- Prerequisite: Step 7 `completed`.
+- RED: add LOAD-001..004; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "LoadAssembly|EssentialConstraints" --output-on-failure`.
+- GREEN: minimally add aggregate nodal moment/director validation and reuse full DOF load.
+- VERIFY: focused load/constraint tests, then common full VERIFY.
+- Prohibitions: no distributed/equivalent/follower load; no drill load channel.
+
+### Step 9 — `shell-analysis-state`
+
+- Required reading: requirements 026/041-046/048, I/O 6.3-6.5,
+  current `analysis_state.*`/`result_records.hpp` and completed physical recovery types.
+- Prerequisite: Step 8 `completed`.
+- RED: add STATE-001..003; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "AnalysisState" --output-on-failure`.
+- GREEN: minimally add candidate-owned shell row/global evidence containers and
+  validation/commit mechanics.
+- VERIFY: focused AnalysisState tests, then common full VERIFY.
+- Prohibitions: no element calculation, ResultRecovery orchestration or HDF5 writing.
+
+### Step 10 — `shell-result-recovery`
+
+- Required reading: requirements 029/035/041-046/048, formulation 14/16, I/O 6.3-6.5,
+  current `result_recovery.*` and completed Steps 5/9.
+- Prerequisite: Step 9 `completed`.
+- RED: add REC-001..005; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "ResultRecovery" --output-on-failure`.
+- GREEN: minimally orchestrate full-residual evidence and physical shell recovery into
+  a fully validated candidate, then commit it.
+- VERIFY: focused recovery tests, then common full VERIFY.
+- Prohibitions: no new result record type, location averaging, drill recovery or writer edit.
+
+### Step 11 — `shell-hdf5-output`
+
+- Required reading: requirements 039-048, I/O Section 6 exact schema, ADR atomic output,
+  current HDF5 writer/self-check tests, completed recovery records.
+- Prerequisite: Step 10 `completed`.
+- RED: add H5-001..004; build unit target and run
+  `ctest --test-dir .harness/build -C Debug -R "Hdf5ResultsWriter" --output-on-failure`.
+- GREEN: add exact additive shell schema behind unchanged ResultsWriter boundary.
+- VERIFY: focused HDF5 tests, then common full VERIFY.
+- Prohibitions: no B33 schema reinterpretation, CSV solver output, partial final file.
+
+### Step 12 — `shell-linear-static-flow`
+
+- Required reading: requirements 024-030, architecture eight-hook lifecycle, I/O CLI
+  contract, current `linear_static_analysis.*`/application integration tests.
+- Prerequisite: Step 11 `completed`.
+- RED: add FLOW-001..004; build integration target and run
+  `ctest --test-dir .harness/build -C Debug -R "LinearStaticCli|Mitc4ShellCli" --output-on-failure`.
+- GREEN: minimally route shell through existing lifecycle; preserve one factorization.
+- VERIFY: focused integration tests, then common full VERIFY.
+- Prohibitions: no analysis lifecycle reorder, no 0x0 singular conversion, no early state commit.
+
+### Step 13 — `shell-reference-comparison`
+
+- Required reading: requirements 058-072, I/O Section 7, Reference Model contract,
+  the sole declared S4 paths and hashes, and current B33 comparator only as reusable
+  identity/report precedent.
+- Scope ownership: this one test-only reference module owns TASK-13 comparator behavior
+  plus TASK-14 declared-case tests; it owns no solver production module.
+- Prerequisite: Step 12 `completed` and valid MITC4 HDF5 fixture.
+- RED: update REF-003 first so the existing scale-dependent comparator fails against
+  literal fixed-boundary cases around `1.0e-5`; build the reference target and run
+  `ctest --test-dir .harness/build -C Debug -R "Mitc4ReferenceComparison|Mitc4S4Reference" --output-on-failure`.
+  Record the expected old-tolerance assertion failure.
+- GREEN: minimally replace only the MITC4 comparator decision value with fixed absolute
+  `1.0e-5`; keep reference scale as diagnostic report data only.
+- If a declared case exposes a production defect outside this reference-test module,
+  stop the Step and route a focused correction to the owning prior module; do not patch
+  unrelated production layers inside this Step.
+- VERIFY: focused comparator and the declared S4 tests, then common full VERIFY.
+- Prohibitions: no reference artifact writes/Abaqus run, no metadata/README gate, no UR
+  blocking, no B33 comparator change, and no scale/clamp substitution for fixed `1.0e-5`.
+
+After explicit approval, planning may create only:
+
+- `phases/index.json` with task status `pending` and no creation timestamp;
+- `phases/linear-static-mitc4-shell/index.json` with Steps 0-13 initially `pending` and
+  no executor-owned timestamps;
+- `phases/linear-static-mitc4-shell/stepN.md` containing the approved self-contained
+  directions.
+
+Approval does not authorize `python scripts/execute.py linear-static-mitc4-shell` or
+`--push`.
+
+## 13. Open Issues
+
+| id | item | blocking now | resolution owner |
+| --- | --- | --- | --- |
+| OI-001 | Section 12 multi-Step Harness draft was explicitly approved on 2026-08-12 and materialized. | resolved | user |
+| OI-002 | Candidate shell semantic representation and exact C++ value/reference types are not public API decisions. | no | owning RED test in Steps 0/3 |
+| OI-003 | Per-test MITC4 CTest label can be added without relabeling B33 tests인지 확인한다. Exact suite regex remains sufficient. | no | Step 0 and downstream TASK-14 CMake verification |
+
+No mathematical, I/O, reference-inventory or tolerance decision remains open for
+implementation planning. The user separately authorized implementation and Harness
+execution on 2026-08-12; the Executor still owns branch, Step selection, timestamps,
+commits and advancement.

docs/linear-static-mitc4-shell/io.md

@@ -3,17 +3,17 @@
 ## Metadata
 
 - feature_id: `linear-static-mitc4-shell`
-- source_requirement: `docs/requirements/linear-static-mitc4-shell.md`
-- source_research: `docs/research/linear-static-mitc4-shell-research.md`
-- source_formulation: `docs/formulations/mitc4-shell-formulation.md`
-- source_numerical_review: `docs/numerical-reviews/linear-static-mitc4-shell-review.md`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_research: `docs/linear-static-mitc4-shell/research.md`
+- source_formulation: `docs/linear-static-mitc4-shell/formulation.md`
+- source_numerical_review: `docs/linear-static-mitc4-shell/numerical-review.md`
 - source_commits: requirements/research/formulation policy revision `73df844`
 - status: `approved-for-implementation-planning`
 - owner_agent: `io-definition-agent`
-- date: `2026-08-12`
+- date: `2026-08-13`
 - authoritative_output: `results.h5`
 - hdf5_schema_version: `0`
-- reference_cases: read-only S4 at `reference/shell/`; read-only S4R at `reference/shellR/`
+- reference_cases: sole acceptance case is read-only full-integration S4 at `reference/shell/`; S4R is covered by non-reference source-mapping/kernel/metadata tests
 - implementation_planning_authorized: `true`
 
 This document defines the approved semantic input and output contract for the
@@ -23,19 +23,21 @@ comparison-tool C++ APIs and does not claim full Abaqus compatibility or Abaqus
 
 The user approved the following I/O-specific decisions on `2026-08-12`:
 
-1. `S4` and `S4R` reference artifacts remain separate even though both source types
-   map to one internal `FESA-MITC4` formulation.
+1. `S4` and `S4R` source types map to one internal full-integration `FESA-MITC4`
+   formulation, while only S4 is an Abaqus acceptance reference.
 2. The optional positive integer on a `*SHELL SECTION` thickness row is accepted
    without a warning and has no semantic effect on FESA quadrature. FESA always uses
    the formulation's `2 x 2 x 2` rule.
-3. The existing `reference/shell/` S4 and `reference/shellR/` S4R input/displacement
-   files are retained under their current names without mutation. No canonical or
-   legacy-alias classification is required.
+3. The existing `reference/shell/` S4 input/displacement files are the sole declared
+   acceptance pair and remain under their current names without mutation. Existing
+   `reference/shellR/` files are untouched optional evidence and are not consumed by
+   acceptance comparison. No canonical or legacy-alias classification is required.
 4. HDF5 schema version `0` is extended additively with MITC4-specific datasets while
    retaining the existing common metadata, nodal-result, diagnostic, CLI, and
    failure-atomicity conventions.
 
-The approved requirements fix the drilling rule and B33 mixed displacement tolerance.
+The approved requirements fix the drilling rule and the project-wide common
+family-scale reference tolerance.
 Drilling calibration/output, `NR-O03`, `NR-O04`, bundle-administration metadata, and
 an expanded reference portfolio are outside this contract.
 
@@ -440,15 +442,12 @@ Authoritative comparison reads `results.h5` and read-only Abaqus CSV artifacts.
 A deterministic FESA CSV projection is a debugging/review view only; it is neither
 solver output nor a reference artifact.
 
-### 7.1 S4/S4R case separation
+### 7.1 Full-integration reference boundary
 
-- S4 and S4R use their existing distinct input and displacement CSV paths.
-- FESA maps both source types to `FESA-MITC4`, but Abaqus S4 and S4R values are not
-  expected to equal each other on a finite mesh.
-- Reference comparison always compares one FESA run with Abaqus rows generated from
-  the same source model/type.
+- FESA maps S4 and S4R source types to the same full `2 x 2 x 2` FESA-MITC4 path.
+- Reference comparison consumes only the declared S4 input/displacement pair.
 - Equality of FESA internal numerical rows after changing only source type is a
-  separate implementation test, not an Abaqus S4-versus-S4R comparison.
+  separate implementation test; it does not consume or compare Abaqus S4R rows.
 
 ### 7.2 Declared S4 case
 
@@ -463,23 +462,18 @@ The files must not be renamed, rewritten, zero-clamped, normalized, or repaired.
 The reaction and stress CSV files in the same directory are optional review evidence
 and are not comparison inputs.
 
-### 7.3 Declared S4R case
+### 7.3 S4R non-reference coverage
 
-The blocking S4R case uses:
-
-| role | exact path | SHA-256 | observed inventory |
-| --- | --- | --- | --- |
-| input | `reference/shellR/shellR.inp` | `1325940FB42B78961CF25E84379BF2693846FAD22473E7688AC5456B37B18CB4` | `TYPE=S4R`; declared comparison input |
-| displacement | `reference/shellR/shellR displacements.csv` | `8887ACC5ED007CB97583A9FDC1150B48B9297E269A5BA8EBA6C1A5F6306E98CB` | 49 finite data rows; U and UR columns |
-
-These files follow the same immutability rule. The reaction and stress CSV files in
-the directory are optional review evidence and are not comparison inputs. No README,
+No file under `reference/shellR/` is required or consumed by this acceptance
+comparison. S4R source support is verified by parser mapping, S4/S4R common-kernel
+and deterministic-assembly equivalence, plus preserved HDF5 source-type metadata.
+Any existing S4R artifacts remain immutable optional inspection evidence. No README,
 `metadata.json`, canonical name, legacy alias, provenance record, or duplicated model
-description is required for either case.
+description is required.
 
 ### 7.4 Displacement header mapping
 
-After trimming header whitespace, both declared displacement CSVs use:
+After trimming header whitespace, the declared S4 displacement CSV uses:
 
 | CSV column | comparison field/component | FESA HDF5 source |
 | --- | --- | --- |
@@ -521,23 +515,31 @@ Before tolerance evaluation:
 Only `U1/U2/U3` affect pass/fail. `UR1/UR2/UR3` are always compared and reported but
 can emit only an approved deterministic nonblocking warning.
 
-For each model, step/frame, quantity, and component group:
+Matched U rows form one blocking translation family and matched UR rows form one warning-only
+rotation family. For each family:
 
 ```text
-reference_scale_c = max(abs(reference_value_i))
-row_tolerance_c = 1e-9 + 1e-6 * reference_scale_c
-row_pass = abs(fesa_value-reference_value) <= row_tolerance_c
+S = max(abs(reference_value_i))
+error_i = abs(fesa_value_i-reference_value_i)
+near_zero_band = 0.01*S
+
+if abs(reference_value_i) <= near_zero_band:
+    row_pass = error_i <= near_zero_band
+else:
+    row_pass = error_i/abs(reference_value_i) <= 0.05
+
+relative_rms = sqrt(mean(error_i^2))/S
+family_pass = all(row_pass) and relative_rms <= 0.01
 ```
 
-Reference scale uses finite Abaqus values only. A zero scale makes the relative term
-zero. No reference or result value is zero-clamped and no row-specific relative
-denominator replaces the component scale. The formula is the approved B33 rule:
-the `1e-9` floor is in the user-consistent length unit for U and dimensionless for UR.
-U exceedance fails; UR exceedance emits a deterministic warning only.
+Reference values alone define `S`; values are not zero-clamped or omitted and there is no
+independent absolute-error gate. A zero-scale family passes only if every FESA value is exactly
+zero, otherwise it fails without emitting NaN/Inf. U row or RMS exceedance fails; the same UR
+exceedance emits a deterministic warning only.
 
-The report records every U/UR row, blocking/nonblocking decision, absolute error,
-component-scale normalized error, RMS error, displacement/rotation vector-norm
-error, worst source row/component, and every UR warning.
+The report records every U/UR row, blocking/nonblocking decision, family scale,
+near-zero branch, absolute and applicable row-relative error, scale-relative RMS,
+worst source row/component, and every UR warning.
 
 ## 8. CLI and Diagnostics Contract
 
@@ -604,7 +606,7 @@ warning behavior.
 | HDF5 model | exact paths, dtypes, shapes, IDs, source types, directors, sections, and finite values | output, exit 6 |
 | HDF5 results | every mandatory row/location/component exists in stable order and is finite | output, exit 6 |
 | S4 case | exact declared input/displacement paths and unique finite U/UR rows | reference precheck |
-| S4R case | exact declared input/displacement paths and unique finite U/UR rows | reference precheck |
+| S4R source support | parser/common-kernel/deterministic-assembly/HDF5 metadata tests; no Abaqus artifact consumption | implementation verification |
 | comparison | exact normalized row-set; U blocking; UR warning-only; approved tolerance | reference verification |
 
 No successful parser/HDF5 implementation, numerical solution, reference comparison,
@@ -622,24 +624,24 @@ physics review, or release status follows from approval of this document alone.
 | `031-038` | source-independent MITC4 identity, fixed drilling stabilization and full-integration policy; no drilling output | Implementation Planning tests |
 | `039-048` | additive HDF5 v0 paths, mandatory quantities, location identity, atomic output | Reference Model and Implementation Planning |
 | `049-057` | diagnostic/schema hooks and required verification-metric/physical-energy evidence | Numerical Review and planning |
-| `058-064` | normalized U/UR rows, exact B33 tolerance, blocking/warning behavior, report inventory | Reference Verification |
-| `065-072` | exact current S4/S4R paths, immutability and displacement-only gate | Reference Model |
+| `058-064` | common-policy U/UR families, blocking/warning behavior, report inventory | Reference Verification |
+| `065-072` | exact current S4 paths, S4R reference non-consumption, immutability and displacement-only gate | Reference Model |
 
 ## 11. Open Issues and Downstream Handoff
 
 ### 11.1 Numerical Review boundary
 
 No I/O-owned calibration value remains open. Numerical Review shall verify the exact
-fixed drilling rule, basic geometry predicates, required HDF5 inventory, and B33
-tolerance mapping. Drilling calibration/energy output, `NR-O03`, `NR-O04`, bundle
+fixed drilling rule, basic geometry predicates, required HDF5 inventory, and common
+family-scale row/RMS tolerance mapping. Drilling calibration/energy output, `NR-O03`, `NR-O04`, bundle
 administration and reference-portfolio expansion are removed scope.
 
 ### 11.2 Reference Model Agent
 
-- Write `docs/reference-models/linear-static-mitc4-shell-reference-models.md` using
+- Write `docs/linear-static-mitc4-shell/reference-model.md` using
   this exact keyword/HDF5/reference-row contract.
 - Record only the two exact existing input/displacement pairs, comparison components,
-  HDF5 projection, source-row identity, B33 tolerance and immutability rule.
+  HDF5 projection, source-row identity, common family-scale tolerance and immutability rule.
 - Treat reaction/stress artifacts as nonblocking review evidence and do not create
   location-equivalence claims absent from this contract.
 

docs/linear-static-mitc4-shell/numerical-review.md

@@ -0,0 +1,367 @@
+# Linear Static MITC4 Shell Numerical Review
+
+## 1. Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- source_formulation: `docs/linear-static-mitc4-shell/formulation.md`
+- source_requirements: `docs/linear-static-mitc4-shell/requirements.md`
+- source_research: `docs/linear-static-mitc4-shell/research.md`
+- source_io_definition: `docs/linear-static-mitc4-shell/io.md`
+- source_reference_inventory: `docs/linear-static-mitc4-shell/reference-model.md`
+- repository_policy: `AGENTS.md`, `docs/SOLVER_AGENT_DESIGN.md`,
+  `.codex/skills/fesa-numerical-review/SKILL.md`
+- reviewed_head: `cf769aa` (`mathematical implementation baseline`)
+- prior_pass_commit: `60b42f4` (`context-only; verdict not inherited`)
+- status: `pass-for-implementation-planning`
+- owner_agent: `numerical-review-agent`
+- date: `2026-08-13`
+- implementation_planning_authorized: `true`
+- implementation_complete: `false`
+- build_test_complete: `false`
+- reference_comparison_complete: `false`
+- physics_evaluation_complete: `false`
+- release_ready: `false`
+
+이번 재검토는 현재 HEAD의 요구조건, 연구, 정식화, I/O 및 reference-case 계약을
+처음부터 상호 대조했다. 기존 review의 판정과 artifact 관찰 결과는 결론의 전제로
+사용하지 않았고, 이전 finding은 현 문서의 수식으로 다시 검산한 뒤 disposition만
+기록했다. 원 MITC4 local paper는 tying 위치와 covariant shear 보간을 확인하는 데
+read-only로 사용했다.
+
+이 단계에서는 Abaqus, Harness, C++ build/test, FESA 실행 및 reference comparison을
+수행하지 않았다. Reference artifact를 생성, 수정, 복원 또는 정규화하지 않았다.
+
+## 2. Review Verdict
+
+- verdict: `pass-for-implementation-planning`
+- critical_blockers: `none`
+- confirmed_defects: `none in the approved current linear-static scope`
+- open_blocking_questions: `none`
+- reason: 현재 정식화는 24 global DOF와 20 physical DOF의 관계, MITC4 shear
+  tying, plane-stress section law, 공통 `2 x 2 x 2` quadrature, residual/stiffness,
+  고정 drilling 안정화, 물리 recovery 및 검증 불변식을 구현계획으로 옮길 수 있을
+  만큼 명시한다. 요구조건, I/O 및 reference 계약과 모순되는 차원, 부호, 위치 또는
+  pass/fail 의미도 발견되지 않았다.
+- downstream_boundary: 이 판정은 Implementation Planning 진입만 허용한다. 구현,
+  MSVC build/CTest, reference comparison, physics sanity 또는 release를 승인하지 않는다.
+
+정식화 Section 15의 geometrically nonlinear residual/tangent는 future-only다. 완전한
+`Phi: R24 -> R20`, map Hessian, objective drilling potential 및 finite-rotation load work가
+미정인 사실은 미래 nonlinear 구현을 막지만 현재 linear-static 판정은 막지 않는다.
+
+## 3. Critical Findings
+
+### 3.1 Confirmed defects
+
+현재 승인된 선형 정적 범위에서 구현계획 전에 Formulation 또는 Research로 돌려보낼
+confirmed mathematical defect는 없다.
+
+`K20`의 exact-arithmetic 대칭/positive-semidefinite 구조와 20-to-24 congruence는
+일관된다. 다만 실제 구현의 rank, rigid action, patch field와 reference error는 문서
+검토만으로 통과했다고 볼 수 없으며 Section 6의 downstream test evidence가 필요하다.
+
+### 3.2 Previous finding disposition
+
+| previous item | rerun disposition | current independent basis |
+| --- | --- | --- |
+| `NR-C01` Jacobian/geometry inventory | `resolved` | Formulation 9.2-9.3은 center, eight stiffness points, four tying points 및 committed recovery points를 공통 fail-closed inventory로 두고 finite bases, nonzero surface measure와 `J>0`를 요구한다. 승인 범위는 calibrated distortion/warp cutoff를 요구하지 않는다. |
+| `NR-C02` drilling normalization | `resolved` | Formulation 12.2는 `R+`를 오직 8 physical tangent-rotation diagonals의 finite positive 값으로 제한하므로 모든 후보의 단위가 `force*length`로 같다. |
+| `NR-C03` mixed-DOF spectrum scaling | `resolved` | Formulation 12.5의 `(L_e I3,I2)` 및 `(L_e I3,I3)` congruence는 rank/condition evidence에서 translation/rotation 단위 혼합을 제거한다. Raw mixed-unit spectrum은 금지된다. |
+| `NR-C04` 20/24 weak-form mismatch | `resolved` | Formulation 5.2-5.3과 7.1-7.2는 physical, drilling, external work를 모두 `V24`에서 `T_p^T`와 `T_d^T`로 결합한다. |
+| `NR-C05` nonlinear 20-to-24 closure | `resolved for current scope` | Section 15는 physical chart tangent와 conditional global pullback을 분리하고 map-curvature 항을 보존하며, 미정인 global map/objective drill을 future-only blocker로 명시한다. |
+| `NR-D01` drilling-direction moment | `retained and consistent` | Exact-zero moment는 별도 처리하고 nonzero moment에 `rho_M=abs(d dot M)/norm(M)<=1e-12`를 적용한다. Numerical drilling은 거부된 moment를 운반하지 않는다. |
+| `NR-D02` normalized algebraic checks | `retained and consistent` | `1e-12` symmetry/frame/energy와 `1e-10` rigid/residual/equilibrium 기준은 scaled matrices와 unclamped denominators에 적용된다. |
+| `NR-O01` coefficient sweep/plateau | `closed by product decision` | `k_d=1e-3 min(R+)`가 고정 계약이다. Sweep, plateau 및 coefficient optimality는 구현 gate가 아니다. |
+| `NR-O02` drilling-energy ratio | `removed from scope` | Drilling energy는 내부 quadratic identity일 뿐 physical energy나 mandatory output이 아니며 ratio/warning threshold도 요구하지 않는다. |
+| `NR-O03` smooth-director calibration | `removed from scope` | Pairwise positive incident-normal orientation, finite/nonzero averaging 및 duplicate-node fold modeling이 승인된 exact predicate다. 별도 angle calibration은 gate가 아니다. |
+| `NR-O04` distortion/warp calibration | `removed from scope` | Basic topology, finite/nonzero surface measure 및 required-point `J>0`가 승인된 predicate다. Quality sweep이나 cutoff는 gate가 아니다. |
+| `NR-O05` U/UR tolerance | `resolved` | 모든 관련 문서가 sole S4 reference에 공통 family-scale row/RMS 규칙, U blocking, UR warning-only를 동일하게 정의한다. S4R은 reference gate가 아닌 common-path evidence다. |
+
+이전의 `needs-reference-model` 판정에 포함됐던 canonical naming, README,
+`metadata.json`, provenance, expanded portfolio 및 아직 없는 comparison result는 현재
+프로젝트 정책상 formulation verdict의 blocker가 아니다. 현 Reference Model 문서는
+정확한 기존 input/displacement path와 row/tolerance 계약을 제공한다.
+
+### 3.3 Open questions
+
+- current_linear_scope: `none blocking`
+- future_geometric_nonlinearity: finite global rotation coordinate, `Phi`와 그 1/2차
+  미분, chart recentering, objective drilling, nodal-moment work 및 nonlinear output/state
+  계약이 미정이다. 이는 별도 future formulation/review가 소유한다.
+- optional_characterization: near-singular positive-J geometry의 conditioning과 original
+  MITC4의 distorted-curved membrane locking을 더 넓게 정량화할 수 있으나 현재 승인된
+  planning/completion gate는 아니다.
+- downstream_results: implementation rank/patch evidence, S4 reference comparison과
+  S4R common-path test 결과는 해당 후속 Agent가 판정한다. 부재 자체는
+  pre-implementation review의 결함이 아니다.
+
+## 4. Numerical Risk Assessment
+
+| risk label | assessment | required in-scope control |
+| --- | --- | --- |
+| `rigid_body_modes` | Physical `K20`은 six rigid modes와 expected rank 14를 가져야 한다. 24-DOF embedding은 네 drill null coordinates를 더하고 fixed drill block 뒤 expected rank 18/nullity 6이다. | 세 translation과 세 rotation을 명시적으로 구성한다. Rotation mode는 `u_I=omega x X_I`, `theta_I=omega-(omega dot d_I)d_I`, `gamma_I=0`를 사용한다. |
+| `patch_test` | Bilinear membrane/bending field와 MITC tied shear는 required patch states를 표현할 계약을 갖는다. | `E11/E22/G12`, `K11/K22/K12`, `G13/G23`를 독립 시험하고 signs/component order/resultants/stress를 함께 확인한다. |
+| `symmetry` | `B^T C B`, `T_p^T K20 T_p`, `T_d^T(k_d I)T_d`는 exact arithmetic에서 symmetric이다. | Scaled Frobenius check `<=1e-12`; deterministic assembly가 대칭을 깨지 않는지 확인한다. |
+| `positive_definiteness` | Free element는 six-mode semidefinite이고, 충분히 구속된 nonsingular `Kff`는 positive definite가 기대된다. Geometry 또는 supports가 부적절하면 singularity가 정당하다. | Scaled spectrum/rank, non-rigid positive physical energy, constrained solve 및 singular negative cases를 분리한다. |
+| `hourglass` | `1 x 1` reduced integration을 쓰지 않으므로 Abaqus-style hourglass path는 `N/A`다. | Full `2 x 2` midsurface rank test는 유지한다. S4R source label로 reduced rule을 선택하지 않는다. |
+| `shear_locking` | Edge-midpoint MITC projection이 transverse-shear locking을 다루지만 모든 mesh/thickness에서 완전 제거를 주장할 수 없다. | Required shear/bending patch와 declared S4 reference를 통과한다. Broader thin/thick convergence는 nonblocking characterization이다. |
+| `membrane_locking` | Original MITC4는 membrane strain을 수정하지 않아 distorted curved meshes에서 알려진 locking 위험이 남는다. | Known limitation을 유지하고 MITC4+ 성능을 주장하지 않는다. Expanded curved/distorted portfolio는 optional이다. |
+| `volumetric_locking` | 승인된 homogeneous plane-stress shell에는 `N/A`다. | `C5`를 3D nearly-incompressible law로 확장하지 않는다. |
+| `distortion` | 양의 `J`를 유지하는 심한 distortion/warpage는 정확도와 rank/conditioning을 악화할 수 있다. | 모든 required location의 exact predicates와 rank/finite-result checks를 시행한다. 승인되지 않은 quality cutoff를 추가하지 않는다. |
+| `singular_jacobian` | Nonfinite/nonpositive `J`, zero surface measure, invalid reciprocal basis는 mapping을 무효화한다. | Center, stiffness, tying 및 recovery inventory 전체를 omission/clamp 없이 fail closed한다. |
+| `conditioning` | Thin shells, near-degenerate positive-J geometry 및 작은 fixed drill scale에서 `Kff` conditioning이 나빠질 수 있다. | Spectrum/condition evidence는 오직 length-scaled matrix를 사용한다. Threshold calibration은 gate가 아니지만 factorization failure는 결정적으로 진단한다. |
+| `convergence` | Current solve는 direct linear solve라 Newton convergence는 `N/A`; spatial convergence와 locking trend는 모델 의존이다. | Free residual/global equilibrium `<=1e-10`과 declared S4 reference case를 확인한다. Broader mesh sequences는 optional이다. |
+| `drilling_contamination` | Numerical drill이 physical strain/recovery에 들어가면 비물리 결과가 생긴다. | `T_p`/`T_d`를 분리하고 pure drill에서 physical strain/resultant/stress/energy가 zero임을 시험한다. Full residual은 의도대로 stabilized system 전체를 포함한다. |
+| `future_nonlinear_misuse` | Section 15만으로 global nonlinear element를 만들면 nonobjective drill 또는 inconsistent tangent가 된다. | Current plan에서 완전히 제외하고 별도 승인 전 실행하지 않는다. |
+
+## 5. Consistency Checks
+
+### 5.1 Units, dimensions, DOF order, and constrained/free system — `pass`
+
+- Per-node global order is exactly `[UX,UY,UZ,URX,URY,URZ]`; element order is 24
+  global coordinates and 20 physical coordinates
+  `[uX,uY,uZ,alpha,beta]` per node plus four separately selected `gamma` coordinates.
+- `T_p` is `20 x 24`, `T_d` is `4 x 24`, `K20` is `20 x 20`, and both global
+  stiffness contributions are `24 x 24`.
+- Translation-translation, translation-rotation, and rotation-rotation stiffness
+  blocks have units `force/length`, `force`, and `force*length`; `R+` therefore
+  excludes every translational diagonal.
+- The constrained/free equation is `Kff*df=Ff-Kfc*dc`. Stiffness partition and
+  factorization precede load assembly, and an all-constrained valid `0 x 0 Kff` is
+  not reclassified as singular.
+- `r=K*d-F` fixes the internal-minus-external sign. Constrained entries are the
+  required reaction rows and free entries remain residual evidence.
+
+### 5.2 Local/global transforms, congruence, and energy — `pass`
+
+- `(a_I,b_I,d_I)` and `(e1,e2,e3)` are deterministic right-handed orthonormal frames.
+  The least-aligned-axis nodal rule avoids a fixed-axis parallel singularity.
+- `[alpha,beta,gamma]^T=R_I^T theta_I^g` gives
+  `delta d=beta*a-alpha*b=theta x d` with the required sign.
+- `q20=T_p qg` and `gamma=T_d qg` preserve virtual work. Congruence gives
+  `Kphys24=T_p^T K20 T_p` and `Kdrill24=T_d^T(k_d I4)T_d`; the corresponding local and
+  global quadratic energies are identical.
+- A physical rigid rotation uses only the tangent projection of `omega`, so `gamma=0`
+  and drilling does not destroy the six physical rigid modes.
+
+### 5.3 Kinematic operators and MITC tying — `pass`
+
+- Bilinear `N_I` satisfies partition, Kronecker and derivative-sum identities.
+- Membrane and bending content comes from the direct covariant small-strain operator.
+  Only `epsilon_xi-zeta` and `epsilon_eta-zeta` are replaced.
+- `epsilon_xi-zeta` is tied at `(0,-1,0)` and `(0,+1,0)` and interpolated in `eta`;
+  `epsilon_eta-zeta` is tied at `(-1,0,0)` and `(+1,0,0)` and interpolated in `xi`.
+  Each interpolation reproduces its own edge value and is constant along the edge
+  direction, matching the original MITC4 construction.
+- The assumed covariant tensor is reconstructed through reciprocal bases, projected
+  into the stored local Cartesian frame, and converted once to engineering shear
+  `gamma_ij=2 epsilon_ij`. The same projected `B_bar` drives strain, internal force,
+  stiffness and recovery.
+
+### 5.4 Constitutive and section matrices — `pass`
+
+- `Cps=E/(1-nu^2)[[1,nu,0],[nu,1,0],[0,0,(1-nu)/2]]` uses engineering `G12`; its
+  shear coefficient is exactly `G=E/[2(1+nu)]`.
+- `C5=diag(Cps,(5/6)G I2)` is symmetric positive definite for `E>0` and
+  `-1<nu<0.5`. `sigma33=0` and absent thickness stretch remain assumptions.
+- `A=t Cps`, `B=0`, `D=t^3 Cps/12`, and `As=(5/6)Gt I2` have consistent dimensions.
+  Membrane/shear strains are dimensionless, curvature is `1/length`, `N/Q` is
+  `force/length`, `M` is `force`, stress is `force/length^2`, and energy is
+  `force*length`.
+
+### 5.5 Jacobian, derivative transform, and integration — `pass`
+
+- The three-dimensional degenerated mapping uses
+  `X=sum(N X_I)+(t*zeta/2)sum(N d_I)` with a unit nodal director and separate
+  thickness, preventing nodal thickness double counting.
+- `J=det[G_xi,G_eta,G_zeta]` and finite covariant/reciprocal bases are checked at all
+  eight stiffness points, all four midsurface tying points, center, and every
+  committed bottom/middle/top recovery evaluation. Failed points are not skipped,
+  averaged, clamped or repaired.
+- Direct natural derivatives are converted covariantly and then to the local
+  Cartesian tensor through contravariant bases; no flat-element derivative shortcut
+  is substituted for curved/warped accepted geometry.
+- Both S4 and S4R use the common in-plane `2 x 2` points
+  `+-1/sqrt(3)` with unit weights and two identical thickness points, for eight
+  volume evaluations. Tied shear is evaluated at `zeta=0` and reused at both
+  thickness points while the remaining mapping and direct components use the actual
+  thickness point.
+
+### 5.6 Internal force, residual, stiffness, and future tangent — `pass`
+
+- Current `K20=integral(B_bar^T C5 B_bar dV)` and `f_int20=K20 q20` are mutually
+  consistent and symmetric positive semidefinite in exact arithmetic.
+- The complete current weak form is in `V24` and adds the numerical drilling
+  gradient before subtracting the global nodal `CLOAD` vector.
+- No geometric stiffness or nonlinear state enters the current product path.
+  Future Section 15 correctly separates `Kmat` and the stress-dependent `Kgeo` and,
+  conditionally on a future `Phi`, includes both `A^T K20 A` and the
+  residual-weighted map-Hessian term.
+
+### 5.7 Fixed drilling contract — `pass`
+
+- `R+` contains only finite strictly positive diagonals of the eight physical
+  tangent-rotation coordinates. `k_ref=min(R+)`, `k_d=1e-3 k_ref`, and
+  `Kd_local=k_d I4` are dimensionally consistent and deterministic; empty `R+`
+  fails validation.
+- `T_d^T(k_d I4)T_d` is symmetric and positive on the four pure drilling
+  coordinates. It must remove those four nonphysical null modes without changing
+  the physical rank/null modes.
+- Drilling is excluded from generalized strain/resultant/stress and reported
+  physical strain energy. A director-parallel applied nodal moment is rejected as
+  `unsupported-drilling-load`; no numerical drill load channel exists.
+
+### 5.8 Recovery, signs, locations, units, and external comparison — `pass`
+
+- Nodal `[U1,U2,U3,UR1,UR2,UR3]` and full-residual
+  `[RF1,RF2,RF3,RM1,RM2,RM3]` are global and source-node ordered.
+- Four midsurface Gauss rows recover
+  `[E11,E22,G12,K11,K22,K12,G13,G23]` and
+  `[N11,N22,N12,M11,M22,M12,Q13,Q23]` in the stored local frame. The definition
+  `e_m(z)=epsilon0+z*kappa` fixes curvature, moment, and bottom/top stress signs.
+- `[S11,S22,S12]` is evaluated directly at `zeta=-1,0,+1`; `S33=0` is documented but
+  not emitted, and `S13/S23` point stress is not synthesized. Different natural or
+  section locations are never averaged.
+- Reference comparison first rejects missing, extra, duplicate, nonfinite or
+  identity-mismatched rows. Every U/UR family then uses the common reference-only scale,
+  near-zero ratio `0.01`, row relative tolerance `0.05` and scale-relative RMS tolerance
+  `0.01`. `S=0` requires exact-zero FESA values; there is no zero clamp or independent
+  absolute-error gate. U1/U2/U3 is blocking and UR1/UR2/UR3 is warning-only.
+- Source S4 and S4R select the same FESA MITC4 kernel/quadrature/recovery path while
+  preserving source type. This is an input mapping, not an Abaqus formulation,
+  integration, stabilization or recovery equivalence claim.
+
+## 6. Verification Readiness
+
+### 6.1 Downstream unit and invariant tests
+
+Implementation Planning shall convert the following to `RED -> GREEN -> VERIFY`:
+
+1. Shape identities; nodal/integration frame orthonormality, handedness and axis
+   tie-break determinism.
+2. `T_p`/`T_d` dimensions, orthogonal channel selection, virtual-work equality and
+   nonzero transformation-energy equality.
+3. Hand-calculated direct membrane/bending columns, all four covariant tying values,
+   interpolation weights and engineering-shear factors.
+4. `Cps/C5/A/D/As` coefficients, symmetry, positive definiteness, dimensions and
+   force/length unit-rescaling invariance.
+5. Common `2 x 2 x 2` point/weight order and an independent analytical or
+   higher-order flat-element stiffness/recovery cross-check.
+6. Required-location geometry validation: valid planar/rotated/warped cases and
+   duplicate, bow-tie/self-intersecting, zero-area, reversed, nonfinite,
+   nonpositive-J and opposed-normal negative cases.
+7. Scaled symmetry `<=1e-12`, physical rigid action `<=1e-10`, expected physical
+   rank 14, stabilized rank 18/nullity six, and positive non-rigid physical energy.
+8. Exact `R+` membership, exclusion of translations, fixed coefficient, empty-`R+`
+   failure, pure drill action and zero physical recovery/energy.
+9. Stable COO/reduction, source/result/diagnostic order and thread-count
+   repeatability.
+10. `Kff/Kfc` effective RHS, nonzero prescribed values, full-residual reaction,
+    singular-support negative case, and valid all-constrained `0 x 0 Kff` case.
+11. Exact-zero and accepted/rejected `rho_M` moment projections, including proof
+    that rejected drilling moments never reach stabilization.
+12. Mandatory HDF5 locations/components/units, physical-only energy, nonfinite
+    recovery failure and atomic finalization.
+
+### 6.2 Required patch and sign tests
+
+- independent constant `E11`, `E22`, and `G12` membrane fields with `N` and
+  middle-stress signs;
+- pure `K11` and `K22` bending with `M` order and bottom/top stress reversal;
+- pure `K12` twist with `M12` sign;
+- constant `G13` and `G23` transverse shear with `Q13/Q23` order;
+- six physical rigid states and four pure drilling states;
+- source-type-only S4/S4R variants producing identical FESA numeric rows and
+  different preserved source metadata.
+
+### 6.3 Reference and physics handoff readiness
+
+The sole declared read-only acceptance pair is:
+
+- `reference/shell/shell.inp` with
+  `reference/shell/shell displacements.csv` for full-integration S4.
+
+S4R remains required input coverage through source mapping, common-kernel,
+deterministic-assembly and HDF5 metadata tests, but no S4R Abaqus artifact is
+consumed by reference acceptance.
+
+The Reference Model and I/O documents define deterministic HDF5-to-CSV identity,
+precheck and tolerance sufficiently for later comparison. This review did not assert
+that `results.h5` exists or that any row passes. Reference Verification owns numeric
+U/UR outcome; Physics Evaluation owns force/moment balance, displacement direction,
+symmetry, result signs, recovered-resultant consistency and physical plausibility.
+
+### 6.4 Missing evidence classification
+
+- blocking_for_current_formulation: `none`
+- required_after_implementation: invariant, patch, MSVC build/CTest, declared
+  reference comparison and physics evidence above
+- nonblocking_optional: drilling coefficient sweep/energy ratio, `NR-O03`, `NR-O04`,
+  canonical naming, README/metadata/provenance, expanded benchmark portfolio and
+  broader mesh convergence studies
+- future_only: nonlinear directional-derivative/objectivity/Newton evidence after
+  its missing formulation decisions are separately approved
+
+## 7. Required Revisions
+
+### Formulation Agent
+
+- None for the approved current linear-static implementation scope.
+- Do not promote Section 15 to executable status until a separate formulation closes
+  the nonlinear global coordinate map, objective drilling and load-work contracts.
+
+### Research Agent
+
+- None before current Implementation Planning.
+- Optional locking/convergence characterization must remain clearly outside the
+  approved implementation gate and must not imply MITC4+ or Abaqus equivalence.
+
+### I/O Definition Agent
+
+- None for the current numerical verdict. Preserve exact physical/full-residual
+  distinction, source identity, locations, units and U-versus-UR decision rule.
+
+### Reference Model Agent
+
+- None for the current numerical verdict. Preserve the two declared S4 files and every
+  optional existing reference artifact read-only, exclude S4R artifacts from acceptance
+  comparison, and do not add administrative or portfolio gates.
+
+## 8. Downstream Handoff
+
+### Implementation Planning Agent
+
+Implementation Planning is authorized and shall:
+
+- trace the required tests in Section 6 to the approved requirement IDs before
+  production work;
+- keep `24 global -> 20 physical + 4 drilling` transforms, covariant MITC tying,
+  common `2 x 2 x 2` integration, fixed drilling and physical recovery as explicit
+  independent test seams;
+- preserve stiffness assembly/partition/factorization-before-load, stable reduction,
+  full-residual reaction and failure-atomic HDF5 lifecycle;
+- keep future nonlinear execution, coefficient calibration, drilling output,
+  `NR-O03/NR-O04`, reference-artifact mutation and Abaqus-equivalence claims outside
+  the plan.
+
+This handoff authorizes planning only. It does not authorize Harness execution,
+production implementation, reference artifact changes, or completion claims.
+
+### Reference Verification Agent
+
+- Compare authoritative FESA HDF5 rows directly with the matching declared Abaqus
+  displacement CSV after exact row-set precheck.
+- Let only U1/U2/U3 affect pass/fail; report every UR1/UR2/UR3 warning without
+  changing the verdict.
+
+### Physics Evaluation Agent
+
+- After reference verification, independently evaluate force and global moment
+  balance, free residual, reaction sign, displacement direction, symmetry, positive
+  physical energy and consistency of local resultants/stresses.
+
+### Coordinator and Release Agents
+
+- Record the Numerical Review gate as passed for planning at HEAD `a058ef7`.
+- Do not infer implementation or release completion. Build/test, reference,
+  physics-sanity and release-readiness gates remain pending.

docs/linear-static-mitc4-shell/physics-evaluation.md

@@ -0,0 +1,331 @@
+# Linear Static MITC4 Shell Physics Evaluation Report
+
+## Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- model_id: `shell-s4`
+- evaluated_head: `820ba30c717b3d0e113775608e20dfd5fbc05d53`
+- source_build_test_report:
+  `docs/linear-static-mitc4-shell/build-test.md`
+- source_reference_verification_report:
+  `docs/linear-static-mitc4-shell/reference-comparison.md`
+- source_reference_model:
+  `docs/linear-static-mitc4-shell/reference-model.md`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_formulation: `docs/linear-static-mitc4-shell/formulation.md`
+- source_numerical_review:
+  `docs/linear-static-mitc4-shell/numerical-review.md`
+- source_io_definition: `docs/linear-static-mitc4-shell/io.md`
+- status: `pass-for-release-agent`
+- owner_agent: `physics-evaluation-agent`
+- date: `2026-08-13`
+
+This verdict means that the documented physical checks pass and the Release Agent
+may audit release readiness. It does not approve release readiness, re-approve the
+reference comparison, or change its tolerance decision.
+
+## Input Evidence
+
+The prerequisite reference-verification report has status
+`pass-for-physics-evaluation`. The existing S4 reference-test route was rebuilt and
+executed to regenerate fresh FESA evidence. Abaqus and other reference solvers were
+not executed.
+
+| evidence | exact path or identity | status | notes |
+| --- | --- | --- | --- |
+| build/test report | `docs/linear-static-mitc4-shell/build-test.md` | `pass-for-reference-verification` | Clean MSVC x64 Debug build, focused `87/87`, lifecycle `10/10`, reference `8/8`, and full `144/144` CTest passed. |
+| reference verification | `docs/linear-static-mitc4-shell/reference-comparison.md` | `pass-for-physics-evaluation` | Required prerequisite satisfied; 147/147 blocking U rows passed and no UR warning was emitted. |
+| solver HDF5 | `.harness/build/reference/mitc4-shell-s4-comparison/results.h5` | present and readable | Freshly generated, 95,024 bytes; observed raw SHA-256 `E102D80E82BA133EBDF1C5532F3A0A4FE9984AB6CC36D00264399F7308D9230F` (inventory only). |
+| comparison ledger | `.harness/build/reference/mitc4-shell-s4-comparison/comparison.json` | present and passing | 94,349 bytes; SHA-256 `8E8DEA51B6F7C663BACC41FDA6103A4596DB26E02F1EAD6069D458F51E0102E6`; `passed=true`. |
+| declared S4 input | `reference/shell/shell.inp` | present, unchanged, read-only | SHA-256 `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF`. |
+| declared displacement CSV | `reference/shell/shell displacements.csv` | present, unchanged, read-only | SHA-256 `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B`. |
+| optional Abaqus reaction/stress CSVs | `reference/shell/shell reactions.csv`, `reference/shell/shell stresses.csv` | not used as equality gates | Their presence did not expand the approved U-only blocking boundary. |
+| solver CSV views | N/A | not generated | HDF5 was inspected directly. |
+| fresh S4 route | `.harness/build`, MSVC x64 Debug | pass | `Mitc4S4Reference.*` passed `2/2`. |
+
+The HDF5 contains 49 nodes, 36 source-`S4` elements, internal formulation
+`FESA-MITC4`, and the required full `2 x 2 x 2` result inventory. All inspected
+numeric model and result values were finite. The diagnostics dataset contains four
+approved warning rows for ignored `PREPRINT`, `RESTART`, and `OUTPUT` keywords and
+no error diagnostic.
+
+Raw HDF5 bytes are not a physics decision identity and may change across valid
+regenerations. The projected values and deterministic comparison ledger are the
+relevant evidence.
+
+### Documented model expectation
+
+The reference input is a flat `10 x 10`, thickness `0.5`, homogeneous isotropic
+square plate with `E=2.1e11` and `nu=0.3`. Its 24 perimeter nodes are fully clamped
+in all six global DOFs, and center node 2 at the global origin carries one
+`F3=-100000` concentrated force. The geometry, boundary, material, and load are
+symmetric about both global `x=0` and `y=0` planes.
+
+The documented physical expectations are therefore:
+
+- total constrained reaction force opposes the applied `-Z` force;
+- total force and moment about the global origin balance within the approved
+  normalized `1e-10` limit;
+- perimeter displacements are zero and the interior deflects in `-Z`, with the
+  largest magnitude at the center;
+- the displacement and rotation field obeys the polar/axial-vector reflection
+  parity about both symmetry planes;
+- a linear, flat transverse plate response has no membrane strain/resultant or
+  middle-surface in-plane stress, while bottom and top in-plane stresses reverse;
+- stored physical strain energy is finite, strictly positive, and consistent with
+  the recovered generalized fields and `0.5 F^T d`;
+- free residual, finite results, rank/rigid-mode evidence, and the approved
+  one-case coverage show no rigid-body or unsupported-mode symptom.
+
+## Physics Checks
+
+| check | documented expectation | observed evidence | verdict | classification |
+| --- | --- | --- | --- | --- |
+| global equilibrium | Applied plus constrained reaction force/moment about origin; normalized metrics `<=1e-10` | force residual `[0,0,-1.12049747258425e-9]`; moment residual `[6.45741238258779e-10,-2.63753463514149e-10,0]`; normalized force/moment `1.12049747258425e-14` / `2.35981341581908e-15` | pass | N/A |
+| reaction consistency | Perimeter-only reactions from full residual oppose `F3=-100000`; constrained motion is zero | summed constrained force `[0,0,99999.9999999989]`; constrained displacement exactly zero | pass | N/A |
+| displacement direction | Interior bends in `-Z`; center is maximum; no in-plane or drilling leakage | center `U3=-2.35504417650849e-5`; all 25 free-node U3 values nonpositive; `U1/U2/UR3` exactly zero | pass | N/A |
+| symmetry | Correct polar/axial reflection parity about `x=0` and `y=0` | max absolute reflection mismatch `4.27126230127111e-20` / `2.29087480670778e-20` | pass | N/A |
+| element force balance | No documented direct element-end-action output exists for shell GP resultants | Global assembled residual/equilibrium passed; direct element-end-action balance is skipped rather than inferred from mismatched locations | skipped | N/A |
+| recovered resultant consistency | `N=A*epsilon0`, `M=D*kappa`, `Q=As*gamma0` at each of 144 GP rows | absolute L2 residual `9.9301874551584e-12`; normalized L2 `5.56240926027087e-17` | pass | N/A |
+| stress/location/sign sanity | Local `[S11,S22,S12]` at bottom/middle/top; middle zero and faces reverse for this linear flat bending field | absolute L2 constitutive residual `1.29488994752188e-9`; normalized `3.05323934859307e-16`; middle stress exactly zero; face reversal exact | pass | N/A |
+| rigid body/nonfinite | Complete constraints, finite solution, acceptable residual, no abnormal uncontrolled mode | zero nonfinite values; `FREE_RESIDUAL_NORMALIZED=1.07747756058988e-14`; rank/rigid-mode tests passed upstream | pass | N/A |
+| physical energy | Positive and consistent physical-only energy; no drilling-energy gate | `1.17752208825422`; independently recovered value identical to relative `1.88569375589575e-16`; `0.5F^Td=1.17752208825424` | pass | N/A |
+| model coverage | Approved sole S4 case plus declared invariant/patch/common-path portfolio | S4 case covers end-to-end symmetric plate response; focused build/test portfolio passed; expanded models are explicitly nonblocking | pass | N/A |
+
+### 1. Global force and moment equilibrium
+
+The reaction dataset is the assembled full residual `K*d-F`. Applying the
+constraint mask gives
+
+```text
+sum constrained RF = [0, 0, 99999.9999999989]
+sum applied F       = [0, 0, -100000]
+force residual      = [0, 0, -1.12049747258425e-9]
+```
+
+The load acts at the global origin and therefore contributes no origin moment.
+The constrained force and moment rows give
+
+```text
+moment residual = [ 6.45741238258779e-10,
+                   -2.63753463514149e-10,
+                    0 ]
+```
+
+These independently reconstructed values match the stored HDF5 equilibrium vector
+exactly. The serialized normalized metrics are:
+
+| metric | value | threshold | result |
+| --- | ---: | ---: | --- |
+| `FREE_RESIDUAL_NORMALIZED` | `1.07747756058988e-14` | `1.0e-10` | pass |
+| `FORCE_BALANCE_NORMALIZED` | `1.12049747258425e-14` | `1.0e-10` | pass |
+| `MOMENT_BALANCE_NORMALIZED` | `2.35981341581908e-15` | `1.0e-10` | pass |
+
+### 2. Reaction consistency and constrained/free meaning
+
+The 24 perimeter nodes provide 144 constrained DOFs; the remaining 150 DOFs are
+free. The HDF5 constraint mask agrees with the input set exactly, all prescribed
+values are zero, and the maximum constrained displacement is exactly zero.
+
+The total positive `RF3=99999.9999999989` opposes the center `F3=-100000`. The
+nearly zero total `RF1/RF2` and origin `RM1/RM2/RM3` are required by the centered
+load and double symmetry. The largest raw free residual component is
+`7.62156560085714e-10`; because translational and rotational entries have different
+dimensions, the decision uses the documented normalized metric rather than this
+raw maximum. The normalized free residual passes by more than three orders of
+magnitude.
+
+### 3. Displacement direction and deformation mode
+
+Center node 2 has
+
+```text
+[U1,U2,U3,UR1,UR2,UR3]
+= [0, 0, -2.35504417650849e-5, 0, 0, 0]
+```
+
+Every free-node `U3` is nonpositive, the center is the maximum-magnitude
+translation, and every clamped boundary displacement is zero. Across the complete
+model, maximum absolute `U1`, `U2`, and `UR3` are exactly zero. Maximum absolute
+`UR1` and `UR2` are `7.55060026636931e-6` and
+`7.55060026636933e-6`, respectively. This is the expected symmetric plate-bending
+mode under a negative transverse center load, with no in-plane or drilling
+deformation leakage.
+
+### 4. Reflection symmetry and expected zeros
+
+For reflection through `x=0`, polar displacement components transform as
+`[-U1,+U2,+U3]`, while the axial rotation vector transforms as
+`[+UR1,-UR2,-UR3]`. Reflection through `y=0` similarly uses
+`[+U1,-U2,+U3,-UR1,+UR2,-UR3]`.
+
+All 294 scalar nodal component comparisons per reflection were paired by source
+coordinates. The maximum absolute mismatch was `4.27126230127111e-20` for the
+`x` reflection and `2.29087480670778e-20` for the `y` reflection. The equality
+evidence is numerical roundoff, not a new acceptance tolerance.
+
+### 5. Shell generalized resultants and physical work signs
+
+At every one of 36 elements times four midsurface Gauss locations, an independent
+constitutive reconstruction applied the documented centered-section identities:
+
+```text
+N = A * epsilon0
+M = D * kappa
+Q = As * gamma0
+```
+
+The complete 1,152-component resultant comparison has absolute L2 residual
+`9.9301874551584e-12` and normalized L2 residual
+`5.56240926027087e-17`. All membrane strains and `N` resultants are exactly zero,
+as required for this linear flat plate bending response.
+
+The generalized work density `generalized_strain dot section_resultant` is strictly
+positive at all 144 locations: minimum `4.46321446508071e-4`, maximum
+`1.27059464107458e-1`, negative count `0`. This supports the physical sign and
+component-order interpretation.
+
+The shell contract does not expose element-end nodal actions and explicitly forbids
+averaging mismatched result locations. Consequently a separate GP-to-element-end
+force-balance assertion is not documented and is marked `skipped`; the assembled
+global residual and equilibrium checks provide the required force-balance evidence.
+
+### 6. Stress component, location, and sign sanity
+
+HDF5 stores local in-plane `[S11,S22,S12]` directly at each GP and ordered section
+positions `BOTTOM(-1), MIDDLE(0), TOP(+1)`. Independent reconstruction used
+`Cps*(epsilon0+z*kappa)` at all 432 section locations. The absolute L2 residual is
+`1.29488994752188e-9`, or normalized `3.05323934859307e-16`, against a maximum
+absolute stored stress of `313250.779137971`.
+
+All middle-surface in-plane stresses are exactly zero. For every component and GP,
+bottom and top stresses are equal in magnitude and opposite in sign. This is the
+documented curvature/stress-location convention for a centered homogeneous section.
+No `S33`, `S13`, `S23`, nodal stress, or Abaqus stress equality was inferred.
+
+### 7. Physical energy, residual, and rigid-body symptoms
+
+The stored `PHYSICAL_STRAIN_ENERGY` is finite and positive:
+
+```text
+HDF5 physical energy              = 1.17752208825422
+independent GP generalized energy = 1.17752208825422
+0.5 * F^T * d                     = 1.17752208825424
+```
+
+The HDF5-versus-recovered normalized difference is
+`1.88569375589575e-16`; the final fresh HDF5-versus-external-work normalized
+difference is `9.14561471609432e-15`. The HDF5 value excludes numerical drilling stabilization,
+as required. No drilling-energy ratio, calibration, or warning criterion was
+introduced.
+
+The solution is finite, the full perimeter support is enforced, the expected
+deformation mode is smooth and symmetric, all three normalized verification
+metrics pass `1e-10`, and the build/test evidence passes six-rigid-mode,
+stabilized-rank, non-rigid positive-energy, and pure-drill separation tests. No
+rigid-body, singular, nonfinite, or abnormal-mode symptom is present.
+
+### 8. Approved model coverage
+
+The sole approved reference model provides end-to-end evidence for the
+full-integration `S4 -> FESA-MITC4` path, a nonzero transverse bending response,
+fully constrained reaction recovery, global force/moment balance, two-axis
+symmetry, physical energy, and shell result recovery.
+
+The prerequisite focused `87/87` build/test portfolio supplies the documented
+complementary element evidence: membrane, bending, transverse-shear and twist patch
+fields; six physical rigid modes; expected physical/stabilized rank; frame and
+energy invariance; fixed numerical drilling and physical-recovery separation; and
+S4/S4R common-kernel/source-metadata behavior. The reference verification supplies
+the approved external displacement comparison.
+
+This is the complete approved coverage for the feature. Pinched-cylinder,
+hemisphere, Scordelis-Lo, locking/convergence sequences, drilling calibration,
+expanded distorted/curved portfolios, and `NR-O01` through `NR-O04` are explicitly
+nonblocking or out of scope and were not invented as new physics gates.
+
+## Execution Evidence
+
+| order | exact command or read-only operation | exit code | result |
+| ---: | --- | ---: | --- |
+| 1 | `cmake --build .harness/build --config Debug --target fesa_reference_tests` | `0` | Existing S4 reference route rebuilt. |
+| 2 | `ctest --test-dir .harness/build -C Debug -R '^Mitc4S4Reference\.' --output-on-failure` | `0` | Fresh S4 tests `2/2` passed; FESA HDF5 and ledger regenerated. |
+| 3 | `h5dump.exe -n .harness/build/reference/mitc4-shell-s4-comparison/results.h5` with HDF5/oneAPI runtime on `PATH` | `0` | Required HDF5 inventory inspected read-only. |
+| 4 | `h5dump.exe -y -w 0 -m '%.17g' -d <dataset> .harness/build/reference/mitc4-shell-s4-comparison/results.h5` for model nodes/elements/mask/prescribed values, nodal U/R, shell frames/strain/resultant/stress, energy/equilibrium/metrics/diagnostics | `0` | High-precision read-only numerical audit supplied all reported values. |
+| 5 | PowerShell independent input/mask, reaction/equilibrium, reflection, constitutive recovery, stress-location, energy, finite, and diagnostics audit over the `h5dump` values | `0` | All documented physics assertions passed. |
+| 6 | `cmake --build .harness/build --config Debug --target fesa_unit_tests fesa_reference_tests`, then `ctest --test-dir .harness/build -C Debug -R 'Mitc4Shell(Kernel|Patch|Drilling|PhysicalRecovery)|ResultRecovery\..*Shell|Mitc4S4Reference' --output-on-failure` | `0` | Final focused verification passed `17/17`; the regenerated ledger retained `passed=true`, 294 rows, and zero warnings. |
+
+Three intermediate audit-helper incidents were diagnostic-only and did not alter
+the product or verdict: a compound-record regular expression initially omitted the
+last node, a PowerShell array-expression initially bound subtraction to an array,
+and a final optional `h5ls` operand-order probe exited `1`. The root causes were
+confirmed as audit-command/parser usage, while successful high-precision HDF5
+inspection established 49/49 nodes and the complete evidence above. They are not
+solver, HDF5-schema, reference, or physics failures.
+
+## Failure Classification
+
+- classification: `N/A`
+- primary_failure: `N/A`
+- evidence: all documented force/moment equilibrium, reaction, displacement,
+  symmetry, recovery, stress-location, physical-energy, residual, finite-result,
+  rigid-mode and approved coverage checks passed
+- correction_handoff: `N/A`
+- non_gating_incident: resolved read-only audit helper parsing/usage errors only
+
+## Evaluation Verdict
+
+- verdict: `pass-for-release-agent`
+- reason: the prerequisite reference status is valid and fresh FESA S4 evidence
+  satisfies every documented physical expectation under its approved threshold;
+  no implementation, formulation, I/O, model-coverage, nonfinite, rigid-body, or
+  environment failure remains
+- release_approval: `not granted by this report`
+- reference_reapproval: `not performed by this report`
+
+## Handoff Recommendation
+
+| target_agent | reason | required_input |
+| --- | --- | --- |
+| Release Agent | All documented MITC4 physics checks passed. | This report, the build/test and reference-verification reports, the fresh build-local HDF5/comparison evidence, and the nonblocking limitations below. |
+
+## No-Change Assertion
+
+- source_files_modified: `false`
+- test_files_modified: `false`
+- cmake_files_modified: `false`
+- requirements_modified: `false`
+- formulation_modified: `false`
+- numerical_review_modified: `false`
+- io_contract_modified: `false`
+- reference_model_contract_modified: `false`
+- build_test_report_modified: `false`
+- reference_verification_report_modified: `false`
+- phase_files_modified: `false`
+- tolerance_policies_modified: `false`
+- reference_artifacts_modified: `false`
+- Abaqus_or_other_reference_solver_executed: `false`
+- owned_report_created: `true`
+- generated_build_local_evidence: `true`, ignored under `.harness/build/`
+- pre_existing_untracked_build_test_report_preserved: `true`
+- pre_existing_untracked_reference_verification_report_preserved: `true`
+
+## Open Issues
+
+- Nonblocking coverage limitation: the approved external reference inventory is one
+  flat, symmetric, fully clamped S4 plate. The approved analytical/unit portfolio
+  supplies membrane, shear, twist, rigid-mode, rank, drilling-separation and S4R
+  common-path evidence. No broader Abaqus or benchmark portfolio is claimed.
+- Known original-MITC4 limitations remain: transverse-shear locking is mitigated by
+  the tying field, but distorted-curved membrane locking and broader thin/thick mesh
+  convergence are not characterized by this release gate.
+- Source `S4R` maps to the same full-integration FESA kernel and is not compared to
+  the optional Abaqus S4R artifacts. This is an approved boundary, not missing
+  physics evidence.
+- The fixed numerical drilling stiffness is not a physical stress/resultant/energy
+  channel. No drilling calibration, coefficient sweep, energy-ratio gate, or
+  director-parallel moment support is claimed.
+- Geometrically nonlinear execution remains outside the current feature even though
+  future-only residual/tangent equations are documented. No nonlinear release claim
+  is made.
+- No open issue blocks Release Agent review.

docs/linear-static-mitc4-shell/reference-comparison.md

@@ -0,0 +1,271 @@
+# Linear Static MITC4 Shell Reference Verification Report
+
+## Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- model_id: `shell-s4`
+- source_head: `820ba30c717b3d0e113775608e20dfd5fbc05d53`
+- source_build_test_report:
+  `docs/linear-static-mitc4-shell/build-test.md`
+- source_reference_models:
+  `docs/linear-static-mitc4-shell/reference-model.md`
+- source_io_definition: `docs/linear-static-mitc4-shell/io.md`
+- source_requirements: `docs/linear-static-mitc4-shell/requirements.md`
+- historical_status: `pass-for-physics-evaluation`
+- current_status: `pass-for-physics-evaluation`
+- superseded_on: `2026-08-18`
+- revalidated_on: `2026-08-18`
+- owner_agent: `reference-verification-agent`
+- date: `2026-08-13`
+
+The fixed-absolute-tolerance evidence below is preserved as a historical record. ADR-022
+supersedes that decision with the common family-scale row/RMS policy. The following revalidation
+is the current reference-gate evidence.
+
+## 2026-08-18 Common-policy Revalidation
+
+`cmake --build .harness/build --config Debug` passed, followed by
+`ctest --test-dir .harness/build -C Debug --output-on-failure`: 214/214 tests passed.
+The S4 reference test generated
+`.harness/build/reference/mitc4-shell-s4-comparison/comparison.json` from the declared
+read-only input/CSV pair.
+
+- exact canonical rows: 294
+- blocking translation family: 147 rows, all row/RMS gates passed
+- warning-only rotation family: 147 rows, all row/RMS gates passed; warnings 0
+- identity/schema/nonfinite failures: 0
+- overall verdict: `passed=true`
+
+| family | rows | scale | near-zero rows | max absolute error | scale-relative RMS |
+| --- | ---: | ---: | ---: | ---: | ---: |
+| translation `U1/U2/U3` | 147 | `2.37408203e-5` | 122 | `1.903785349151439e-7` | `1.0332756650397166e-3` |
+| rotation `UR1/UR2/UR3` | 147 | `7.60725743e-6` | 107 | `6.882854962740428e-8` | `2.9771372973060382e-3` |
+
+All values above come from the generated JSON ledger. The historical report body starts below
+and must not be read as the current identity/tolerance evidence.
+
+The prerequisite build/test report has status `pass-for-reference-verification`.
+This report verifies only the approved full-integration S4 displacement case. It
+does not compare S4R artifacts, claim Abaqus formulation equivalence, approve
+physics sanity, or approve release readiness.
+
+## Verdict
+
+The reference-comparison gate passes. The fresh FESA result has an exact
+294-row identity match to the read-only Abaqus displacement CSV. All 147 blocking
+`U1/U2/U3` rows satisfy the fixed absolute tolerance `1.0e-5`. All 147
+warning-only `UR1/UR2/UR3` rows also satisfy `1.0e-5`, so no rotational warning
+was emitted.
+
+- declared cases compared: `1/1`
+- blocking rows passed: `147/147`
+- blocking failures: `0`
+- warning-only rows compared: `147/147`
+- warning-only exceedances and warnings: `0/0`
+- missing / extra / duplicate / nonfinite rows: `0 / 0 / 0 / 0`
+- largest blocking normalized error: `0.0190378534915144`
+- comparison verdict: `passed=true`
+
+## Artifact Inventory
+
+| item | exact path | status | direct evidence |
+| --- | --- | --- | --- |
+| declared S4 input | `reference/shell/shell.inp` | present, unchanged | 4,770 bytes; SHA-256 `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF` |
+| required displacement CSV | `reference/shell/shell displacements.csv` | present, unchanged | 5,592 bytes; SHA-256 `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B`; 49 wide rows and six finite components per row |
+| authoritative solver output | `.harness/build/reference/mitc4-shell-s4-comparison/results.h5` | present, valid | 95,024 bytes after a fresh verification run; observed raw SHA-256 `E88E5E5E1F83AB95CAE545AD1F28F896D3FCB82F3EA2B8006A3A703FED07A5CB` (non-decision inventory only) |
+| deterministic comparison ledger | `.harness/build/reference/mitc4-shell-s4-comparison/comparison.json` | present, valid | 94,349 bytes; SHA-256 `8E8DEA51B6F7C663BACC41FDA6103A4596DB26E02F1EAD6069D458F51E0102E6`; `passed=true` |
+| S4R reference artifacts | `reference/shellR/` | not consumed | Excluded from the approved acceptance comparison. |
+| reaction and stress CSVs | other files under `reference/shell/` | not consumed | Optional inspection evidence only; not an equality gate. |
+| FESA-derived CSV | N/A | not generated | HDF5 was compared directly to the Abaqus CSV. |
+
+The input and required CSV had identical SHA-256 values and identical UTC
+modification times (`2026-08-11T11:44:40.9520527Z`) before and after the fresh
+comparison runs. `git diff --name-only -- reference` and
+`git status --short -- reference` each returned zero paths.
+
+The HDF5 hash above inventories one fresh generation, not a stable content
+identity. Raw HDF5 bytes changed across regenerations while the projected values
+and deterministic JSON ledger remained identical; concurrent executions of the
+same ignored build-local route may replace the file again. This report does not
+claim byte-identical HDF5 container output.
+
+### Authoritative HDF5 projection audit
+
+The comparator and a read-only HDF5 2.1.1 inspection established:
+
+- `/metadata`: schema version `0`, feature `linear-static-mitc4-shell`, solver
+  version `0.1.0`, `Step-1`, frame `0`, internal formulation `FESA-MITC4`, and
+  integration rule `2x2x2-gauss; mitc4-edge-midpoint-shear`;
+- source input identity:
+  `path=C:/Users/baram/orca/workspaces/FESADev/MITC4/reference/shell/shell.inp;content_identity=fnv1a64:cf0795df753d377e`;
+- `/model/nodes`: 49 compound rows with ordered fields
+  `internal_node_id`, `instance_name`, `source_label`, and `coordinates`;
+- `/model/elements`: 36 compound rows, source element type `S4`, and internal
+  formulation `FESA-MITC4`;
+- `/steps/Step-1/frames/0/nodal/displacement`: float64 shape `[49,6]`, global
+  nodal components `[UX,UY,UZ,URX,URY,URZ]`, mapped to
+  `[U1,U2,U3,UR1,UR2,UR3]`;
+- all projected source identities and displacement values are unique and finite.
+
+## Comparison Contract
+
+- case identity: `shell-s4`
+- authoritative FESA dataset:
+  `/steps/Step-1/frames/0/nodal/displacement`
+- HDF5 node identity source: `/model/nodes.instance_name` and
+  `/model/nodes.source_label`
+- normalized key:
+  `(case_id, instance_name, source_node_label, component)`
+- observed instance: `Part-1-1`
+- observed source nodes: 49 unique labels
+- CSV projection: trimmed `Part Instance Name`, `Node Label`, `U-U1`, `U-U2`,
+  `U-U3`, `UR-UR1`, `UR-UR2`, and `UR-UR3`
+- stable component order: `[U1,U2,U3,UR1,UR2,UR3]`
+- blocking inventory: `U1/U2/U3`
+- warning-only inventory: `UR1/UR2/UR3`
+- row precheck: exact key set, unique rows, finite HDF5/reference values, and
+  exact required HDF5 schema before tolerance evaluation
+- tolerance source: requirements `FESA-REQ-LSMITC4-058` through `064`, the I/O
+  definition Section 7, and the approved reference-model contract Section 5
+- row tolerance: `abs(fesa-reference) <= 1.0e-5`
+- tolerance-normalized error: `absolute_error / 1.0e-5`
+- `reference_scale`: diagnostic only; it does not alter the tolerance or verdict
+- zero policy: no clamp, substitution, omission, averaging, or row-order matching
+
+Independent ledger reconstruction found 294 projected CSV keys, 294 comparison
+keys, zero duplicate keys, and exact set equality. All 294 serialized row
+tolerances equal `1.0e-5`. The complete per-row decisions remain in the
+identified deterministic `comparison.json`.
+
+## Execution Evidence
+
+The existing FESA S4 reference route was used; Abaqus and other reference solvers
+were not executed. The first run regenerated the evidence after its test-owned
+output directory cleanup. Further exact reruns confirmed the ledger bytes and the
+final verification assertions.
+
+| order | exact command | exit code | result |
+| ---: | --- | ---: | --- |
+| 1 | `cmake --build .harness/build --config Debug --target fesa_reference_tests` | `0` | Reference test executable built; duration `1.830 s`. |
+| 2 | `ctest --test-dir .harness/build -C Debug -R '^Mitc4S4Reference\.' --output-on-failure` | `0` | Fresh S4 E2E tests `2/2` passed; duration `0.760 s`. |
+| 3 | PowerShell CSV/JSON identity and metric audit | `0` | 294 exact keys; missing/extra/duplicate/nonfinite/nonfixed-tolerance counts all zero. |
+| 4 | `ctest --test-dir .harness/build -C Debug -R '^Mitc4S4Reference\.' --output-on-failure` | `0` | Repeat S4 E2E tests `2/2` passed; duration `0.911 s`; JSON hash unchanged. |
+| 5 | `h5dump.exe -A -g '/metadata' <results.h5>` with HDF5 and Intel oneAPI 2026.1 runtime directories prepended to `PATH` | `0` | Required metadata values inspected read-only. |
+| 6 | `h5dump.exe -H -d '/model/nodes' <results.h5>` | `0` | Compound schema and shape `[49]` inspected. |
+| 7 | `h5dump.exe -H -d '/model/elements' <results.h5>` | `0` | Compound schema and shape `[36]` inspected. |
+| 8 | `h5dump.exe -H -d '/steps/Step-1/frames/0/nodal/displacement' <results.h5>` | `0` | Float64 displacement schema and shape `[49,6]` inspected. |
+| 9 | Final `ctest --test-dir .harness/build -C Debug -R '^Mitc4S4Reference\.' --output-on-failure` plus report/metric assertion script | `0` | S4 E2E tests `2/2`; `verification=PASS`; final generated artifacts retained. |
+| 10 | `git diff --check` | `0` | Report patch has no whitespace errors. |
+| 11 | target-only Git scope and reference no-change audit | `0` | Only this owned report and the pre-existing untracked build/test report are visible; reference scope remains clean. |
+
+An initial optional `h5dump` probe without the Intel runtime directory exited
+`-1073741515` because `libmmd.dll` was not on that process's `PATH`. The same
+read-only inspection passed after adding the installed oneAPI runtime path. This
+was an inspection-environment incident, not a solver, schema, comparison, or
+reference-artifact failure.
+
+The repeated JSON ledger was byte-identical:
+
+- size: `94,349` bytes
+- SHA-256 before and after repeat: `8E8DEA51B6F7C663BACC41FDA6103A4596DB26E02F1EAD6069D458F51E0102E6`
+- row decisions: `294`
+- component metrics: `6`
+- node vector metrics: `49`
+- warnings: `0`
+- verdict: `passed=true`
+
+## Quantity Results
+
+`max_rel` is calculated only for nonzero Abaqus values and is diagnostic. Tiny
+reference residues make it ill-conditioned; the acceptance decision uses only
+the fixed absolute tolerance and its fixed-tolerance-normalized error.
+
+| quantity | components | behavior | compared | missing | extra | duplicate | nonfinite | max abs | max rel | max normalized | RMS | L2 norm | worst ID/component by abs error | result |
+| --- | --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- | --- |
+| displacement | `U1/U2/U3` | blocking | 147 | 0 | 0 | 0 | 0 | `1.90378534915144e-7` | `1.0` | `0.0190378534915144` | `2.45308118840709e-8` | `2.97420287738877e-7` | `Part-1-1`, node 2, `U3` | pass |
+| rotation | `UR1/UR2/UR3` | warning-only | 147 | 0 | 0 | 0 | 0 | `6.88285496274043e-8` | `35.7165926082675` | `0.00688285496274043` | `2.26478498250615e-8` | `2.74590586054375e-7` | `Part-1-1`, node 44, `UR1` | pass, no warning |
+
+The translational maximum relative error `1.0` occurs on near-zero reference
+residue, including node 1 `U3=-1.17769804e-32`. The rotational maximum relative
+error `35.7165926082675` occurs at node 2 `UR2`, where the reference is
+`1.3860685e-22` and the absolute error is only `4.95056439416524e-21`.
+Neither value was clamped or omitted, and neither relative diagnostic controls
+the verdict.
+
+### Component metrics
+
+Each component contains 49 matched rows. `reference_scale` is retained only as
+diagnostic information. `L2 norm` is the comparator's component
+`vector_norm_error`, and `worst_row` is the zero-based index in the JSON ledger.
+
+| component | behavior | reference scale | tolerance | max abs | max rel | max normalized | RMS | L2 norm | worst row | worst identity | result |
+| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- | --- |
+| `U1` | blocking | `3.11730945e-23` | `1.0e-5` | `3.11730945e-23` | `1.0` | `3.11730945e-18` | `1.13587076092006e-23` | `7.95109532644045e-23` | 66 | node 12 / `U1` | pass |
+| `U2` | blocking | `3.11730945e-23` | `1.0e-5` | `3.11730945e-23` | `1.0` | `3.11730945e-18` | `1.13587076092006e-23` | `7.95109532644045e-23` | 61 | node 11 / `U2` | pass |
+| `U3` | blocking | `2.37408203e-5` | `1.0e-5` | `1.90378534915144e-7` | `1.0` | `0.0190378534915144` | `4.24886125341252e-8` | `2.97420287738877e-7` | 8 | node 2 / `U3` | pass |
+| `UR1` | warning-only | `7.60725743e-6` | `1.0e-5` | `6.88285496274043e-8` | `11.9344619221920` | `0.00688285496274043` | `2.77378379212910e-8` | `1.94164865449037e-7` | 261 | node 44 / `UR1` | pass, no warning |
+| `UR2` | warning-only | `7.60725743e-6` | `1.0e-5` | `6.88285496273990e-8` | `35.7165926082675` | `0.00688285496273990` | `2.77378379212909e-8` | `1.94164865449036e-7` | 202 | node 34 / `UR2` | pass, no warning |
+| `UR3` | warning-only | `5.27113701e-25` | `1.0e-5` | `5.27113701e-25` | `1.0` | `5.27113701e-20` | `2.12986098533393e-25` | `1.49090268973375e-24` | 203 | node 34 / `UR3` | pass, no warning |
+
+### Worst rows and node-vector metrics
+
+- global worst and worst blocking row: `shell-s4`, `Part-1-1`, node 2,
+  `U3`; FESA `-2.35504417650849e-5`, Abaqus `-2.37408203e-5`, absolute
+  error `1.90378534915144e-7`, tolerance `1.0e-5`, normalized error
+  `0.0190378534915144`, within tolerance;
+- worst warning-only row: `shell-s4`, `Part-1-1`, node 44, `UR1`; FESA
+  `1.39850632037260e-6`, Abaqus `1.46733487e-6`, absolute error
+  `6.88285496274043e-8`, normalized error `0.00688285496274043`, within
+  tolerance;
+- maximum per-node displacement-vector error: `1.90378534915144e-7` at
+  `Part-1-1`, node 2;
+- maximum per-node rotation-vector error: `7.21144690026432e-8` at
+  `Part-1-1`, node 39.
+
+## Failure Classification
+
+- classification: `N/A`
+- primary_failure: `N/A`
+- evidence: declared artifacts exist; the authoritative HDF5 projection passed
+  schema and identity checks; 294/294 exact row keys are unique and finite;
+  147/147 blocking rows pass; zero rotational exceedances or warnings occurred
+- correction_handoff: `N/A`
+
+The optional first `h5dump` runtime-path incident is classified as a resolved
+inspection-environment incident and does not affect this gate.
+
+## Handoff Recommendation
+
+| target_agent | reason | required input |
+| --- | --- | --- |
+| Physics Evaluation Agent | All declared blocking reference rows passed the exact approved tolerance, and the warning-only inventory has no exceedance. | This report, `.harness/build/reference/mitc4-shell-s4-comparison/results.h5`, and its deterministic `comparison.json`. |
+
+The handoff status is `pass-for-physics-evaluation` only. Physics Evaluation
+remains responsible for equilibrium, reaction sign, displacement direction,
+symmetry, physical energy, result signs, and model-coverage adequacy.
+
+## No-Change Assertion
+
+- source_files_modified: `false`
+- test_files_modified: `false`
+- cmake_files_modified: `false`
+- requirements_modified: `false`
+- formulation_modified: `false`
+- numerical_review_modified: `false`
+- io_contract_modified: `false`
+- reference_model_contract_modified: `false`
+- phase_files_modified: `false`
+- tolerance_policies_modified: `false`
+- reference_artifacts_modified: `false`
+- Abaqus_or_other_reference_solver_executed: `false`
+- owned_report_created: `true`
+- generated_build_local_evidence: `true`, ignored under `.harness/build/`
+- pre_existing_untracked_build_test_report_preserved: `true`
+
+## Open Issues
+
+- None blocking Physics Evaluation.
+- Nonblocking audit note: raw HDF5 container hashes differed across fresh
+  regenerations, while the authoritative projected rows and serialized comparison
+  ledger were stable and the JSON was byte-identical. No HDF5 byte-determinism
+  claim is made by this report.

docs/linear-static-mitc4-shell/reference-model.md

@@ -3,18 +3,19 @@
 ## Metadata
 
 - feature_id: `linear-static-mitc4-shell`
-- source_requirement: `docs/requirements/linear-static-mitc4-shell.md`
-- source_io_definition: `docs/io-definitions/linear-static-mitc4-shell-io.md`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_io_definition: `docs/linear-static-mitc4-shell/io.md`
 - status: `approved-for-implementation-planning`
 - owner_agent: `reference-model-agent`
-- date: `2026-08-12`
+- date: `2026-08-13`
+- comparison_revalidated_on: `2026-08-18`
 - artifact_policy: `read-only-existing-files`
 - authoritative_fesa_output: `results.h5`
 
 ## 1. Purpose and boundary
 
-This document is the lightweight inventory and comparison contract for the two
-approved MITC4 reference cases. Abaqus supplies an external displacement reference;
+This document is the lightweight inventory and comparison contract for the sole
+approved full-integration S4 reference case. Abaqus supplies an external displacement reference;
 it does not define FESA element equations, integration, stabilization, recovery, or
 implementation structure. FESA and Abaqus are not required to operate identically.
 
@@ -37,13 +38,12 @@ input, CSV, row mapping, or tolerance.
 | --- | --- | --- | --- | --- | --- |
 | `shell-s4` | `S4` | input | `reference/shell/shell.inp` | `4005851E1AB22FD3A16AC17A8D5DA3E051233F69F37419079F3553AD134ECFCF` | `TYPE=S4`; one linear-static case |
 | `shell-s4` | `S4` | required reference | `reference/shell/shell displacements.csv` | `C81D94E0B4A849F87AA0F79C83A79B94D5661AC79E44ED826919AB432C87746B` | 49 data rows; U and UR components |
-| `shell-s4r` | `S4R` | input | `reference/shellR/shellR.inp` | `1325940FB42B78961CF25E84379BF2693846FAD22473E7688AC5456B37B18CB4` | `TYPE=S4R`; one linear-static case |
-| `shell-s4r` | `S4R` | required reference | `reference/shellR/shellR displacements.csv` | `8887ACC5ED007CB97583A9FDC1150B48B9297E269A5BA8EBA6C1A5F6306E98CB` | 49 data rows; U and UR components |
 
-The existing reaction and stress CSVs in both directories are optional inspection
-evidence only. They are not pass/fail inputs. No agent may rename, rewrite, normalize,
-repair, regenerate, or restore any reference artifact unless a later phase explicitly
-authorizes that operation.
+The existing reaction and stress CSVs in `reference/shell/` are optional inspection
+evidence only. Existing files under `reference/shellR/` are also optional inspection
+evidence and are not consumed by acceptance comparison. They are not pass/fail inputs.
+No agent may rename, rewrite, normalize, repair, regenerate, or restore any reference
+artifact unless a later phase explicitly authorizes that operation.
 
 ## 3. Required CSV mapping
 
@@ -62,8 +62,8 @@ Trim surrounding whitespace from header names, then apply this exact projection:
 
 The normalized row key is
 `(case_id,instance_name,source_node_label,component)`. Stable comparison order is
-case order `shell-s4`, `shell-s4r`; then FESA instance declaration order; stable
-source-node order; and component order `[U1,U2,U3,UR1,UR2,UR3]`.
+case `shell-s4`; then FESA instance declaration order; stable source-node order; and
+component order `[U1,U2,U3,UR1,UR2,UR3]`.
 
 ## 4. Precheck
 
@@ -81,34 +81,41 @@ zero-clamped and mismatched rows are not omitted, averaged, or synthesized.
 
 ## 5. Tolerance and decision rule
 
-For each case and component `c` independently, using only finite Abaqus rows:
+Matched `U1/U2/U3` rows form one blocking translation family and matched
+`UR1/UR2/UR3` rows form one warning-only rotation family. For each family:
 
 ```text
-reference_scale_c = max(abs(abaqus_value_i))
-tolerance_c = 1e-9 + 1e-6 * reference_scale_c
-absolute_error_i = abs(fesa_value_i - abaqus_value_i)
+S = max(abs(abaqus_value_i))
+absolute_error_i = abs(fesa_value_i-abaqus_value_i)
+near_zero_band = 0.01*S
+
+near-zero row: absolute_error_i <= near_zero_band
+other row: absolute_error_i/abs(abaqus_value_i) <= 0.05
+relative_rms = sqrt(mean(absolute_error_i^2))/S <= 0.01
 ```
 
-The `1e-9` floor is expressed in the model's user-consistent length unit for
-`U1/U2/U3` and is dimensionless for `UR1/UR2/UR3`. A zero reference scale leaves only
-the absolute floor; no alternate denominator is introduced.
+The scale uses read-only Abaqus values only. No independent absolute-error gate, zero clamp or
+row omission is permitted. A zero-scale family passes only when all FESA values are exactly zero;
+otherwise it reports `zero-reference-scale-nonzero-error` without NaN/Inf. The U family controls
+the verdict and the UR family uses the same numeric rule for deterministic warning-only evidence.
 
-- Every matched `U1/U2/U3` row must satisfy `absolute_error_i <= tolerance_c`.
-  Any U exceedance fails that case and the feature reference comparison.
-- `UR1/UR2/UR3` uses the same formula. Every exceedance produces a deterministic
-  warning containing the case, source row, component, error, and tolerance, but does
-  not change pass/fail.
+- Every matched `U1/U2/U3` row and the translation family RMS must satisfy the common
+  row/RMS rule. Any U row or family-RMS exceedance fails the feature reference comparison.
+- `UR1/UR2/UR3` uses the same family-scale row/RMS rule. Every row or family-RMS
+  exceedance produces deterministic warning evidence but does not change pass/fail.
 
-The comparison report records every U/UR row decision, maximum absolute error,
-component-scale normalized error, RMS error, vector-norm error, worst source
+The comparison report records every U/UR row decision, family scale, near-zero branch,
+maximum absolute error, row relative error, scale-relative RMS, worst source
 row/component, and every UR warning.
 
 ## 6. Coverage and handoff
 
-The two cases are the complete required reference inventory for this feature:
+The S4 case is the complete required reference inventory for this feature:
 
-- `shell-s4` proves the approved `S4 -> FESA-MITC4` input path against Abaqus U;
-- `shell-s4r` proves the approved `S4R -> FESA-MITC4` input path against Abaqus U.
+- `shell-s4` checks the approved full-integration `S4 -> FESA-MITC4` input path against Abaqus U.
+- `S4R -> FESA-MITC4` support is verified by parser, common-kernel,
+  deterministic-assembly and HDF5 source-metadata tests without consuming an Abaqus
+  S4R reference artifact.
 
 They do not prove Abaqus formulation equivalence or general MITC4 accuracy outside
 the modeled cases. Additional flat/thin/thick/distorted/curved models, mesh studies,

docs/linear-static-mitc4-shell/release.md

@@ -0,0 +1,300 @@
+# Linear Static MITC4 Shell Release Report
+
+## Metadata
+
+- feature_id: `linear-static-mitc4-shell`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
+- source_research: `docs/linear-static-mitc4-shell/research.md`
+- source_formulation: `docs/linear-static-mitc4-shell/formulation.md`
+- source_numerical_review: `docs/linear-static-mitc4-shell/numerical-review.md`
+- source_io_definition: `docs/linear-static-mitc4-shell/io.md`
+- source_reference_model: `docs/linear-static-mitc4-shell/reference-model.md`
+- source_implementation_plan: `docs/linear-static-mitc4-shell/implementation-plan.md`
+- source_implementation_evidence: `phases/linear-static-mitc4-shell/index.json`
+- source_implementation_report: `N/A`; the project Harness phase index is the implementation completion ledger for this execution
+- source_correction_report: `N/A`; no final Build/Test, Reference Verification, or Physics Evaluation failure was handed to Correction Agent
+- source_build_test_report: `docs/linear-static-mitc4-shell/build-test.md`
+- source_reference_verification_report: `docs/linear-static-mitc4-shell/reference-comparison.md`
+- source_physics_evaluation_report: `docs/linear-static-mitc4-shell/physics-evaluation.md`
+- audited_source_head: `820ba30c717b3d0e113775608e20dfd5fbc05d53`
+- audited_branch: `feat-linear-static-mitc4-shell`
+- historical_status: `ready-for-release`
+- current_status: `superseded-pending-revalidation`
+- superseded_on: `2026-08-18`
+- owner_agent: `release-agent`
+- date: `2026-08-13`
+- release_boundary: internal FESA feature release readiness only; no publish, deploy, package, tag, commit, or external release was performed
+
+The readiness evidence below is historical. The 2026-08-18 common-policy reference revalidation
+passed, but fresh physics and release audits are still required before a current
+`ready-for-release` verdict may be issued.
+
+## Release Scope
+
+| item | included | excluded | notes |
+| --- | --- | --- | --- |
+| analysis type | One small-displacement, small-rotation `*STEP, *STATIC` | Multiple steps, geometric/material nonlinearity, buckling, modal, dynamic, contact, thermal | Existing factorize-before-load linear-static lifecycle is retained. |
+| element type | Four-node source `S4` and `S4R` mapped to one full-integration `FESA-MITC4` path | `S3`, `S8`, `S8R`, continuum/solid/axisymmetric shells, MITC4+, MITC4/D, reduced integration/hourglass behavior | Source type is preserved as metadata; no Abaqus formulation-equivalence claim is made. |
+| nodal degrees of freedom | Global `[UX,UY,UZ,URX,URY,URZ]`; 20 physical element coordinates plus four numerical drilling coordinates | Physical drilling strain/load/result channel | Director-parallel nodal moment is rejected. |
+| material and section | Centered, single-layer, homogeneous isotropic linear elasticity with finite valid `E`, `nu`, and constant positive thickness | Composite/laminated or orthotropic sections, offsets, variable/nodal thickness, user orientation | Multiple valid materials/sections/element sets are allowed with exactly one assignment per element. |
+| geometry | Valid four-node quadrilaterals and smooth shell patches with deterministic geometry-derived directors | Explicit normals/directors, shared-node sharp folds, hinges, shell-beam joints, calibrated distortion/warp thresholds | A physical fold requires duplicated source nodes. |
+| loads and constraints | Nodal `*BOUNDARY` and global nodal `*CLOAD` on DOFs 1-6, subject to the drilling-moment projection rule | `*DLOAD`, pressure, gravity, body/edge/follower load | Output-request keywords in the allowlist remain warning no-ops. |
+| authoritative output | HDF5 `results.h5`: nodal displacement/full residual, shell frames, generalized strain/resultant, bottom/middle/top in-plane stress, physical energy, equilibrium, verification metrics, diagnostics | Official CSV solver output, drilling output/energy, `S33`, pointwise `S13/S23`, nodal shell-stress averaging | HDF5 schema version remains `0` and is extended additively. |
+| external acceptance | Read-only `reference/shell/shell.inp` and `reference/shell/shell displacements.csv`; `U1/U2/U3` blocking at fixed absolute `1.0e-5`; `UR1/UR2/UR3` warning-only at the same value | S4R artifact consumption; Abaqus reaction/stress/strain/resultant equality | Direct HDF5-to-CSV source identity mapping is authoritative. |
+
+## Gate Evidence Inventory
+
+| gate | source | expected status/evidence | observed status/evidence | verdict |
+| --- | --- | --- | --- | --- |
+| requirements | `docs/linear-static-mitc4-shell/requirements.md` | approved scope, acceptance criteria, tolerance and reference boundary | `approved`; requirements `001-072` are all `must` and covered without gaps | pass |
+| research | `docs/linear-static-mitc4-shell/research.md` | approved evidence with applicability limits | `approved`; no research-owned blocking decision remains | pass |
+| formulation | `docs/linear-static-mitc4-shell/formulation.md` | implementation-ready current linear formulation | `approved-for-implementation-planning`; future nonlinear Section 15 remains explicitly non-executable | pass |
+| numerical review | `docs/linear-static-mitc4-shell/numerical-review.md` | `pass-for-implementation-planning` | `pass-for-implementation-planning`; no current-scope blocker | pass |
+| I/O definition | `docs/linear-static-mitc4-shell/io.md` | approved input/HDF5/comparison contract | `approved-for-implementation-planning`; exact S4-only acceptance boundary and fixed tolerance are present | pass |
+| reference model | `docs/linear-static-mitc4-shell/reference-model.md` | approved exact artifact inventory and row mapping | `approved-for-implementation-planning`; both declared files exist and match hashes | pass |
+| implementation plan | `docs/linear-static-mitc4-shell/implementation-plan.md` | approved TDD plan tracing every `must` requirement | `ready-for-implementation`; ranges cover `001-072` exactly once | pass |
+| implementation | `phases/linear-static-mitc4-shell/index.json` at source HEAD | completed RED/GREEN/VERIFY execution | Steps `0-13` are `completed`; all 14 `stepN-output.json` records have `exitCode=0`; top-level phase is `completed` | pass |
+| correction | final downstream reports and commit/Harness history | no unresolved implementation-owned failure | `N/A`; final Build/Test and Reference Verification classify correction handoff as `N/A`; historical pre-gate Step 13 retries are closed | pass |
+| build/test | `docs/linear-static-mitc4-shell/build-test.md` | `pass-for-reference-verification` | `pass-for-reference-verification`; clean MSVC x64 Debug build, focused `87/87`, `10/10`, `8/8`, full `144/144`, Harness Python `7/7`, zero warnings | pass |
+| reference verification | `docs/linear-static-mitc4-shell/reference-comparison.md` | `pass-for-physics-evaluation` | `pass-for-physics-evaluation`; exact `294/294` row identity, blocking U `147/147`, UR warnings `0`, invalid rows `0` | pass |
+| physics evaluation | `docs/linear-static-mitc4-shell/physics-evaluation.md` | `pass-for-release-agent` | `pass-for-release-agent`; equilibrium, reaction, direction, symmetry, recovery, stress signs, residual and physical energy pass | pass |
+
+### Gate Consistency and Staleness Audit
+
+- Every gate uses `feature_id=linear-static-mitc4-shell`. Build/Test, Reference Verification, and Physics Evaluation all identify source HEAD `820ba30c717b3d0e113775608e20dfd5fbc05d53`, which is the audited current HEAD.
+- The requirements metadata value `current_product_state=requirements-approved-not-implemented` records the product state when the requirements gate was authored. It is not the requirements gate status and does not alter the later implementation evidence.
+- Numerical Review values `implementation_complete=false`, `build_test_complete=false`, `reference_comparison_complete=false`, `physics_evaluation_complete=false`, and `release_ready=false` are explicit pre-implementation review boundaries. Later same-feature reports at the audited HEAD supply those downstream decisions. They are phase-time markers, not contradictory live statuses.
+- No contract meaning, reference inventory, tolerance, or source identity in those earlier documents conflicts with the final reports. The phase-time markers are therefore nonblocking and are not silently treated as current completion claims.
+- A separate implementation report is nonblocking under the current workflow because the Build/Test report explicitly declares `source_implementation_report=N/A` and points to the Harness index. The index records each RED failure, GREEN behavior, focused/full verification summary and completion timestamp, while all 14 generated step-output records have `exitCode=0`.
+- A correction report is nonblocking `N/A`: Correction Agent reporting is required after a formal downstream failure handoff. The final formal Build/Test, Reference Verification, and Physics Evaluation gates report no such failure or unresolved defect. Historical implementation-phase retries and correction commits precede the clean independent gates and are closed by the final phase ledger and source HEAD.
+- The Numerical Review metadata contains two historical commit references (`reviewed_head=cf769aa` and a downstream note naming `a058ef7`). This is a document-history inconsistency, but its mathematical verdict was re-audited against the current tracked requirements/formulation/I/O/reference contracts and independently exercised at `820ba30...`; it does not create a scope, tolerance, or acceptance contradiction.
+
+## Evidence Hash Inventory
+
+SHA-256 values are lowercase. The build-local HDF5 container is intentionally not a decision identity because valid regenerations can change raw container bytes; the deterministic projected ledger is the stable comparison evidence.
+
+| evidence | bytes | SHA-256 | audit role |
+| --- | ---: | --- | --- |
+| requirements | 35,124 | `c8966f667ee42c3d35d0bf2f1bc1662a93784a2037d5b191b59793fb5ad66c64` | approved must-requirement baseline |
+| research | 37,197 | `45f02291c8b2fdc7066cdb938cf6662a97dd3dceb9023fa743fe1c356a7970b4` | theory/benchmark/applicability evidence |
+| formulation | 57,593 | `117932a112b2ad7f1cb6a8084304733fcd529a675a93cca6364c2f6a16c57cd0` | mathematical implementation contract |
+| numerical review | 24,406 | `c82bd7b4f14f7664fb082d9a860dc443bf76571affa24caa8a7a9acfba04af1e` | independent formulation gate |
+| I/O definition | 41,327 | `6f362b16625e61c85949648e6bc1f215c92ff31e99654f6efd9787b428dd54c4` | input/HDF5/comparison contract |
+| reference model | 6,112 | `978e18485a627b6abf83373f74929cfe564952d29f68574e0cab75ef5e689c44` | exact S4 inventory and row mapping |
+| implementation plan | 56,316 | `8d269455d0d5fe9bb9dc32f140d1ea6f105bab60ebfe73d368ad8c5225af6482` | TDD task/test traceability |
+| Harness implementation index | 7,392 | `8c0a7369bc81fc7a12f9f3a49158a9534d4727b9a924c95b1ece7e915d16880d` | 14-step completion ledger |
+| Build/Test report | 11,970 | `d261e13eb1c6dc20cf24f85ee97a3fcb6b667420486ad12725030e3eebf8dcd1` | independent MSVC/CMake/CTest evidence |
+| Reference Verification report | 14,362 | `2c69814eeb265bd47672c46e17e2c63d55a0f8d834325c2b0d78e0500a83bade` | fixed-tolerance comparison evidence |
+| Physics Evaluation report | 19,596 | `d753521c73ac8509fe2d3e2896e2f6dd23a362c0f17f153fadbeb6227b8ed4c6` | physical plausibility evidence |
+| declared S4 input | 4,770 | `4005851e1ab22fd3a16ac17a8d5da3e051233f69f37419079f3553ad134ecfcf` | immutable reference input |
+| declared S4 displacement CSV | 5,592 | `c81d94e0b4a849f87aa0f79c83a79b94d5661ac79e44ed826919ab432c87746b` | immutable required reference quantity |
+| deterministic comparison ledger | 94,349 | `8e8dea51b6f7c663bacc41fda6103a4596db26e02f1ead6069d458f51e0102e6` | `294` row decisions, `passed=true`, zero warnings |
+
+## Acceptance Traceability
+
+Evidence abbreviations: `PH` = Harness phase index; `BT` = Build/Test report; `RV` = Reference Verification report; `PE` = Physics Evaluation report. `shell-s4` means the sole declared read-only S4 comparison model. Every row below has priority `must` in the approved requirements baseline.
+
+| requirement_id | acceptance criterion | test/phase evidence | reference/physics evidence | release disposition |
+| --- | --- | --- | --- | --- |
+| `001` | Exactly one supported static step; unsupported procedure or second step fails deterministically. | `InpDomainMapping.*Procedure*`; `Mitc4ShellCli.*`; PH Steps 1/12; BT | N/A | released |
+| `002` | S4 and S4R map to one FESA MITC4 path without equivalence claim. | `InpDomainMapping.MapsS4AndS4rThroughOneMitc4Identity`; PH Steps 0/1/7; BT | S4R is intentionally non-reference | released |
+| `003` | Source element type and internal formulation remain separate metadata. | `DomainModel.Mitc4ShellRecordsPreserveSourceAndInternalIdentity`; `Hdf5ResultsWriter.WritesExactShellMetadataAndModelIdentity`; BT | RV HDF5 projection audit | released |
+| `004` | Four distinct ordered nodes and stable source element identity are enforced. | `InpDomainMapping.RejectsInvalidShellConnectivityOptionsAndMixedModels`; Domain/HDF5 tests; BT | RV observed 36 uniquely identified S4 elements | released |
+| `005` | Six global DOFs use the exact order and equation ownership remains in DofManager. | `DofManager.BuildsShellScatterInSourceNodeAndComponentOrder`; HDF5 schema tests; PH Steps 0/6/11; BT | RV `[UX,UY,UZ,URX,URY,URZ]` projection | released |
+| `006` | Valid homogeneous isotropic `E,nu` is accepted and invalid/unsupported material data fails. | `InpDomainMapping.RejectsInvalidShellAssignmentsAndProperties`; `Mitc4ShellConstitutive.*`; BT | PE material audit | released |
+| `007` | One finite positive constant thickness is accepted for a centered single-layer shell section. | Domain/model/mapping tests; `Mitc4ShellConstitutive.*`; BT | PE thickness/model audit | released |
+| `008` | Each element resolves to exactly one section and material; missing/conflicting/duplicate assignments fail. | `DomainModel.ShellOwnershipPreservesResolvedAssignmentsAndOptionalFrames`; `InpDomainMapping.RejectsInvalidShellAssignmentsAndProperties`; BT | N/A | released |
+| `009` | Multiple valid sets/sections/materials are allowed while each element remains homogeneous and single-layer. | Domain/model/mapping suites; PH Steps 0/1; BT | N/A | released |
+| `010` | Composite, offset, variable thickness, explicit normal and orientation meanings fail closed. | `InpDomainMapping.RejectsInvalidShellConnectivityOptionsAndMixedModels`; BT | N/A | released |
+| `011` | Nodal director is a unit vector separate from scalar thickness. | `Mitc4Geometry.BuildsDeterministicFramesForPlanarRotatedAndWarpedElements`; HDF5 director tests; BT | PE finite/model audit | released |
+| `012` | Element normal candidate follows deterministic source order and tangent cross product. | `Mitc4Geometry.*`; PH Step 2; BT | N/A | released |
+| `013` | Smooth shared-node directors use deterministic area-weighted averaging. | `Mitc4Geometry.AreaWeightsSharedDirectorsInStableSourceIdentityOrder`; BT | N/A | released |
+| `014` | Nonfinite/zero/opposed normals fail without an invented smooth-angle threshold. | `Mitc4Geometry.RejectsInvalidSurfaceJacobianAndIncidentOrientationCases`; BT | N/A | released |
+| `015` | Discontinuous shared-node director fields are not silently averaged; folds require duplicate nodes. | Geometry negative tests and mapper unsupported-scope tests; BT | Documented limitation | released |
+| `016` | Duplicate, bow-tie, nonfinite, zero-area and nonpositive-J geometry fails at every required point. | `Mitc4Geometry.RejectsInvalidSurfaceJacobianAndIncidentOrientationCases`; `.ExposesTheCompleteRequiredValidationPointInventory`; BT | PE finite-result audit | released |
+| `017` | Nodal label/set boundary targets support global DOFs 1-6 and zero/nonzero prescribed values. | `EssentialConstraints.PreservesShellSizedNoMixedAndAllConstraintRoundTrips`; `Mitc4ShellCli.AppliesKfcForNonzeroPrescribedDisplacement`; BT | PE constraint-mask audit | released |
+| `018` | Global nodal force/moment CLOAD components aggregate deterministically. | `LoadAssembly.AggregatesAllSixGlobalShellLoadComponentsInSourceOrder`; BT | PE load/reaction audit | released |
+| `019` | Aggregate director-parallel moment fails as `unsupported-drilling-load`. | `LoadAssembly.AcceptsExactlyZeroAggregateShellMoment`; `.EnforcesAggregateShellMomentDirectorProjectionThreshold`; PH Step 8; BT | N/A | released |
+| `020` | Distributed/follower/body/edge load inputs remain unsupported. | `InpDomainMapping.PreservesProcedureLoadAndOutputRequestBoundariesForShells`; BT | Documented limitation | released |
+| `021` | Only the approved identity-preserving keyword subset plus S4/S4R and shell section is accepted. | `InpDomainMapping.*`; full CTest regression; BT | `shell-s4` parsed end-to-end | released |
+| `022` | Identity wrappers/multiple instances preserve source identity; transforms/nesting remain unsupported. | `InpDomainMapping.*`; Domain identity tests; BT | RV exact instance/node identity | released |
+| `023` | Allowlisted output requests do not affect mandatory results; other model-affecting meanings fail. | `InpDomainMapping.PreservesProcedureLoadAndOutputRequestBoundariesForShells`; `Hdf5ResultsWriter.WritesShellInventoryDespiteRequestsAndOmitsForbiddenPaths`; BT | PE observed warning-only diagnostics | released |
+| `024` | Domain owns shell semantics and AnalysisModel remains a non-owning active view. | `DomainModel.*`; PH Step 0 and full lifecycle Step 12; BT | N/A | released |
+| `025` | DofManager solely owns six-DOF numbering, maps, scatter and sparse pattern. | `DofManager.*`; `EssentialConstraints.*`; PH Step 6; BT | N/A | released |
+| `026` | AnalysisState owns only required linear-static shell solution/recovery/evidence rows. | `AnalysisState.OwnsExactShellRowsInStableElementAndLocationOrder`; `.CommitsFiniteShellGlobalEvidence`; BT | PE required-result audit | released |
+| `027` | Element buffers and stable COO/fixed reduction are deterministic across thread paths. | `SparseAssembly.ShellSerialTbbReverseAndRepeatedRunsAreByteIdentical`; PH Step 7; BT | N/A | released |
+| `028` | Stiffness/partition/factorization precedes load/effective RHS/substitution/recovery/commit. | `Mitc4ShellCli.UsesExistingLifecycleAndExactlyOneFactorization`; `.AppliesKfcForNonzeroPrescribedDisplacement`; `.DoesNotWriteAnInvalidRecoveryCandidate`; BT | PE final-state evidence | released |
+| `029` | Reaction/free equilibrium derives from full residual `K*d-F`. | `ResultRecovery.KeepsFullResidualAndComputesGlobalShellEquilibrium`; BT | PE reaction and equilibrium reconstruction | released |
+| `030` | S4/S4R use the same numerical path and may differ only in source metadata. | Mapper identity test; `SparseAssembly.S4AndS4rSemanticFixturesAssembleIdenticalStiffness`; HDF5 metadata tests; BT | S4R reference intentionally N/A | released |
+| `031` | Physical kernel uses three translations plus two tangent rotations; drilling is not physical strain. | `Mitc4ShellKinematics.BuildsRightHandedFramesAndSeparatePhysicalDrillingMaps`; drill/recovery tests; BT | PE no drilling leakage | released |
+| `032` | Deterministic right-handed frames transform global rotations to tangent/drilling coordinates. | `Mitc4ShellKinematics.BuildsRightHandedFramesAndSeparatePhysicalDrillingMaps`; `Mitc4Geometry.*`; BT | PE symmetry/direction evidence | released |
+| `033` | Drilling is a symmetric positive numerical stabilization only. | `Mitc4ShellDrilling.*`; `Mitc4ShellKernel.*`; BT | PE physical-energy audit | released |
+| `034` | `R+`, `k_ref`, `k_d=1e-3*k_ref`, and `I4` use only eight positive physical rotational diagonals. | `Mitc4ShellDrilling.UsesOnlyEightPositivePhysicalRotationDiagonalsAndFixedFactor`; BT | N/A | released |
+| `035` | Drilling contributes to no physical generalized strain/resultant/stress/output. | `Mitc4ShellDrilling.ExcludesPureDrillFromPhysicalRecoveryAndEnergy`; HDF5 forbidden-path test; BT | PE physical-only recovery/energy | released |
+| `036` | Empty/nonfinite valid `R+` fails; no coefficient/energy calibration gate or output exists. | `Mitc4ShellDrilling.FailsNonfiniteReferenceAndStabilizesEachPureDrillCoordinate`; BT | Documented limitation/boundary | released |
+| `037` | S4 and S4R share the approved full `2x2x2` rule; no reduced integration/hourglass selection. | `Mitc4ShellKinematics.UsesOneFixedTwoByTwoByTwoQuadratureOrder`; S4/S4R assembly parity; BT | RV S4-only boundary | released |
+| `038` | Stabilized element keeps six physical rigid modes, symmetry, energy invariance and positive non-rigid modes. | `Mitc4ShellKernel.*`; `Mitc4ShellPatch.*`; BT | PE rigid/rank/energy evidence | released |
+| `039` | HDF5 is authoritative and invalid candidates cannot replace a valid final result. | `Hdf5ResultsWriter.InvalidShellInventoryPreservesExistingFinal`; CLI invalid-candidate test; BT | RV/PE read authoritative HDF5 | released |
+| `040` | Required metadata/model identities, frames, directors, section/material and source/internal types are written. | `Hdf5ResultsWriter.WritesExactShellMetadataAndModelIdentity`; BT | RV HDF5 schema audit | released |
+| `041` | Every node has global six-component displacement and full-residual reaction rows in stable order. | HDF5 mandatory inventory; `ResultRecovery.*`; BT | RV 49x6 displacement; PE reaction audit | released |
+| `042` | Four shell locations contain ordered eight-component generalized strain with correct dimensions. | `Mitc4ShellPhysicalRecovery.*`; `ResultRecovery.RecoversShellRowsInStableElementAndGpOrder`; HDF5 tests; BT | PE 144-GP reconstruction | released |
+| `043` | The same locations contain ordered `N/M/Q` section resultants with correct dimensions. | Physical recovery/result recovery/HDF5 suites; BT | PE 1,152-component resultant reconstruction | released |
+| `044` | Bottom/middle/top local `[S11,S22,S12]` is emitted; S33/S13/S23 is not. | `ResultRecovery.RecoversDirectBottomMiddleTopShellStress`; HDF5 inventory/forbidden paths; BT | PE stress/location/sign reconstruction | released |
+| `045` | Rows preserve source element, exact natural/section location, frame and component order without averaging. | AnalysisState, ResultRecovery stable-order and HDF5 schema tests; BT | PE location-specific audit | released |
+| `046` | Free residual, force/moment balance and deterministic physical shell energy are output without drilling datasets. | `ResultRecovery.*ShellEquilibrium`; `.SumsOnlyPhysicalShellEnergyInSourceOrder`; HDF5 global rows; BT | PE equilibrium and energy checks | released |
+| `047` | Abaqus output requests neither suppress nor expand mandatory HDF5; CSV remains auxiliary. | Mapper output-request test; `Hdf5ResultsWriter.WritesShellInventoryDespiteRequestsAndOmitsForbiddenPaths`; BT | RV compared HDF5 directly; no FESA CSV | released |
+| `048` | Nonfinite/incomplete recovery or HDF5 inventory fails without partial success. | `AnalysisState.InvalidShellCandidatesLeavePriorStateUnchanged`; `ResultRecovery.InvalidLaterShellLeavesEntirePriorStateUnchanged`; HDF5 invalid-inventory test; BT | N/A | released |
+| `049` | Each production behavior has RED/GREEN/VERIFY evidence, related C++ tests, clean MSVC Debug build and no new warning. | PH Steps 0-13 and outputs; BT clean build, zero warnings, `144/144` | N/A | released |
+| `050` | Frames, Jacobians, symmetry, transform energy, six modes, positivity and repeatability are tested. | Geometry, Kinematics, Kernel, Patch and SparseAssembly suites; BT | PE rigid/rank summary | released |
+| `051` | Approved normalized `1e-12` and `1e-10` algebraic/equilibrium criteria are applied. | Kernel/frame tests; `ResultRecovery.UsesScaleAwareShellMetricsAndRejectsExcess`; BT | PE verification metrics all below `1e-10` | released |
+| `052` | Independent membrane, bending, shear and twist patch/sign/component checks pass. | `Mitc4ShellPatch.ReproducesIndependentMembraneBendingShearAndTwistFields`; PhysicalRecovery tests; BT | PE recovery sign checks | released |
+| `053` | Formulation invariants/patches and declared S4 displacement reference case pass; expanded portfolio is not required. | Kernel/Patch/PhysicalRecovery suites; `Mitc4S4Reference.*`; BT | RV blocking U `147/147`; PE coverage audit | released |
+| `054` | Exact geometry acceptance/rejection tests pass without NR-O03/NR-O04 calibration. | `Mitc4Geometry.*`; BT | N/A | released |
+| `055` | Additional curved/locking/convergence benchmarks remain optional and nonblocking. | Implementation-plan scope audit; BT/PE coverage inventory | PE records limitation and approved coverage | released |
+| `056` | Exact drilling formula, symmetry/positivity/rank/determinism and recovery exclusion are verified. | `Mitc4ShellDrilling.*`; Kernel/Sparse/Recovery suites; BT | PE drilling separation evidence | released |
+| `057` | Physics verifies equilibrium, direction, symmetry, energy, signs and resultant consistency. | ResultRecovery/CLI suites; BT | PE all documented checks pass | released |
+| `058` | Every matched U row uses fixed absolute `1.0e-5` and blocks on exceedance. | `Mitc4ReferenceComparison.AppliesFixedAbsoluteToleranceWithoutScaleClampOrRowDenominator`; `Mitc4S4Reference.*`; BT | RV U `147/147` pass | released |
+| `059` | Reference scale, row denominator, component magnitude and zero clamp never alter MITC4 tolerance. | Fixed-tolerance comparator test; BT | RV all 294 tolerances exactly `1.0e-5` | released |
+| `060` | U tolerance is in user-consistent length units and independent of B33 policy. | Comparator contract test and scope audit; BT | RV reports independent fixed policy | released |
+| `061` | UR uses fixed absolute `1.0e-5`; exceedance warns deterministically and never blocks. | `Mitc4ReferenceComparison.RotationExceedanceWarnsWithoutBlockingTranslationVerdict`; BT | RV UR `147/147`, zero warnings | released |
+| `062` | UR tolerance is dimensionless and no extra UR/drilling threshold is required. | Comparator/report tests; BT | RV fixed-tolerance metrics | released |
+| `063` | Missing/extra/duplicate/nonfinite/schema/identity mismatch fails before tolerance. | `Mitc4ReferenceComparison.RejectsInvalidInventoryBeforeNumericComparison`; BT | RV all invalid counts zero | released |
+| `064` | Report contains every U/UR decision, maximum, normalized, RMS, vector and worst-row evidence. | `Mitc4ReferenceComparison.ReportsMetricsVectorsWorstRowAndJsonDeterministically`; BT | RV complete quantity/component/worst-row tables | released |
+| `065` | Exact S4 input and displacement CSV paths are used without mutation. | `Mitc4S4Reference.*`; comparator required-file test; BT hash audit | `shell-s4`; immutable hashes match | released |
+| `066` | S4R artifacts are not required/consumed; parser/common-kernel/assembly/HDF5 tests cover S4R. | Mapper identity, S4/S4R assembly parity, HDF5 source-type tests; BT | RV confirms `reference/shellR/` not consumed | released |
+| `067` | Only declared input/CSV, FESA HDF5, mapping and tolerance/precheck rules gate readiness. | `Mitc4ReferenceComparison.RequiresOnlyDeclaredInputCsvAndHdf5`; BT | RV no administrative artifact gate | released |
+| `068` | Reference rows are unique, finite and deterministically source-mapped without internal-equivalence inference. | Comparator mapping and invalid-inventory tests; BT | RV exact `294/294` key set | released |
+| `069` | Sole Abaqus case is full-integration S4; S4R coverage remains non-reference. | Common quadrature/S4R mapping tests and `Mitc4S4Reference.*`; BT | RV one declared case, S4R excluded | released |
+| `070` | Authoritative FESA HDF5 rows are compared directly by model/step/frame/source/component. | Comparator mapping/E2E tests; BT | RV direct HDF5 projection, no FESA CSV | released |
+| `071` | Reaction/stress/strain/resultants do not become undeclared equality gates. | Comparator only projects U/UR; BT | RV and PE keep optional CSVs out of pass/fail | released |
+| `072` | No Abaqus/reference-solver execution or reference artifact mutation occurs. | PH/BT Git and hash audits | RV and PE no-change assertions; immutable hashes match | released |
+
+No requirement is deferred or blocked. Out-of-scope behavior is documented as a limitation rather than silently accepted.
+
+## Validation Evidence
+
+| command or report | expected | observed | verdict |
+| --- | --- | --- | --- |
+| config-resolved VS18/MSVC x64 Debug fresh configure and clean build | exit `0`, no new warning | exit `0`; warning/error scan `0/0` | pass |
+| focused model/kernel/assembly/recovery CTest | pass | `87/87` | pass |
+| focused linear-static lifecycle CTest | pass | `10/10` | pass |
+| focused MITC4 reference CTest | pass | `8/8` | pass |
+| CTest discovery and full regression | nonzero inventory and all pass | `144` discovered; `144/144` pass; current read-only discovery includes 57 test names containing `Mitc4` or `Shell` | pass |
+| Harness Python verification | pass because Harness runner changed on the feature branch | `7/7` pass | pass |
+| final release-stage confirmation | current HEAD build, full CTest and Harness self-test pass | fresh `cmake --build` exit `0`; CTest `144/144`; Harness pytest `7/7` | pass |
+| Harness implementation ledger | every Step completed successfully | Steps `0-13` completed; `14/14` output `exitCode=0` | pass |
+| reference verification | `pass-for-physics-evaluation` | exact 294 rows; U `147/147`; invalid rows/warnings `0` | pass |
+| physics evaluation | `pass-for-release-agent` | all equilibrium/direction/symmetry/recovery/energy checks pass | pass |
+| immutable reference audit | exact expected hashes, no Git reference diff/status | both hashes match; zero changed reference paths | pass |
+
+The Release Agent did not run Abaqus. It independently read the complete upstream evidence chain, enumerated current CTest discovery read-only, checked all Harness output exit codes, verified source HEAD/branch/status, and recomputed the declared artifact/report hashes. The primary audit then performed a fresh current-HEAD incremental full build, full CTest run and Harness self-test as the final completion check; all passed as recorded above.
+
+## Release Checklist
+
+- [x] One `feature_id` across all required gate documents.
+- [x] Requirements, research, formulation, numerical review, I/O, reference-model and implementation-plan gates are approved/passing for their intended downstream phase.
+- [x] Harness implementation is complete at the audited source HEAD with 14 successful Step outputs.
+- [x] Independent Build/Test status is `pass-for-reference-verification`.
+- [x] Reference Verification status is `pass-for-physics-evaluation`.
+- [x] Physics Evaluation status is `pass-for-release-agent`.
+- [x] All `must` requirements `001-072` trace to acceptance criteria, test/phase evidence, reference/physics evidence where applicable, and a released disposition.
+- [x] Sole required comparison files exist, match approved hashes, and were not modified.
+- [x] Fixed MITC4 tolerance and U-blocking/UR-warning-only policy are unchanged.
+- [x] No unresolved build, test, reference, physics, schema, identity, nonfinite or environment failure remains.
+- [x] Known limitations, unsupported Abaqus meanings and accepted risks are documented below and in the Release Notes Draft.
+- [x] Release Agent changed only this report and performed no external release action.
+
+## Known Limitations
+
+| limitation | category | user impact | disposition |
+| --- | --- | --- | --- |
+| One small-displacement/small-rotation linear-static step only | analysis | Nonlinear, buckling, modal, dynamic, contact and thermal models are rejected/not executed. | documented |
+| S4 and S4R are source labels for one full-integration FESA-MITC4 formulation | input/numerical | No Abaqus S4/S4R internal algorithm, reduced-integration or hourglass equivalence is claimed. | documented |
+| External acceptance uses one flat symmetric fully clamped S4 plate | verification | General curved/distorted/thin-thick convergence is not established by the source-solver comparison. | accepted risk; optional future evidence |
+| Original MITC4 membrane field is unmodified | numerical | Distorted curved meshes may exhibit membrane locking; MITC4+ performance is not claimed. | documented |
+| S4R has no blocking Abaqus reference comparison | verification | S4R support is proven by parser/common-kernel/deterministic-assembly/HDF5 metadata tests only. | approved boundary |
+| Homogeneous isotropic centered single layer with constant thickness | material/section | Composite, orthotropic, offset and variable-thickness models are unsupported. | documented |
+| Smooth geometry-derived director field only | geometry | Explicit normals and shared-node folds/hinges/intersections are unsupported; folds require duplicated nodes. | documented |
+| Nodal loads/constraints only | load | Pressure, gravity, DLOAD, body/edge/follower load are unsupported. Director-parallel moments fail as `unsupported-drilling-load`. | documented |
+| Fixed numerical drilling stabilization | numerical | It is not a physical strain/load/result channel; no calibration sweep or drilling-energy ratio is supplied. | accepted fixed policy |
+| In-plane stress only at bottom/middle/top | output | `S33` is a plane-stress assumption; `S13/S23` point stress and nodal averaging are not emitted. | documented |
+| Raw HDF5 container bytes are not a declared deterministic identity | output/evidence | Regeneration may change the raw file hash; ordered projected rows and the JSON comparison ledger are deterministic evidence. | documented, nonblocking |
+| Target-wide CTest label remains the legacy beam label | test infrastructure | MITC4 selection uses the approved exact suite regexes instead of a dedicated label. | documented, nonblocking |
+
+## Release Notes Draft
+
+### Feature Summary
+
+- Adds end-to-end single-step linear-static analysis for four-node shell inputs labeled `S4` or `S4R`, both mapped to the independent full-integration `FESA-MITC4` formulation.
+- Adds deterministic geometry-derived nodal directors, six global nodal DOFs, MITC4 transverse-shear tying, fixed nonphysical drilling stabilization, deterministic sparse assembly, full-residual reactions, physical shell recovery and failure-atomic HDF5 output.
+- Adds mandatory HDF5 nodal displacement/reaction, shell local frames, generalized strain/resultant, bottom/middle/top in-plane stress, physical energy, equilibrium and verification metrics.
+
+### Verification Scope
+
+- Clean Visual Studio 18 2026/MSVC x64 Debug build and full `144/144` CTest regression passed with zero compiler/linker warnings.
+- All 14 Harness implementation Steps completed with recorded RED/GREEN/VERIFY evidence and successful output exit codes.
+- The sole declared S4 case matched all `147` blocking translational rows within fixed absolute `1.0e-5`; all `147` rotation rows were within the warning threshold and produced no warnings.
+- Physics checks passed for global force/moment equilibrium, full-residual reactions, displacement direction, double symmetry, generalized resultant/stress consistency, free residual, finite results and positive physical energy.
+
+### Main Limitations
+
+- The released scope is linear static, small deformation, homogeneous isotropic single-layer MITC4 with nodal loads and constraints only.
+- Abaqus compatibility and Abaqus S4/S4R formulation equivalence are not claimed. S4R uses the same FESA full-integration kernel and is not an Abaqus reference gate.
+- Broader curved/distorted/thin-thick convergence, original-MITC4 membrane-locking characterization, nonlinear execution, distributed loads, composites, physical drilling behavior and expanded stress output remain outside this release.
+
+### Artifacts
+
+- Release report: `docs/linear-static-mitc4-shell/release.md`
+- Build/Test report: `docs/linear-static-mitc4-shell/build-test.md`
+- Reference Verification report: `docs/linear-static-mitc4-shell/reference-comparison.md`
+- Physics Evaluation report: `docs/linear-static-mitc4-shell/physics-evaluation.md`
+- Declared reference pair: `reference/shell/shell.inp`, `reference/shell/shell displacements.csv`
+- Build-local deterministic ledger: `.harness/build/reference/mitc4-shell-s4-comparison/comparison.json`
+
+## Release Verdict
+
+- verdict: `ready-for-release`
+- reason: all required upstream gates are present and passing; the final three mandatory handoff statuses are exactly `pass-for-reference-verification`, `pass-for-physics-evaluation`, and `pass-for-release-agent`; every `must` requirement `001-072` has a released disposition; required S4 artifacts and row mappings are present and valid; no unresolved defect, missing comparison file, tolerance conflict, nonfinite result, identity gap, or physics blocker remains.
+- approval_scope: internal feature release closure only
+- external_release_action: `none`
+
+## Handoff Recommendation
+
+| target_agent | reason | required_input |
+| --- | --- | --- |
+| Coordinator Agent | Close the feature workflow with `ready-for-release`. | This report, source HEAD `820ba30c717b3d0e113775608e20dfd5fbc05d53`, and the three passing downstream gate reports. |
+| Correction Agent | N/A; no implementation-owned release blocker remains. | N/A |
+| Reference Verification Agent | N/A; the required S4 comparison is complete and passing. | N/A |
+| Physics Evaluation Agent | N/A; physics evaluation is complete and passing. | N/A |
+
+## No-Change Assertion
+
+- source_files_modified: `false`
+- test_files_modified: `false`
+- cmake_files_modified: `false`
+- requirements_modified: `false`
+- research_modified: `false`
+- formulation_modified: `false`
+- numerical_review_modified: `false`
+- io_contract_modified: `false`
+- reference_model_contract_modified: `false`
+- implementation_plan_or_phase_files_modified: `false`
+- build_test_report_modified: `false`
+- reference_verification_report_modified: `false`
+- physics_evaluation_report_modified: `false`
+- reference_artifacts_modified: `false`
+- tolerance_policies_modified: `false`
+- Abaqus_or_other_reference_solver_executed: `false`
+- publish_deploy_package_tag_commit_performed: `false`
+- owned_report_created: `true`
+- pre_existing_untracked_reports_preserved: `true`
+- notes: before this report was created, the worktree had no tracked/staged diff and contained only the three upstream gate reports as untracked files. This audit adds only `docs/linear-static-mitc4-shell/release.md`; generated build-local evidence remains ignored under `.harness/build/`.
+
+## Open Issues
+
+- blocking: `none`
+- deferred requirements: `none`
+- unresolved defects: `none`
+- missing required comparison files or mappings: `none`
+- nonblocking administrative note: the Build/Test, Reference Verification, Physics Evaluation and this Release report are untracked workspace reports pending whatever repository-integration action the user or Coordinator Agent separately authorizes. Their untracked state does not invalidate their content or the current internal readiness verdict.
+- nonblocking evidence note: raw `results.h5` hashes can differ across valid regenerations; no byte-identical HDF5 container claim is made.

docs/linear-static-mitc4-shell/requirements.md

@@ -6,11 +6,11 @@
 - title: `Linear Static MITC4 Shell`
 - status: `approved`
 - owner_agent: `requirement-agent`
-- date: `2026-08-12`
+- date: `2026-08-13`
 - approval_basis: 사용자와 확정한 선형 정적 범위, `S4`/`S4R` 매핑, 6자유도 외부 계약, drilling 안정화, 자동 director 생성, 결과 및 검증 계약
 - current_product_state: `requirements-approved-not-implemented`
-- formulation_alignment: `docs/formulations/mitc4-shell-formulation.md`는 이 baseline의 6자유도 및 고정 drilling 안정화 계약과 정렬함
-- reference_inventory_state: `reference/shell/`의 S4와 `reference/shellR/`의 S4R input/displacement CSV를 기존 경로와 이름 그대로 사용함
+- formulation_alignment: `docs/linear-static-mitc4-shell/formulation.md`는 이 baseline의 6자유도 및 고정 drilling 안정화 계약과 정렬함
+- reference_inventory_state: full-integration FESA-MITC4의 Abaqus acceptance comparison은 `reference/shell/`의 S4 input/displacement CSV만 기존 경로와 이름 그대로 사용함; S4R source support는 reference artifact 없이 mapping/kernel/HDF5 tests로 검증함
 
 ## Purpose
 
@@ -39,7 +39,7 @@ Formulation, Numerical Review, I/O, Reference Model, Implementation Planning 및
 - `docs/PRD.md`, `docs/ARCHITECTURE.md`, `docs/ADR.md`: end-to-end feature boundary,
   ownership, linear-static lifecycle, deterministic assembly, HDF5, reference immutability 및
   failure atomicity
-- `docs/formulations/mitc4-shell-formulation.md`: 후속 정렬이 필요한 선행 draft이며 이
+- `docs/linear-static-mitc4-shell/formulation.md`: 후속 정렬이 필요한 선행 draft이며 이
   approved requirements baseline을 변경하는 근거로 사용하지 않음
 
 ## In Scope
@@ -59,7 +59,7 @@ Formulation, Numerical Review, I/O, Reference Model, Implementation Planning 및
 - DOF 1~6의 nodal `*BOUNDARY`와 nodal `*CLOAD`
 - deterministic assembly와 기존 linear-static partition/factorization/substitution lifecycle
 - HDF5 nodal displacement/reaction, shell generalized strain/resultant, in-plane stress, residual 및 energy output
-- element invariant, patch, 현재 S4/S4R displacement reference 및 physics verification
+- element invariant, patch, 현재 S4 displacement reference 및 physics verification
 
 ## Out Of Scope
 
@@ -165,7 +165,7 @@ Formulation, Numerical Review, I/O, Reference Model, Implementation Planning 및
 - **FESA-REQ-LSMITC4-050** — Element invariant tests shall cover frame orthonormality/handedness, Jacobian sign, stiffness symmetry, coordinate-transformation energy invariance, six physical rigid modes, deformation-mode positivity and deterministic repeatability.
 - **FESA-REQ-LSMITC4-051** — Normalized algebraic acceptance shall use `1e-12` for symmetry, frame orthonormality and transformation-energy invariance and `1e-10` for rigid-mode action, linear-system residual and global equilibrium unless Numerical Review approves and documents an evidence-backed scale-aware replacement before Implementation Planning.
 - **FESA-REQ-LSMITC4-052** — Patch verification shall independently cover constant membrane strain/stress, pure bending, transverse shear and twist, including sign and component-order checks for generalized strain, resultant and recovered in-plane stress.
-- **FESA-REQ-LSMITC4-053** — The implementation shall pass the formulation-defined element invariants and patch/manufactured tests plus the two declared S4/S4R displacement reference cases; an expanded locking, distortion or curved-shell benchmark portfolio is not an implementation-completion gate for this feature.
+- **FESA-REQ-LSMITC4-053** — The implementation shall pass the formulation-defined element invariants and patch/manufactured tests plus the declared S4 displacement reference case; an expanded locking, distortion or curved-shell benchmark portfolio is not an implementation-completion gate for this feature.
 - **FESA-REQ-LSMITC4-054** — Geometry validation tests shall cover the exact accepted/rejected conditions defined by the formulation and I/O contract; `NR-O03` smooth-director calibration and `NR-O04` distortion/warp threshold sweeps are not required tests.
 - **FESA-REQ-LSMITC4-055** — Additional pinched-cylinder, hemispherical-shell, Scordelis-Lo or mesh-convergence studies may be added later as nonblocking research or release evidence, but are not required for Implementation Planning or feature completion.
 - **FESA-REQ-LSMITC4-056** — Drilling verification shall check the fixed formula in Requirement 034, symmetry, positivity, deterministic assembly, removal of the four nonphysical local drilling null modes, and exclusion from physical recovery; coefficient sweeps and drilling-energy checks are not required.
@@ -184,24 +184,25 @@ Formulation, Numerical Review, I/O, Reference Model, Implementation Planning 및
 
 ## Tolerance Policy
 
-- **FESA-REQ-LSMITC4-058** — Abaqus reference pass/fail shall apply only to matched global `U1/U2/U3` rows using `tolerance_c = 1e-9 + 1e-6 * reference_scale_c`, exactly reusing the approved B33 displacement rule.
-- **FESA-REQ-LSMITC4-059** — `reference_scale_c` shall be computed only from finite Abaqus values in the same model, step/frame, quantity and component group; reference values shall not be zero-clamped and row-specific relative denominators shall not replace the group scale.
-- **FESA-REQ-LSMITC4-060** — The `1e-9` absolute floor is expressed in the model's user-consistent length unit for U; no additional MITC4 tolerance calibration is required.
-- **FESA-REQ-LSMITC4-061** — Global `UR1/UR2/UR3` rows shall use the same component-scale formula as Requirement 058 and shall be fully reported; an exceedance emits a deterministic nonblocking warning and never changes pass/fail.
-- **FESA-REQ-LSMITC4-062** — The `1e-9` UR floor is dimensionless. No separate UR large-error or drilling-energy threshold is required.
+- **FESA-REQ-LSMITC4-058** — Abaqus reference pass/fail shall apply only to matched global `U1/U2/U3` rows grouped as one blocking translation family with Abaqus-only scale `S=max(abs(reference))`.
+- **FESA-REQ-LSMITC4-059** — A U row with `abs(reference)<=0.01*S` shall pass when `abs(fesa-reference)<=0.01*S`; every other U row shall pass when its relative error is at most `0.05`; the U family shall also satisfy `RMS(error)/S<=0.01`.
+- **FESA-REQ-LSMITC4-060** — MITC4 shall use the project-wide family-scale policy without an independent absolute-error gate, zero clamp or row omission; a zero-scale family passes only when every FESA value is exactly zero.
+- **FESA-REQ-LSMITC4-061** — Global `UR1/UR2/UR3` rows shall form one warning-only rotation family using the same near-zero, relative-row and scale-relative RMS constants; an exceedance emits a deterministic nonblocking warning and never changes pass/fail.
+- **FESA-REQ-LSMITC4-062** — No separate UR large-error, fixed absolute or drilling-energy threshold is required; every row decision and family metric shall remain finite and auditable.
 - **FESA-REQ-LSMITC4-063** — Missing, extra, duplicate, nonfinite, schema-mismatched or source-identity-mismatched rows shall fail artifact/schema validation before numeric tolerance evaluation for both U and UR inventories.
-- **FESA-REQ-LSMITC4-064** — The comparison report shall record each U/UR row decision, maximum absolute error, component-scale normalized error, RMS error, vector-norm error and worst source row/component; nonblocking UR warnings shall not be omitted from an otherwise passing report.
+- **FESA-REQ-LSMITC4-064** — The comparison report shall record each U/UR row branch/decision, family identity/components, scale, near-zero band/count, maximum absolute error, scale-relative RMS, vector-norm diagnostic, worst source row/component and zero-scale diagnostic; nonblocking UR warnings shall not be omitted from an otherwise passing report.
 
 ## Reference Artifact Requirements
 
-The approved lightweight inventory uses the existing S4 and S4R paths below. These files
-remain read-only; their names are identities, not canonical/legacy-policy decisions.
+The approved lightweight acceptance inventory uses only the existing S4 paths below.
+Every file under `reference/` remains read-only; its name is an identity, not a
+canonical/legacy-policy decision.
 
 - **FESA-REQ-LSMITC4-065** — The S4 case shall use `reference/shell/shell.inp` and `reference/shell/shell displacements.csv` without creating, renaming, rewriting or repairing either file.
-- **FESA-REQ-LSMITC4-066** — The S4R case shall use `reference/shellR/shellR.inp` and `reference/shellR/shellR displacements.csv` without creating, renaming, rewriting or repairing either file.
+- **FESA-REQ-LSMITC4-066** — No S4R Abaqus artifact shall be required or consumed by this feature's reference acceptance comparison. S4R source support shall instead be verified by parser, common-kernel, deterministic assembly and HDF5 source-metadata tests; any existing `reference/shellR/` files remain untouched optional evidence.
 - **FESA-REQ-LSMITC4-067** — Reference readiness requires only the declared input and required displacement CSV, FESA `results.h5`, deterministic source-node/component mapping, and Requirements 058-063 tolerance/precheck rules. README, `metadata.json`, canonical naming, provenance, Abaqus version, duplicated model semantics, and a schema version are not required gates; a present `metadata.json` is optional read-only context.
 - **FESA-REQ-LSMITC4-068** — Artifact validation shall require unique finite displacement rows and deterministic source-node/component identity before comparison; it shall not attempt to establish Abaqus internal formulation equivalence.
-- **FESA-REQ-LSMITC4-069** — The two declared cases satisfy the required source-label coverage: `reference/shell/` covers S4 and `reference/shellR/` covers S4R. No expanded reference portfolio is required for this feature.
+- **FESA-REQ-LSMITC4-069** — The sole declared Abaqus comparison case is `reference/shell/` S4 because FESA-MITC4 uses full `2 x 2 x 2` integration. S4R source-label coverage is satisfied by the non-reference tests in Requirement 066; no expanded reference portfolio is required for this feature.
 - **FESA-REQ-LSMITC4-070** — Reference verification shall compare FESA HDF5 global nodal displacement rows directly against Abaqus displacement CSV rows by model, step/frame, source node and component identity; a FESA-generated CSV view shall not become the authoritative comparison source.
 - **FESA-REQ-LSMITC4-071** — Abaqus reaction, stress, strain and shell force/moment output may be retained as review evidence if present but shall not change the approved `U1/U2/U3` pass/fail boundary or become an undeclared equality gate.
 - **FESA-REQ-LSMITC4-072** — FESA agents shall not execute Abaqus or another reference solver and shall not generate, modify, restore or normalize reference artifacts during requirements, research, formulation, implementation or verification unless a later phase explicitly authorizes that operation.
@@ -224,12 +225,12 @@ without gaps or overlap.
 | `024-030` | Ownership, deterministic assembly and linear-static execution | architecture/execution | PRD; ADR-004/007/008/009/016/017 | must | Unit, orchestration and repeated-thread-count tests | Ownership boundaries, event order, residual reaction and deterministic bytes/rows match | `1e-12` deterministic numeric target where applicable | Numerical Review; Implementation Planning | approved |
 | `031-038` | 5-DOF physics embedded in 6-DOF with fixed drilling stabilization | numerical boundary | User approval; MITC literature and thesis 6-DOF discussion | must | Formulation review, invariant and rank tests | Exact `10^-3` positive rotational-diagonal rule; physical outputs exclude drilling | Fixed by Requirements 033-036 | Formulation; Numerical Review; Implementation Planning | approved |
 | `039-048` | Mandatory HDF5 output and failure atomicity | output | User approval; ADR-005/016/018 | must | Recovery, schema, identity, nonfinite and atomicity tests | Every quantity/location/unit/identity exists; failure commits no partial success | Exact component/location inventory; I/O Definition owns schema | Formulation; I/O Definition; Implementation Planning | approved |
-| `049-057` | TDD, invariants, patch, declared references and physics | verification | User approval; shell formulation evidence; project process | must | CTest evidence, analytical/patch tests, two reference cases and physics review | Required tests pass; removed calibration/portfolio checks are not reintroduced | `1e-12` symmetry/frame; `1e-10` rigid/residual | Numerical Review; Implementation Planning | approved |
-| `058-060` | Translational displacement pass/fail tolerance | tolerance | User approval; B33 baseline | must | Comparator unit/integration tests and report review | Every matched U row uses `1e-9 + 1e-6*reference_scale_c` without clamp/omission | Fixed by Requirements 058-060 | Reference Verification | approved |
-| `061-062` | Rotational warning-only comparison | tolerance/warning | User approval; B33 baseline | must | Comparator/diagnostic tests and report review | UR never changes pass/fail; same mixed-tolerance exceedance emits a deterministic warning | Fixed by Requirements 061-062 | Reference Verification | approved |
+| `049-057` | TDD, invariants, patch, declared reference and physics | verification | User approval; shell formulation evidence; project process | must | CTest evidence, analytical/patch tests, the S4 reference case and physics review | Required tests pass; removed calibration/portfolio checks are not reintroduced | `1e-12` symmetry/frame; `1e-10` rigid/residual | Numerical Review; Implementation Planning | approved |
+| `058-060` | Translational displacement pass/fail tolerance | tolerance | User approval; ADR-022 | must | Comparator unit/integration tests and report review | Every matched U row and U-family RMS use the common family-scale policy without clamp or omission | `0.01`, `0.05`, `0.01` | Reference Verification | approved |
+| `061-062` | Rotational warning-only comparison | tolerance/warning | User approval; ADR-022 | must | Comparator/diagnostic tests and report review | UR never changes pass/fail; common-policy exceedance emits a deterministic warning | `0.01`, `0.05`, `0.01` | Reference Verification | approved |
 | `063-064` | Row/schema failure and report completeness | reference verification | User approval; ADR-005/014/018 | must | Negative comparator and report-schema tests | Invalid inventory fails before numeric comparison; all U/UR metrics remain visible | No ignored invalid rows | I/O Definition; Reference Verification | approved |
-| `065-068` | Exact S4/S4R reference-case inventory and row validity | reference | User declaration; ADR-019 | must | Read-only inventory and source-row/component precheck | Four declared paths exist; required rows are unique, finite and deterministically mapped | Requirements `058-063` | Reference Model; Reference Verification | approved |
-| `069-071` | S4/S4R coverage and displacement-only comparison | reference | User approval | must | HDF5-to-CSV comparison | Declared S4/S4R cases; only U blocks and UR only warns | Requirements `058-064` | Reference Verification; Physics Evaluation | approved |
+| `065-068` | Exact S4 reference-case inventory, S4R exclusion and row validity | reference | User declaration; ADR-019 | must | Read-only S4 inventory, source-row/component precheck and S4R non-consumption test | Two declared S4 paths exist; required rows are unique, finite and deterministically mapped; S4R artifacts are not required or consumed | Requirements `058-063` | Reference Model; Reference Verification | approved |
+| `069-071` | S4 reference coverage and displacement-only comparison | reference | User approval | must | HDF5-to-CSV comparison | Declared S4 case only; U blocks and UR only warns; S4R mapping remains independently tested | Requirements `058-064` | Reference Verification; Physics Evaluation | approved |
 | `072` | Reference solver/artifact immutability | governance | User/project policy; ADR-010 | must | Process audit and Git diff | No unapproved execution or artifact mutation | Exact zero mutations | All downstream agents | approved |
 
 ## Open Questions and Required Downstream Decisions
@@ -249,7 +250,7 @@ and tangent derivation may remain in the formulation document.
 
 ### Formulation Agent
 
-- Revise `docs/formulations/mitc4-shell-formulation.md` to align with global 6-DOF input/output and a physical 5-DOF MITC4 kernel plus numerical drilling embedding.
+- Revise `docs/linear-static-mitc4-shell/formulation.md` to align with global 6-DOF input/output and a physical 5-DOF MITC4 kernel plus numerical drilling embedding.
 - Keep current-product equations strictly linear static; retain geometric-nonlinear residual/tangent only as clearly separated future formulation.
 - Define local frames, transformations, generalized component order, quadrature/tying, stress/resultant recovery and consistent units/signs.
 - Do not introduce distributed-load product support or make `S4R` select reduced integration.
@@ -267,12 +268,13 @@ and tangent derivation may remain in the formulation document.
 
 ### Reference Model Agent
 
-- Record the exact existing S4 and S4R input/displacement CSV paths from Requirements 065-066 and keep them read-only.
-- Define only the HDF5-to-CSV source-node/component projection and the already approved B33 mixed tolerance; do not add bundle administration or portfolio gates.
+- Record only the exact existing S4 input/displacement CSV paths from Requirement 065 as acceptance artifacts and keep every existing reference artifact read-only.
+- Do not consume the S4R bundle in reference verification; route S4R source support to parser/common-kernel/HDF5 tests from Requirement 066.
+- Define only the HDF5-to-CSV source-node/component projection and the approved common family-scale tolerance; do not add bundle administration or portfolio gates.
 
 ### Implementation Planning Agent
 
 - Do not start until Research, revised Formulation, Numerical Review, I/O and lightweight Reference Model inventory are mutually consistent.
 - Trace every `must` requirement to RED/GREEN/VERIFY tests and preserve current solver ownership, deterministic assembly and failure-atomic HDF5 boundaries.
 - Use the project Harness skill to propose self-contained implementation Steps for user approval, then write only the approved phase-planning files; do not execute Harness without a separate explicit request.
-- Include tests for source-type mapping, auto directors, fixed drilling rank/separation, required recovery quantities, row failures and U-versus-UR comparison behavior; exclude coefficient sweeps, drilling energy and `NR-O03`/`NR-O04`.
+- Include tests for source-type mapping, auto directors, fixed drilling rank/separation, required recovery quantities, row failures and S4-only U-versus-UR comparison behavior; retain non-reference S4R source mapping coverage and exclude coefficient sweeps, drilling energy and `NR-O03`/`NR-O04`.

docs/linear-static-mitc4-shell/research.md

@@ -3,13 +3,13 @@
 ## Metadata
 
 - feature_id: `linear-static-mitc4-shell`
-- source_requirement: `docs/requirements/linear-static-mitc4-shell.md`
+- source_requirement: `docs/linear-static-mitc4-shell/requirements.md`
 - status: `approved`
 - owner_agent: `research-agent`
-- date: `2026-08-12`
+- date: `2026-08-13`
 - product_scope: small-displacement, small-rotation, single-step linear static analysis
 - evidence_route: local papers in `docs/reference-papers/MITC4/`, the configured FEM wiki, original peer-reviewed papers, and official Abaqus documentation
-- reference_inventory_state: existing read-only S4 case at `reference/shell/` and S4R case at `reference/shellR/`; exact comparison paths are fixed by the approved requirements
+- reference_inventory_state: the sole acceptance comparison is the existing read-only full-integration S4 case at `reference/shell/`; S4R source support is verified without consuming an Abaqus reference artifact
 - source_policy: each external claim below is assigned a reliability tier; FESA decisions are labeled `Project contract`, and derived recommendations are labeled `Inference` or `Research recommendation`
 
 This brief supplies evidence to the Formulation, Numerical Review, I/O Definition, and
@@ -25,7 +25,7 @@ formulation-equivalent to Abaqus S4 or S4R.
 4. What evidence and dimensional restriction support the approved fixed drilling stabilization without turning it into a physical strain or load channel?
 5. What evidence supports connectivity-derived thickness directions, nodal-normal smoothing, local tangent frames, and geometry rejection?
 6. What may and may not be inferred when Abaqus S4 and S4R input types are both mapped to one FESA MITC4 formulation?
-7. Which element-level checks and the declared S4/S4R source-solver cases fit the approved implementation scope?
+7. Which element-level checks and the declared full-integration S4 source-solver case fit the approved implementation scope while retaining non-reference S4R mapping coverage?
 
 ## Source Reliability Tiers
 
@@ -52,7 +52,7 @@ formulation-equivalent to Abaqus S4 or S4R.
 | S12 | Abaqus, [LE3 Hemispherical Shell with Point Loads](https://docs.software.vt.edu/abaqusv2024/English/SIMACAEBMKRefMap/simabmk-c-le3.htm) and [The Pinched Cylinder Problem](https://docs.software.vt.edu/abaqusv2025/English/SIMACAEBMKRefMap/simabmk-c-pinchcyl.htm) | Tier 1 | authoritative point-load-compatible curved-shell benchmark definitions, target displacements, S4/S4R convergence, and distorted-mesh evidence | official input decks contain semantics such as explicit normals or symmetry shorthand that require an approved FESA-subset adaptation |
 | S13 | Abaqus, [Shell Thickness and Section Points](https://docs.software.vt.edu/abaqusv2024/English/SIMACAEGSARefMap/simagsa-c-shlthick.htm) and [Whole and Partial Model Variables](https://docs.software.vt.edu/abaqusv2024/English/SIMACAEOUTRefMap/simaout-c-std-wholeandpartialmodelvariables.htm) | Tier 1 | bottom/middle/top linear-elastic stress recovery precedent and separate reporting of artificial energy that includes drill constraints | no drill-specific acceptable artificial-energy percentage is stated |
 | S14 | configured FEM wiki pages `[[MITC4 Shell Element]]`, `[[MITC Shell Kinematics]]`, `[[Assumed Transverse Shear Strain Interpolation]]`, `[[Shell Locking Phenomenon]]`, `[[Shell Element Benchmark Testing]]`, and `[[Scordelis-Lo Shell Benchmark]]` | Tier 3 synthesis | navigation between local raw sources; locking, tying, and benchmark terminology | key claims are cited to S1–S13 rather than relying on the wiki alone |
-| P1 | `docs/requirements/linear-static-mitc4-shell.md`, `docs/PRD.md`, `docs/ARCHITECTURE.md`, and `docs/ADR.md` | Project contract | exact FESA scope, six-global-DOF interface, source identity, output, verification, lifecycle, and reference immutability | does not prove numerical correctness by itself |
+| P1 | `docs/linear-static-mitc4-shell/requirements.md`, `docs/PRD.md`, `docs/ARCHITECTURE.md`, and `docs/ADR.md` | Project contract | exact FESA scope, six-global-DOF interface, source identity, output, verification, lifecycle, and reference immutability | does not prove numerical correctness by itself |
 
 The informal `docs/reference-papers/MITC4/MITC공부/` notes were used only as a
 navigation aid. No key numerical decision relies on them.
@@ -146,7 +146,7 @@ problems must not be mixed.
 - **Locking and convergence:** thin/thick cantilevers, pinched cylinder, LE3, and Scordelis–Lo remain useful future studies. They are not additional completion gates for the approved two-case implementation scope.
 - **Geometry verification:** tests must evaluate every formulation-required Gauss and tying location, not only the element center. Director smoothing and Jacobian quality are separate checks; a smooth director cannot rescue a self-intersecting or inverted mapping.
 - **Drilling verification:** verify the fixed formula, dimensional family, symmetry, positivity, four-mode regularization, and absence from physical `E/N/M/Q/stress` recovery. Sensitivity and artificial-energy evidence are excluded.
-- **Reference comparison:** the declared S4 and S4R cases test source mapping and global displacement. They cannot prove formulation identity. Missing, extra, duplicate, nonfinite, or source-node/component-mismatched required rows fail before P1's mixed displacement tolerance is evaluated.
+- **Reference comparison:** the declared S4 case tests full-integration source mapping and global displacement. It cannot prove formulation identity. Missing, extra, duplicate, nonfinite, or source-node/component-mismatched required rows fail before P1's mixed displacement tolerance is evaluated. S4R source support is established separately by FESA mapping/kernel/metadata tests.
 - **Physics sanity:** force and global moment balance, symmetry, displacement direction, reaction sign, positive physical energy, free residual, and consistency of recovered resultants remain mandatory even when all reference displacement rows pass.
 - **Validation boundary:** the identified sources provide analytical, benchmark, and source-solver verification. No experimental dataset was established for the approved homogeneous linear-static feature; physical validation remains N/A unless separately added.
 
@@ -180,8 +180,8 @@ problems must not be mixed.
 
 1. **Drilling:** P1 fixes the positive physical-rotational-diagonal scale and `10^-3` factor. Alternative-family comparison, coefficient sweep, conditioning plateau and artificial-energy threshold are not required.
 2. **Director/geometry:** `NR-O03` and `NR-O04` are removed. The absence of a calibrated smooth angle or distortion/warp cutoff is not missing evidence.
-3. **Reference tolerance:** P1 reuses the B33 formula `1e-9 + 1e-6*reference_scale_c`; U exceedance fails and UR exceedance only warns.
-4. **Reference cases:** the existing `reference/shell/` S4 and `reference/shellR/` S4R input/displacement pairs are the complete required inventory. Administrative bundle metadata and an expanded portfolio are not gates.
+3. **Reference tolerance:** The approved downstream contract uses fixed absolute `1.0e-5` for every U/UR row without a reference-scale decision term; U exceedance fails and UR exceedance only warns. The B33 mixed tolerance remains unchanged and separate.
+4. **Reference case:** the existing `reference/shell/` S4 input/displacement pair is the complete required acceptance inventory because FESA-MITC4 uses full integration. `reference/shellR/` is not consumed by acceptance comparison. Administrative bundle metadata and an expanded portfolio are not gates.
 
 No research-owned numerical decision remains blocking for Implementation Planning.
 
@@ -195,8 +195,8 @@ No research-owned numerical decision remains blocking for Implementation Plannin
 | `037` S4/S4R common path | F-25–F-27, `MITC4-S4-S4R-SAME-PATH` | source mapping supported only as a FESA product decision, never an Abaqus formulation claim |
 | `039-048` shell outputs | F-09–F-11/F-18 | physical shell output and bottom/mid/top stress locations; no drilling-specific dataset |
 | `049-057` element verification | F-28/F-29 and Candidate Benchmarks | required invariant/patch/fixed-drill checks; broader portfolio remains optional |
-| `058-064` U/UR tolerance | F-25–F-27 plus P1 | exact B33 mixed tolerance; U blocking and UR warning-only |
-| `065-072` reference artifacts | S11/S12 and current inventory state | exact existing S4/S4R paths, source-row/component mapping, and immutability |
+| `058-064` U/UR tolerance | F-25–F-27 plus approved user decision | fixed absolute `1.0e-5`; U blocking and UR warning-only |
+| `065-072` reference artifacts | S11/S12 and current inventory state | exact existing S4 paths, source-row/component mapping, S4R non-consumption, and immutability |
 
 ## Downstream Handoff
 
@@ -223,7 +223,7 @@ No research-owned numerical decision remains blocking for Implementation Plannin
 ### Reference Model Agent
 
 - Record the exact existing S4 and S4R input/displacement paths without creating, renaming, repairing, or normalizing artifacts.
-- Define deterministic HDF5-to-CSV source-node/component mapping and the fixed B33 tolerance; do not require provenance, naming policy, README/metadata, duplicated model descriptions, or an expanded portfolio.
+- Define deterministic HDF5-to-CSV source-node/component mapping and the fixed absolute MITC4 tolerance `1.0e-5`; do not require provenance, naming policy, README/metadata, duplicated model descriptions, or an expanded portfolio.
 
 ### Implementation Planning Agent
 

docs/numerical-reviews/README.md

@@ -1,104 +0,0 @@
-# 수치 검토 리포트 작성 가이드
-
-이 디렉터리는 Numerical Review Agent가 작성하거나 제안한 기능별 수치 검토 리포트를 보관하는 위치다.
-
-기본 파일명은 `docs/numerical-reviews/<feature-id>-review.md` 형식을 사용한다. 각 리포트는 Formulation Agent의 정식화 문서를 독립 검토해, 구현 계획 단계로 넘겨도 되는지 판단한다.
-
-## Numerical Review Agent 역할
-
-Numerical Review Agent는 정식화의 수학적 일관성, 수치 안정성 위험, 검증 준비 상태를 검토한다.
-
-수행한다:
-- 수식의 차원, 부호, 좌표 변환, 적분 규칙을 검토한다.
-- `B` matrix 또는 kinematic operator, constitutive contract, element equation을 검토한다.
-- rigid body modes, patch test, symmetry, positive definiteness를 확인한다.
-- hourglass, shear locking, volumetric locking, distortion, singular Jacobian, conditioning 위험을 식별한다.
-- 구현 계획 전에 필요한 정식화 수정, 연구 보강, reference model 요구사항을 작성한다.
-
-수행하지 않는다:
-- C++ 코드를 구현하지 않는다.
-- 정식화 문서를 직접 수정하지 않는다.
-- C++ API나 파일 구조를 설계하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- release readiness를 승인하지 않는다.
-- 레퍼런스 결과와 구현 솔버 결과의 일치 여부를 판정하지 않는다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Numerical Review
-
-## Metadata
-- feature_id: <feature-id>
-- source_formulation: docs/formulations/<feature-id>-formulation.md
-- status: pass-for-implementation-planning | needs-formulation-revision | needs-research | blocked
-- owner_agent: numerical-review-agent
-- date: <YYYY-MM-DD>
-
-## Review Verdict
-- verdict: pass-for-implementation-planning | needs-formulation-revision | needs-research | blocked
-- reason: <판정 이유>
-
-## Critical Findings
-- <구현 전 반드시 수정할 수식 또는 수치 결함>
-
-## Numerical Risk Assessment
-- rigid_body_modes: <check/risk>
-- patch_test: <check/risk>
-- symmetry: <check/risk>
-- positive_definiteness: <check/risk>
-- hourglass: <check/risk or N/A>
-- shear_locking: <check/risk or N/A>
-- volumetric_locking: <check/risk or N/A>
-- distortion: <check/risk>
-- singular_jacobian: <check/risk>
-- conditioning: <check/risk>
-- convergence: <check/risk or N/A>
-
-## Consistency Checks
-- units: <pass/fail/TBD>
-- dimensions: <pass/fail/TBD>
-- signs: <pass/fail/TBD>
-- dof_ordering: <pass/fail/TBD>
-- coordinate_transforms: <pass/fail/TBD>
-- matrix_vector_dimensions: <pass/fail/TBD>
-- integration_weights: <pass/fail/TBD>
-- output_locations: <pass/fail/TBD>
-
-## Verification Readiness
-- unit_tests: <필요한 단위 테스트>
-- patch_tests: <필요한 patch test>
-- mms_or_mes: <MMS/MES 후보 또는 N/A>
-- benchmark_reference_comparison: <필요한 benchmark/reference 비교>
-- missing_evidence: <누락된 검증 근거>
-
-## Required Revisions
-
-### Formulation Agent
-- <정식화 수정 지시>
-
-### Research Agent
-- <연구 보강 지시>
-
-### Reference Model Agent
-- <optional downstream test note; not a formulation blocker>
-
-## Downstream Handoff
-
-### Implementation Planning Agent
-- <테스트와 acceptance criteria로 변환할 항목>
-
-### Reference Model Agent
-- <테스트 모델로 변환할 항목>
-```
-
-## 품질 기준
-
-- 확인된 결함, 위험, open question을 구분해야 한다.
-- `pass-for-implementation-planning`은 구현 계획으로 넘겨도 된다는 뜻이며 기능 완료나 release 승인이 아니다.
-- 정식화 문서를 직접 수정하지 않고 필요한 수정을 명확히 지시해야 한다.
-- 모든 검토는 dimension, sign, DOF ordering, coordinate transform, Jacobian, integration weight, element equation, output recovery를 포함해야 한다.
-- numerical risk는 approved feature scope가 요구하는 항목만 포함한다. 이후 Reference Model
-  문서, canonical naming, README, metadata, provenance, expanded portfolio 또는 아직 없는
-  comparison result는 formulation verdict의 blocker가 아니다.

docs/numerical-reviews/linear-static-mitc4-shell-review.md

@@ -1,346 +0,0 @@
-# Linear Static MITC4 Shell Numerical Review
-
-## Metadata
-
-- feature_id: `linear-static-mitc4-shell`
-- source_formulation: `docs/formulations/mitc4-shell-formulation.md`
-- source_requirements: `docs/requirements/linear-static-mitc4-shell.md`
-- source_research: `docs/research/linear-static-mitc4-shell-research.md`
-- source_io_definition: `docs/io-definitions/linear-static-mitc4-shell-io.md`
-- source_reference_inventory: `docs/reference-models/linear-static-mitc4-shell-reference-models.md`
-- repository_policy: `AGENTS.md`, `docs/ADR.md`, `docs/ARCHITECTURE.md`,
-  `docs/SOLVER_AGENT_DESIGN.md`
-- reviewed_revisions: `73df844`, `22a3238`
-- status: `pass-for-implementation-planning`
-- owner_agent: `numerical-review-agent`
-- date: `2026-08-12`
-- implementation_planning_authorized: `true`
-- implementation_complete: `false`
-- reference_comparison_complete: `false`
-
-## Review Verdict
-
-- verdict: `pass-for-implementation-planning`
-- reason: The current linear-static formulation closes the physical 20-DOF MITC4
-  kernel, its global 24-DOF embedding, fixed drilling regularization, Jacobian and
-  quadrature rules, residual/stiffness equations, recovery signs, and verification
-  invariants without a mathematical inconsistency in the approved feature scope.
-- critical_blockers: `none`
-- remaining_formulation_revisions: `none for the current linear-static scope`
-- downstream_boundary: Implementation Planning may begin. This verdict does not
-  claim implementation, build/test, reference-comparison, physics-sanity, or release
-  completion.
-
-The future geometrically nonlinear material in Formulation Section 15 remains
-explicitly non-executable. Its unresolved global finite-rotation map and objective
-drilling potential do not block the current linear-static implementation plan.
-
-## Critical Findings
-
-No confirmed mathematical defect remains in the approved linear-static formulation.
-The previous review's `needs-reference-model` verdict is not a valid current
-formulation verdict: the current numerical-review gate is based on numerical and
-formulation consistency, while downstream artifact administration and comparison
-execution are separate gates.
-
-### 1. Previous finding disposition
-
-| previous item | current disposition | evidence and strict consequence |
-| --- | --- | --- |
-| `NR-C01` Jacobian/geometry inventory | resolved | Formulation Sections 9.2-9.3 enumerate center, stiffness, tying, and recovery locations and require finite bases, nonzero area, and `J>0`; Requirements 014/016 intentionally define no calibrated smooth-angle, distortion, or warp threshold. |
-| `NR-C02` drilling normalization | resolved | Formulation Section 12.2 and Requirements 033-036 now define one exact dimensional rule using only positive physical tangent-rotation diagonals. No calibration decision remains. |
-| `NR-C03` mixed-DOF algebraic scaling | resolved | Formulation Sections 12.3 and 12.5 define physical length scaling separately from the physical/drilling stiffness split and provide normalized rank, symmetry, and rigid-action evidence. |
-| `NR-C04` 20/24-DOF weak-form mismatch | resolved | Formulation Sections 5.2-5.3 and 7.1-7.2 place physical, drilling, and external work in the common global 24-DOF test space using the required transpose maps. |
-| `NR-C05` nonlinear global tangent closure | resolved for current scope | Formulation Section 15 now labels the nonlinear equations non-executable and identifies the missing nonlinear `Phi` map, map Hessian, and objective drilling potential. Those items block only a future nonlinear feature. |
-| `NR-D01` drilling-direction nodal moment | retained/resolved | The exact-zero branch and `rho_M=|d dot M|/||M|| <= 1e-12` rule are consistent in Formulation Section 6.2 and I/O Section 4.3. |
-| `NR-D02` normalized algebraic checks | retained/resolved | Formulation Section 17.1 defines scale-aware symmetry, rigid-action, frame, transformation-energy, residual, and equilibrium checks without a denominator clamp. |
-| `NR-O01` drilling coefficient/plateau | resolved by approved fixed rule | `k_d=10^-3 min(R+)` replaces the former coefficient-family/sweep question. A sweep, plateau, response sensitivity, or condition-number calibration is not an acceptance gate. |
-| `NR-O02` drilling-energy warning | removed from approved scope | Drilling is an internal numerical potential only. No drilling-energy ratio, warning threshold, or drilling-specific output is required. |
-| `NR-O03` smooth-director angle | removed from approved scope | Requirements 014/054 and Formulation Sections 4.2 and 9.3 use exact orientation/finite/nonzero predicates and explicitly remove `NR-O03`. It is not an open numerical decision. |
-| `NR-O04` distortion/warp calibration | removed from approved scope | Requirements 016/054 and Formulation Sections 9.2-9.3 require exact finite/positive validity checks and explicitly remove `NR-O04`. It is not an open numerical decision. |
-| `NR-O05` U/UR tolerance | resolved | Requirements 058-062, Formulation Section 17.5, I/O Section 7.6, and Reference Case Section 5 all use the exact approved B33 component-scale formula. |
-
-### 2. Required policy classifications
-
-#### 2.1 Fixed drilling rule — resolved and implementation-ready
-
-Let `R+` contain only the finite, strictly positive diagonal entries of the physical
-local stiffness `K20` associated with the eight director-tangent rotational DOFs.
-The formulation fixes
-
-```text
-k_ref = min(R+)
-k_d = 1e-3 * k_ref
-K_drill_local = k_d * I4
-K_drill_24 = T_d^T * K_drill_local * T_d
-```
-
-All entries in `R+` have rotational-stiffness dimension `force*length`; translations,
-off-diagonals, nonpositive values, and nonfinite values are excluded. Therefore
-`k_d` has the correct dimension, `K_drill_24` is symmetric positive on the four pure
-drilling coordinates, and the physical and drilling channels are algebraically
-separate. An empty `R+` is a deterministic numerical-validation failure. This is a
-complete algorithm contract, not a calibration placeholder.
-
-The deterministic nodal frames fix the local-coordinate representation, and
-Formulation Section 5.3 supplies the virtual-work/energy congruence used by the
-coordinate-transformation check. No unselected drilling coefficient or family
-remains for Implementation Planning.
-
-#### 2.2 No drilling outputs — resolved and consistent
-
-Formulation Sections 7.2, 12.3, 12.5, 14, and 17.4 keep drilling out of physical
-strain, resultant, stress, and reported physical shell energy. The internal identity
-`E_drill = 0.5 gamma^T K_drill_local gamma` is permissible verification algebra; it
-does not create an external result quantity. Requirements 035/036/046 and I/O
-Sections 6.1/6.4/6.5 consistently require no drilling coefficient, stiffness, ratio,
-or energy dataset. There is no output-contract defect.
-
-#### 2.3 Exact B33 U/UR tolerance — resolved
-
-For each case and component `c`, using only finite Abaqus rows,
-
-```text
-reference_scale_c = max(abs(reference_value_i))
-tolerance_c = 1e-9 + 1e-6 * reference_scale_c
-```
-
-The `1e-9` floor is in the model's user-consistent length unit for `U1/U2/U3` and is
-dimensionless for `UR1/UR2/UR3`. No row is zero-clamped and no row-specific relative
-denominator replaces the component scale. `U1/U2/U3` exceedance is blocking at the
-later Reference Verification gate; `UR1/UR2/UR3` uses the same formula but is
-warning-only. The formula is exact and needs no MITC4-specific calibration.
-
-#### 2.4 Removed and administrative items — nonblocking
-
-`NR-O03`, `NR-O04`, drilling sweeps, drilling-energy criteria, and expanded flat,
-thin/thick, distorted, warped, curved-shell, or mesh-convergence portfolios are not
-part of the approved implementation-planning gate. They may remain future research
-or release evidence but shall not be reintroduced as missing numerical evidence.
-
-Canonical reference naming, a bundle README, `metadata.json`, provenance, an Abaqus
-version, duplicated units/coordinates/model/step/frame descriptions, or a schema
-version are administrative information and cannot block this formulation verdict.
-The absence of optional `metadata.json` is therefore not a defect. If such a file is
-added later, it remains read-only context and cannot override the approved exact
-paths, row mapping, or tolerance.
-
-## Numerical Risk Assessment
-
-| risk | current assessment | required in-scope control |
-| --- | --- | --- |
-| Transverse-shear locking | controlled for the original MITC4 scope, not claimed eliminated for every mesh | Use the exact edge-midpoint covariant shear projection of Sections 10.2-10.4 and pass transverse-shear patch/reference checks. |
-| Curved/distorted-mesh membrane locking | known limitation of the original MITC4 family | Preserve the documented limitation. An expanded convergence portfolio is nonblocking and does not authorize MITC4+. |
-| Volumetric locking | not applicable to the approved plane-stress shell contract | Do not reinterpret `C5` as a full 3D nearly incompressible material law. |
-| Hourglass modes | no reduced-integration/hourglass path is approved | Both source `S4` and `S4R` use the same full `2x2x2` FESA integration and MITC tying path. |
-| Degenerate, inverted, or self-intersecting geometry | fail-closed contract is present | Enforce distinct connectivity, non-self-intersection, finite nonzero surface measure, finite reciprocal bases, and `J>0` at every required location. |
-| Near-singular but still positive geometry | conditioning may degrade because no calibrated quality threshold is in scope | Preserve finite checks and deterministic solver failure diagnostics; do not invent `NR-O04` thresholds. |
-| Opposed or invalid nodal directors | would corrupt frames, signs, and tying | Reject nonfinite/zero candidates, nonpositive incident-normal agreement, and nonfinite/zero averages; use duplicate source nodes for folds. |
-| Drilling singularity | four nonphysical modes would remain without regularization | Apply the exact fixed `R+` rule and verify stabilized nullity six. |
-| Drilling contamination of physics | possible if drill enters the physical operator or recovery | Keep `T_p` and `T_d` separate and verify pure drill has zero physical strain/resultant/stress and no physical energy contribution. |
-| Rigid-mode test contamination | a full spatial rotation vector can contain director-parallel rotation | Construct physical rigid rotation with `u_I=omega x X_I`, `delta d_I=omega x d_I`, and `gamma_I=0` as specified in Section 8.3. |
-| Wrong shear component/factor | would cause patch failure or incorrect shear energy | Keep the `xi-zeta`/`eta-zeta` tying pairs and engineering factor `gamma_ij=2 epsilon_ij` explicit. |
-| Recovery sign/location drift | could hide a correct stiffness behind wrong outputs | Reuse stiffness frames, tying, material, and thickness quadrature; preserve four Gauss identities and bottom/middle/top positions without averaging. |
-| Future nonlinear misuse | current Section 15 does not define a complete global nonlinear element | Keep it non-executable until a separate approved nonlinear formulation closes `Phi`, map curvature, objective drill, load work, and state. |
-
-## Consistency Checks
-
-### 1. DOF order, director sign, and coordinate transforms — pass
-
-- Global order is exactly `[UX,UY,UZ,URX,URY,URZ]` per node.
-- `R_I=[a_I b_I d_I]` is right-handed and orthonormal, with
-  `[alpha,beta,gamma]^T=R_I^T theta_I^g`.
-- The director variation `delta d_I=beta_I a_I-alpha_I b_I` has the correct sign for
-  `theta_I x d_I`.
-- `T_p` is `20x24`; `T_d` is `4x24`. The physical operator receives only `q20`, and
-  the drilling potential receives only `gamma`.
-- The transpose maps in Formulation Sections 5.3 and 7.2 preserve virtual work and
-  energy in the common 24-DOF space.
-
-### 2. Shape functions, geometry, and B operator — pass
-
-- The bilinear shape functions satisfy partition of unity, nodal interpolation, and
-  derivative-sum identities.
-- The degenerated geometry uses a dimensionless unit director and separate thickness
-  factor `t*zeta/2`, avoiding thickness double-counting.
-- The direct covariant strain column is the symmetric gradient written in covariant
-  bases. The two transverse covariant shear components alone are replaced by the
-  canonical MITC4 edge-midpoint interpolation.
-- Reconstructing with reciprocal bases before local projection preserves the tensor
-  meaning. Engineering shear factors are applied once in the local five-component
-  vector.
-- The same projected `B_bar` is used in strain, residual, stiffness, and recovery;
-  no direct/tied shear mismatch remains.
-
-### 3. Constitutive matrix and dimensional consistency — pass
-
-- For finite `E>0` and `-1<nu<0.5`, the plane-stress block and
-  `kappa_s G I2`, with `kappa_s=5/6`, are symmetric positive definite.
-- `B_bar^T C5 B_bar J dxi deta dzeta` has stiffness-consistent dimensions because
-  the geometry Jacobian contains the through-thickness scale.
-- Membrane/shear strain is dimensionless, curvature is `1/length`, `N/Q` is
-  `force/length`, `M` is `force`, stress is `force/length^2`, and physical energy is
-  `force*length`.
-- The drilling reference uses rotational stiffness only, so no translation/rotation
-  unit mixing occurs.
-
-### 4. Integration, residual, stiffness, and modes — pass
-
-- Stiffness uses deterministic two-point Gauss quadrature in each of `xi`, `eta`, and
-  `zeta`, with points `+-1/sqrt(3)` and unit weights.
-- Both source types select this one rule; FESA does not emulate Abaqus S4/S4R internal
-  integration or hourglass behavior.
-- `K20 = integral(B_bar^T C5 B_bar dV)` is symmetric positive semidefinite. The
-  expected physical rank is 14: 20 physical coordinates minus six rigid modes.
-- `Kphys24=T_p^T K20 T_p` adds four drilling null coordinates. The fixed positive
-  drilling block removes those four, leaving exactly six physical rigid modes.
-- `f_int=K_e q_g` and `r=K d-F` use a consistent linear sign. Partitioning uses
-  `Kff df=Ff-Kfc dc`, including the valid `0x0 Kff` all-constrained case.
-
-### 5. Recovery and external result meaning — pass
-
-- Nodal reactions are constrained entries of the assembled full residual; free
-  entries remain equilibrium evidence.
-- Generalized strains are thickness moments with order
-  `[E11,E22,G12,K11,K22,K12,G13,G23]`.
-- Resultants use `[N11,N22,N12,M11,M22,M12,Q13,Q23]` and the stated centered-layer
-  `A/D/A_s` cross-check.
-- Bottom/middle/top `[S11,S22,S12]` are direct section-position evaluations. `S33=0`
-  is documented but not emitted, and `S13/S23` point stress is outside the output
-  contract.
-- Physical shell energy excludes numerical drilling stabilization, matching the I/O
-  schema.
-
-### 6. Architecture and deterministic lifecycle — pass for planning
-
-The formulation and I/O handoff match ADR-007/008/009/016/017 and the architecture:
-stable element-local computation, deterministic COO/reduction, stiffness assembly
-and partition before load assembly, factorization before substitution, full residual
-recovery, stable row identity, and failure-atomic HDF5 commit. These are planning and
-later implementation-test obligations, not unresolved equations.
-
-## Verification Readiness
-
-### Required element and algebraic tests
-
-Implementation Planning shall trace RED/GREEN/VERIFY tests for:
-
-1. shape-function identities and deterministic right-handed nodal/integration frames;
-2. valid and invalid geometry at every center, Gauss, tying, and recovery location;
-3. `T_p`/`T_d` dimensions, virtual-work equality, and transformation-energy equality;
-4. direct versus tied shear component construction and engineering-shear factors;
-5. constitutive symmetry/positive definiteness and exact `2x2x2` quadrature;
-6. normalized symmetry at `1e-12`, rigid action at `1e-10`, physical rank 14, and
-   stabilized rank 18/nullity six;
-7. exact `R+`, `k_ref`, `k_d`, and `K_drill_local` construction, including empty-`R+`
-   failure and exclusion of translational diagonals;
-8. pure drilling: positive drill action, zero physical strain/resultant/stress, and
-   no drilling-specific output;
-9. deterministic assembly/recovery ordering and thread-count repeatability;
-10. partition/effective-RHS/full-residual reaction behavior, including all-constrained
-    `0x0 Kff` handling.
-
-For a nonzero scaled stiffness, use the formulation's normalized metrics without a
-fallback denominator. Exactly zero constructed energy cases are classified by their
-separate rigid/null action tests rather than clamped to pass.
-
-### Required patch and sign tests
-
-- constant membrane strain/stress and `N` sign;
-- pure bending about both local axes, curvature/moment order, and bottom/top stress sign;
-- constant transverse shear and `Q13/Q23` order;
-- pure twist and `K12/M12` convention;
-- zero physical recovery from a pure drilling vector.
-
-### Declared reference readiness
-
-The lightweight inventory identifies these read-only required pairs:
-
-- `reference/shell/shell.inp` and
-  `reference/shell/shell displacements.csv` for source `S4`;
-- `reference/shellR/shellR.inp` and
-  `reference/shellR/shellR displacements.csv` for source `S4R`.
-
-Read-only inspection confirmed that all four declared files exist, their SHA-256
-values match the Reference Case inventory, and each required displacement CSV has
-49 data rows. This is inventory evidence only. No FESA output exists yet in this
-review, and no reference-comparison pass/fail decision was made.
-
-The later comparator must require exact normalized source-row/component sets,
-finite/unique values, U blocking, UR warning-only, and the approved mixed tolerance.
-The two Abaqus cases are not expected to equal one another, while identical supported
-FESA models labeled S4 or S4R must take the same internal numerical path.
-
-### Nonblocking evidence
-
-The following cannot change this formulation verdict:
-
-- absent README, `metadata.json`, provenance, canonical name, schema-version record,
-  or duplicated bundle descriptions;
-- no coefficient sweep, drilling-energy ratio, smooth-angle calibration, or
-  distortion/warp threshold sweep;
-- no expanded flat/thin/thick/distorted/warped/curved/convergence benchmark portfolio;
-- no implementation result, build/test result, Abaqus run, or completed comparison
-  at this pre-implementation gate.
-
-## Required Revisions
-
-### Formulation Agent
-
-- None for the approved current linear-static implementation scope.
-- Keep Formulation Section 15 non-executable until a separately approved nonlinear
-  feature closes its global coordinate map, consistent tangent, objective drilling,
-  and load-work decisions.
-
-### Research Agent
-
-- None before current Implementation Planning.
-- Broader original-MITC4 locking and convergence studies remain optional future
-  characterization and must not silently widen the implementation gate.
-
-### Reference Model Agent
-
-- None for this formulation verdict. Preserve the four declared artifacts read-only.
-- Optional administrative metadata, if later added by an authorized phase, does not
-  replace the approved exact paths, matching, and tolerance contract.
-
-## Downstream Handoff
-
-### Implementation Planning Agent
-
-Implementation Planning is authorized. The plan shall:
-
-- cover the deterministic director/frame preprocessing, `24 -> 20 + 4` transforms,
-  covariant MITC tying, full `2x2x2` integration, fixed drilling split, and recovery;
-- trace every approved must-requirement to TDD tests, including the invariant, patch,
-  fixed-drilling, failure, schema, row-matching, and U/UR decision behaviors above;
-- preserve one internal `FESA-MITC4` numerical path for source S4 and S4R while keeping
-  source metadata distinct;
-- keep drilling out of physical recovery and HDF5 results;
-- exclude future nonlinear execution, calibration sweeps, removed `NR-O03/NR-O04`,
-  expanded portfolios, and administrative reference requirements.
-
-This handoff authorizes planning only. It does not authorize Harness execution,
-production-code changes, reference-artifact mutation, or a claim of implementation
-completion.
-
-### I/O Definition Agent
-
-The current I/O contract is numerically consistent with the formulation. Planning
-shall preserve its exact source identity, load projection, output units/locations,
-physical-energy meaning, reference row mapping, and U/UR decision rule.
-
-### Reference Verification and Physics Evaluation Agents
-
-These remain downstream of implementation and build/test. Reference Verification
-will decide U/UR comparison outcomes; Physics Evaluation will independently assess
-equilibrium, signs, symmetry, and physical plausibility. Neither result is asserted
-by this review.
-
-## Review Evidence
-
-This review used the repository policy/design files, the approved requirements,
-research, formulation, I/O definition and reference-case inventory listed in
-Metadata, plus read-only inspection of the four declared artifacts. Local FEM wiki
-material cross-checked MITC4 kinematics, edge-midpoint assumed shear and known locking
-risks; the approved repository documents remain the feature source of truth.

docs/physics-evaluations/README.md

@@ -1,170 +0,0 @@
-# Physics Evaluation Report 문서 작성 가이드
-
-이 디렉터리는 Physics Evaluation Agent가 작성하거나 제안하는 기능별 physics evaluation report를 보관하는 위치다.
-
-Physics Evaluation Agent는 Reference Verification Agent가 `pass-for-physics-evaluation`로 넘긴 결과에 대해 물리적 타당성을 검토한다. 이 agent는 reference tolerance를 다시 판정하지 않고, FESA `results.h5`, Abaqus reference CSV files, optional FESA debug CSV view를 읽기 전용 evidence로 사용해 평형, 반력 부호, 변위 방향, 대칭성, 요소내력/응력 위치, rigid body mode 징후, energy/residual sanity, 테스트 모델 coverage를 검토한다.
-
-기본 문서명은 `docs/physics-evaluations/<feature-id>-physics-evaluation.md` 형식을 사용한다.
-
-## Physics Evaluation Agent 역할
-
-수행한다:
-- Reference Verification report가 `pass-for-physics-evaluation`인지 확인한다.
-- documented physical expectation이 있는 항목만 pass/fail로 판정한다.
-- global equilibrium, reaction consistency, displacement direction, symmetry, element force balance, stress/strain sanity, rigid body mode, energy/residual, model coverage를 검토한다.
-- 물리 검토 실패를 equilibrium-failure, reaction-inconsistency, displacement-direction-failure, symmetry-failure, stress-location-failure, element-force-inconsistency, rigid-body-mode-suspected, nonfinite-result, model-coverage-gap, upstream-contract, environment로 분류한다.
-- 실패 원인에 따라 Correction Agent, Reference Model Agent, Formulation Agent, I/O Definition Agent, Coordinator Agent로 handoff한다.
-- 모든 물리 검토가 문서화된 기대값을 만족하면 Release Agent로 handoff한다.
-
-수행하지 않는다:
-- source code를 수정하지 않는다.
-- tests를 수정하지 않는다.
-- CMake files를 수정하지 않는다.
-- requirements, formulations, I/O contracts, reference model contracts를 수정하지 않는다.
-- reference artifacts 또는 tolerance policies를 수정하지 않는다.
-- Abaqus, Nastran 또는 reference solver를 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- reference tolerance를 다시 판정하지 않는다.
-- release readiness를 승인하지 않는다.
-- release notes 또는 final release checklist를 작성하지 않는다.
-
-## 실행 순서
-
-Physics Evaluation Agent는 다음 순서를 따른다.
-
-```text
-EVIDENCE CHECK -> PHYSICS CHECKS -> CLASSIFY -> REPORT
-```
-
-`EVIDENCE CHECK`에서 다음 항목을 확인한다.
-
-- Reference Verification report status가 `pass-for-physics-evaluation`인지 여부
-- checked solver HDF5 file: `<solver output directory>/results.h5`
-- checked feature-declared Abaqus reference CSV files
-- optional FESA deterministic CSV view derived from `results.h5` for review only
-- compared quantities
-- model purpose
-- documented physical expectations
-- sign convention
-- units and coordinate system
-- output location and component naming
-
-## Physics Checks
-
-- `global equilibrium`: 적용 하중, 반력, sign convention이 문서화된 경우 전체 힘/모멘트 평형을 검토한다.
-- `reaction consistency`: constrained DOF와 reaction component가 경계조건과 일관적인지 검토한다.
-- `displacement direction`: 하중 방향, 구속조건, 예상 변형 모드와 변위 부호/방향이 맞는지 검토한다.
-- `symmetry`: symmetry, antisymmetry, expected zero component가 모델 목적과 일치하는지 검토한다.
-- `element force balance`: element internal force가 외력/반력 균형 또는 부호 일관성을 만족하는지 검토한다.
-- `stress/strain`: stress/strain 부호, component, coordinate system, output location이 모델 expectation과 일치하는지 검토한다.
-- `rigid body mode`: 불완전 구속, 비정상적으로 큰 변위, near-zero stiffness 징후를 검토한다.
-- `energy/residual`: `csv/energy_or_residual.csv` 또는 residual output이 있을 때 energy balance, residual trend, convergence sanity를 검토한다.
-- `model coverage`: reference model이 claimed feature를 충분히 검증하는지 검토한다.
-
-## Failure Classification
-
-- `equilibrium-failure`: 전체 하중-반력 또는 모멘트 평형이 기대와 다르다.
-- `reaction-inconsistency`: constrained DOF reaction이 boundary condition 또는 load path와 일관되지 않는다.
-- `displacement-direction-failure`: 변위 방향 또는 부호가 하중/구속조건과 물리적으로 맞지 않는다.
-- `symmetry-failure`: expected symmetry, antisymmetry, zero component가 깨졌다.
-- `stress-location-failure`: stress/strain output location 또는 component 해석이 물리 expectation과 맞지 않는다.
-- `element-force-inconsistency`: element internal force balance 또는 sign이 일관되지 않는다.
-- `rigid-body-mode-suspected`: rigid body mode 또는 under-constrained model 징후가 있다.
-- `nonfinite-result`: NaN 또는 infinite value가 있다.
-- `model-coverage-gap`: reference model이 기능을 충분히 검증하지 못한다.
-- `upstream-contract`: physical expectation, sign convention, unit, coordinate, output location 계약이 누락 또는 충돌한다.
-- `environment`: 로컬 실행/읽기 환경 문제로 평가가 불가능하다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Physics Evaluation Report
-
-## Metadata
-- feature_id: <feature-id>
-- source_reference_verification_report: docs/reference-verifications/<feature-id>-reference-verification.md
-- source_reference_model: docs/reference-models/<feature-id>-reference-models.md
-- source_requirement: docs/requirements/<feature-id>.md
-- source_formulation: docs/formulations/<feature-id>-formulation.md
-- status: pass-for-release-agent | needs-correction | needs-reference-model | needs-formulation-review | needs-io-decision | needs-upstream-decision | blocked
-- owner_agent: physics-evaluation-agent
-- date: <YYYY-MM-DD>
-
-## Input Evidence
-
-| evidence | path_or_source | status | notes |
-| --- | --- | --- | --- |
-| reference_verification | docs/reference-verifications/<feature-id>-reference-verification.md | pass-for-physics-evaluation | <summary> |
-| solver_hdf5 | <solver output directory>/results.h5 | present | missing | <summary> |
-| solver_csv_views | <solver output directory>/csv/ | present | missing | <summary or N/A> |
-| declared_reference_csv | <exact feature-declared path> | present | missing | <blocking/warning quantity summary> |
-| model_purpose | docs/reference-models/<feature-id>-reference-models.md | documented | missing | <summary> |
-| physical_expectations | <source docs> | documented | missing | <summary> |
-
-## Physics Checks
-
-| check | documented_expectation | observed_evidence | verdict | classification |
-| --- | --- | --- | --- | --- |
-| global equilibrium | <expected force/moment balance> | <observed> | pass | fail | skipped | equilibrium-failure |
-| reaction consistency | <expected reaction behavior> | <observed> | pass | fail | skipped | reaction-inconsistency |
-| displacement direction | <expected direction/sign> | <observed> | pass | fail | skipped | displacement-direction-failure |
-| symmetry | <expected symmetry/zero components> | <observed> | pass | fail | skipped | symmetry-failure |
-| element force balance | <expected element force relation> | <observed> | pass | fail | skipped | element-force-inconsistency |
-| stress/strain sanity | <expected sign/location/component> | <observed> | pass | fail | skipped | stress-location-failure |
-| rigid body mode | <expected constrained behavior> | <observed> | pass | fail | skipped | rigid-body-mode-suspected |
-| energy/residual | <expected energy/residual sanity> | <observed> | pass | fail | skipped | upstream-contract |
-| model coverage | <claimed feature coverage> | <observed> | pass | fail | skipped | model-coverage-gap |
-
-## Failure Classification
-- classification: equilibrium-failure | reaction-inconsistency | displacement-direction-failure | symmetry-failure | stress-location-failure | element-force-inconsistency | rigid-body-mode-suspected | nonfinite-result | model-coverage-gap | upstream-contract | environment | N/A
-- primary_failure: <short summary>
-- evidence: <short relevant evidence>
-
-## Evaluation Verdict
-- verdict: pass-for-release-agent | needs-correction | needs-reference-model | needs-formulation-review | needs-io-decision | needs-upstream-decision | blocked
-- reason: <short reason>
-
-## Handoff Recommendation
-
-| target_agent | reason | required_input |
-| --- | --- | --- |
-| Correction Agent | <implementation-owned physical failure> | <failed check and evidence> |
-| Reference Model Agent | <model coverage or missing physical expectation issue> | <coverage gap> |
-| Formulation Agent | <formulation or sign convention issue> | <failed physics check> |
-| I/O Definition Agent | <output location/component/unit/coordinate ambiguity> | <contract gap> |
-| Coordinator Agent | <blocked or repeated ambiguity> | <classification and open issue> |
-| Release Agent | <all documented physics checks passed> | <evaluation report> |
-
-## No-Change Assertion
-- source_files_modified: false
-- test_files_modified: false
-- cmake_files_modified: false
-- reference_artifacts_modified: false
-- tolerance_policies_modified: false
-- notes: <observed no-change evidence or exception>
-
-## Open Issues
-- <missing physical expectation, incomplete model coverage, contradictory sign convention, or unavailable energy/residual evidence>
-```
-
-## 상태 값
-
-- `pass-for-release-agent`: documented physics checks가 통과했고 Release Agent가 release readiness를 평가할 수 있다.
-- `needs-correction`: implementation-owned physical failure가 있어 Correction Agent가 필요하다.
-- `needs-reference-model`: feature-required case 또는 명시적으로 요구된 physical expectation이 누락됐다.
-- `needs-formulation-review`: formulation 또는 numerical review 재검토가 필요하다.
-- `needs-io-decision`: output location, component naming, sign convention, unit, coordinate mapping이 evaluation을 막는다.
-- `needs-upstream-decision`: physical expectation, sign convention, model purpose, acceptance criterion이 누락 또는 충돌한다.
-- `blocked`: 사용자 또는 Coordinator Agent 결정 없이는 안전하게 진행할 수 없다.
-
-## 품질 기준
-
-- Reference Verification report가 `pass-for-physics-evaluation`이 아니면 physics pass를 판정하지 않는다.
-- documented expectation이 없는 항목은 `skipped`로 두며 expanded portfolio나 calibration을
-  새 gate로 만들지 않는다. Feature-required expectation 자체가 누락된 경우에만
-  `needs-upstream-decision` 또는 `needs-reference-model`을 사용한다.
-- 평형 검토는 적용 하중, 반력, element/internal force sign convention이 문서화된 경우에만 수행한다.
-- stress/strain 검토는 output location, component naming, coordinate system, units가 정의된 경우에만 수행한다.
-- FESA `results.h5`, Abaqus reference CSV files, optional FESA debug CSV view는 읽기 전용 evidence로만 사용한다.
-- pass는 Release Agent로 넘길 수 있다는 뜻이며 release readiness 승인이 아니다.
-- reference artifacts와 tolerance policies는 수정하지 않는다.

docs/reference-models/README.md

@@ -1,91 +0,0 @@
-# Reference Case 문서 작성 가이드
-
-이 디렉터리는 Reference Model Agent가 작성하는 기능별 lightweight reference-case
-inventory를 보관한다. Abaqus는 외부 수치 reference이며 FESA의 formulation 또는 내부
-동작 계약이 아니다.
-
-기본 파일명은 `docs/reference-models/<feature-id>-reference-models.md`다. Agent는 Abaqus
-또는 다른 reference solver를 실행하지 않고 `reference/` 파일을 생성, 수정, rename,
-repair 또는 normalize하지 않는다.
-
-## 필수 내용
-
-기능이 실제로 비교하는 case마다 다음만 기록한다.
-
-- case id와 목적
-- existing `.inp` exact path
-- blocking 또는 warning-only quantity의 existing CSV exact path
-- FESA HDF5 dataset
-- source identity와 component mapping
-- missing/extra/duplicate/nonfinite row precheck
-- approved tolerance
-- artifact presence/readability status
-
-다음은 기본 readiness 조건이 아니다.
-
-- canonical directory/file naming 또는 legacy-alias 승인
-- bundle `README.md` 또는 `metadata.json`
-- Abaqus version/generation provenance
-- duplicated units, coordinates, model, step/frame, material, section, thickness, element type
-- reference CSV schema version
-- 비교하지 않는 quantity CSV
-- 요구조건이 요청하지 않은 benchmark portfolio
-
-단일 static step/final frame 기능은 input/CSV pair로 result state를 식별한다. Material,
-section, loads, constraints와 source element type은 `.inp`에서 읽는다.
-
-## 문서 템플릿
-
-```markdown
-# <Feature Title> Reference Cases
-
-## Metadata
-- feature_id: <feature-id>
-- status: draft | needs-user-decision | needs-reference-artifacts | ready-for-implementation-planning | blocked
-- owner_agent: reference-model-agent
-- date: <YYYY-MM-DD>
-
-## Reference Acceptance Scope
-- blocking_quantities: [<quantity/components>]
-- warning_only_quantities: [<quantity/components>]
-- excluded_quantities: [<quantity/reason>]
-
-## Reference Case Inventory
-
-| case_id | purpose | input | required_csv | quantity | behavior | status |
-| --- | --- | --- | --- | --- | --- | --- |
-| <case> | <purpose> | reference/<dir>/<file>.inp | reference/<dir>/<file>.csv | <components> | blocking | warning-only | present | missing |
-
-## HDF5-to-CSV Comparison Mapping
-
-| case_id | hdf5_dataset | source_id | csv_id_column | components | row_precheck |
-| --- | --- | --- | --- | --- | --- |
-| <case> | <path> | <node/element source id> | <column> | <components> | exact set; unique; finite |
-
-## Tolerance and Blocking/Warning Policy
-- reference_scale: <formula>
-- row_tolerance: <formula>
-- zero_policy: no clamp
-- blocking_behavior: <rule>
-- warning_behavior: <rule>
-
-## Readiness Checklist
-- declared input exists and is readable
-- every required CSV exists and is readable
-- source identity and required components are deterministic
-- row-set/nonfinite prechecks are defined
-- tolerance and blocking/warning behavior are approved
-
-## Open Issues and Downstream Handoff
-- <only missing required files, matching, tolerance, or feature-owned decisions>
-```
-
-## 상태 규칙
-
-- `ready-for-implementation-planning`: required input/CSV files, mapping, and tolerance are complete.
-- `needs-reference-artifacts`: a declared input or required comparison CSV is missing.
-- `needs-user-decision`: required quantity, mapping, or tolerance is undefined.
-- `blocked`: no safe progress is possible without an external decision or state change.
-
-Canonical naming, README, metadata, provenance, and unrequested portfolio coverage do not select
-any failure status.

docs/reference-verifications/README.md

@@ -1,107 +0,0 @@
-# Reference Verification Report 작성 가이드
-
-Reference Verification Agent는 Build/Test 통과 뒤 FESA `results.h5`와 기능이 선언한
-Abaqus CSV quantity를 비교한다. 비교 대상은 observable value이며 Abaqus 내부 formulation
-동등성이 아니다.
-
-기본 문서명은 `docs/reference-verifications/<feature-id>-reference-verification.md`다.
-
-## 실행 순서
-
-```text
-ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT
-```
-
-`ARTIFACT CHECK`는 다음만 요구한다.
-
-- declared `.inp`
-- generated FESA `results.h5`
-- every feature-required Abaqus CSV
-- HDF5 dataset projection
-- source-ID/component mapping
-- blocking/warning behavior
-- approved tolerance
-
-Canonical naming, README, metadata, Abaqus version/provenance, duplicated units/coordinates/
-step-frame fields와 reference CSV schema version은 기본 gate가 아니다.
-
-Missing, extra, duplicate 또는 nonfinite required row는 tolerance 전에 실패한다. Row order
-alone으로 대응하지 않는다. FESA/Reference 값을 zero-clamp하거나 tolerance에 맞게 보정하지
-않는다.
-
-## 보고 항목
-
-각 blocking 또는 warning-only quantity에 대해 다음을 보고한다.
-
-- exact input and CSV path
-- FESA HDF5 dataset
-- source ID/component mapping
-- compared, missing, extra, duplicate, nonfinite row counts
-- max absolute and relative error
-- RMS and norm error when required
-- worst source ID/component
-- pass, fail, or warning
-
-## Failure Classification
-
-- `missing-reference-artifact`: declared input or required CSV missing
-- `missing-solver-output`: `results.h5` or comparison command missing
-- `schema-mismatch`: required source ID/component columns unavailable
-- `id-mismatch`: required source row sets do not match
-- `tolerance-failure`: blocking value exceeds approved tolerance
-- `nonfinite-result`: required FESA or reference value is not finite
-- `upstream-contract`: quantity, HDF5 projection, mapping, or tolerance missing/contradictory
-- `environment`: local comparison cannot run
-
-## 문서 템플릿
-
-```markdown
-# <Feature Title> Reference Verification Report
-
-## Metadata
-- feature_id: <feature-id>
-- status: pass-for-physics-evaluation | needs-correction | needs-reference-artifacts | needs-solver-results | needs-upstream-decision | blocked
-- owner_agent: reference-verification-agent
-- date: <YYYY-MM-DD>
-
-## Artifact Inventory
-| item | path | status | notes |
-| --- | --- | --- | --- |
-| input | <exact path> | present | missing | <summary> |
-| required_csv | <exact path> | present | missing | <summary> |
-| solver_hdf5 | <path>/results.h5 | present | missing | <summary> |
-
-## Comparison Contract
-- hdf5_dataset: <path>
-- source_id_mapping: <rule>
-- components: <blocking/warning inventory>
-- row_precheck: exact set; unique; finite
-- tolerance_source: <path/requirement>
-- tolerance_policy: <formula>
-
-## Quantity Results
-| quantity | components | behavior | compared | missing | extra | max_abs | max_rel | rms | norm | worst_id/component | result |
-| --- | --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- | --- |
-| <quantity> | <components> | blocking | warning-only | <values> |
-
-## Failure Classification
-- classification: <value or N/A>
-- evidence: <summary>
-
-## Handoff Recommendation
-- <target and exact evidence>
-
-## No-Change Assertion
-- source_files_modified: false
-- test_files_modified: false
-- cmake_files_modified: false
-- reference_artifacts_modified: false
-- tolerance_policies_modified: false
-
-## Open Issues
-- <only actual comparison blockers>
-```
-
-`pass-for-physics-evaluation`은 required blocking comparison 통과만 의미한다. Warning-only
-quantity는 경고를 남기되 verdict를 바꾸지 않는다. Physics validation과 release readiness는
-각 후속 agent가 판정한다.

docs/releases/README.md

@@ -1,173 +0,0 @@
-# Release Report 문서 작성 가이드
-
-이 디렉터리는 Release Agent가 작성하거나 제안하는 기능별 release readiness report를 보관하는 위치다.
-
-Release Agent는 Physics Evaluation Agent가 `pass-for-release-agent`로 넘긴 기능에 대해 최종 gate evidence를 감사한다. 이 Agent는 source code, tests, CMake, upstream 계약, reference artifacts, tolerance policies를 수정하지 않는다. 또한 Abaqus/Nastran 실행, Abaqus reference CSV 생성 또는 수정, 외부 publish/deploy/package/tag/commit 작업을 수행하지 않는다.
-
-기본 문서명은 `docs/releases/<feature-id>-release.md` 형식을 사용한다.
-
-## Release Agent 역할
-
-수행한다:
-- upstream gate report가 같은 `feature_id`를 대상으로 하는지 확인한다.
-- Build/Test report의 `pass-for-reference-verification` 상태를 확인한다.
-- Reference Verification report의 `pass-for-physics-evaluation` 상태를 확인한다.
-- Physics Evaluation report의 `pass-for-release-agent` 상태를 확인한다.
-- 모든 `must` requirement가 acceptance criterion, test/reference evidence, release scope에 trace되는지 확인한다.
-- known limitations, deferred issues, unsupported Abaqus keyword, accepted risks를 release note에 기록한다.
-- release checklist와 Release Notes Draft를 작성한다.
-
-수행하지 않는다:
-- source code를 수정하지 않는다.
-- tests를 수정하지 않는다.
-- CMake files 또는 build configuration을 수정하지 않는다.
-- requirements, formulations, I/O contracts, numerical review reports, reference verification reports, physics evaluation reports를 수정하지 않는다.
-- reference artifacts 또는 tolerance policies를 수정하지 않는다.
-- Abaqus, Nastran 또는 reference solver를 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- 실패하거나 누락된 upstream gate를 우회하지 않는다.
-- 사용자 명시 요청 없이 publish, deploy, package, tag, commit, external release를 수행하지 않는다.
-
-## 실행 순서
-
-Release Agent는 다음 순서를 따른다.
-
-```text
-GATE AUDIT -> TRACEABILITY CHECK -> RELEASE DOCUMENTATION -> RELEASE VERDICT
-```
-
-`GATE AUDIT`에서는 다음 evidence를 확인한다.
-
-- Physics Evaluation report status: `pass-for-release-agent`
-- Reference Verification report status: `pass-for-physics-evaluation`
-- Build/Test report status: `pass-for-reference-verification`
-- Implementation report와 implementation plan의 feature scope 일치
-- requirements, formulations, numerical reviews, I/O definitions, reference models 문서의 feature scope 일치
-- Build/Test report의 `.harness/config.json` 또는 자동 감지 기반 MSVC build/test command evidence
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Release Report
-
-## Metadata
-- feature_id: <feature-id>
-- source_requirement: docs/requirements/<feature-id>.md
-- source_formulation: docs/formulations/<feature-id>-formulation.md
-- source_numerical_review: docs/numerical-reviews/<feature-id>-review.md
-- source_io_definition: docs/io-definitions/<feature-id>-io.md
-- source_reference_model: docs/reference-models/<feature-id>-reference-models.md
-- source_implementation_plan: docs/implementation-plans/<feature-id>-implementation-plan.md
-- source_build_test_report: docs/build-test-reports/<feature-id>-build-test.md
-- source_reference_verification_report: docs/reference-verifications/<feature-id>-reference-verification.md
-- source_physics_evaluation_report: docs/physics-evaluations/<feature-id>-physics-evaluation.md
-- status: ready-for-release | needs-correction | needs-reference-verification | needs-physics-evaluation | needs-documentation | needs-upstream-decision | blocked
-- owner_agent: release-agent
-- date: <YYYY-MM-DD>
-
-## Release Scope
-
-| item | included | excluded | notes |
-| --- | --- | --- | --- |
-| analysis_type | <supported analysis> | <unsupported analysis> | <notes> |
-| element_type | <supported elements> | <unsupported elements> | <notes> |
-| material_model | <supported materials> | <unsupported materials> | <notes> |
-| Abaqus input subset | <supported keywords> | <unsupported keywords> | <notes> |
-| output_quantities | <supported outputs> | <unsupported outputs> | <notes> |
-
-## Gate Evidence Inventory
-
-| gate | source | expected_status | observed_status | verdict |
-| --- | --- | --- | --- | --- |
-| requirements | docs/requirements/<feature-id>.md | approved or release-ready | <status> | pass | fail | missing |
-| formulation | docs/formulations/<feature-id>-formulation.md | reviewed | <status> | pass | fail | missing |
-| numerical_review | docs/numerical-reviews/<feature-id>-review.md | pass-for-implementation-planning | <status> | pass | fail | missing |
-| io_definition | docs/io-definitions/<feature-id>-io.md | ready | <status> | pass | fail | missing |
-| reference_model | docs/reference-models/<feature-id>-reference-models.md | ready or artifacts present | <status> | pass | fail | missing |
-| implementation | <implementation report> | implemented | <status> | pass | fail | missing |
-| build_test | docs/build-test-reports/<feature-id>-build-test.md | pass-for-reference-verification | <status> | pass | fail | missing |
-| reference_verification | docs/reference-verifications/<feature-id>-reference-verification.md | pass-for-physics-evaluation | <status> | pass | fail | missing |
-| physics_evaluation | docs/physics-evaluations/<feature-id>-physics-evaluation.md | pass-for-release-agent | <status> | pass | fail | missing |
-
-## Acceptance Traceability
-
-| requirement_id | acceptance_criterion | test_id | reference_model_id | verification_report | release_disposition |
-| --- | --- | --- | --- | --- | --- |
-| <req-id> | <criterion> | <test-id> | <model-id> | <report path> | released | deferred | blocked |
-
-## Validation Evidence
-
-| command_or_report | expected | observed | notes |
-| --- | --- | --- | --- |
-| Build/Test report의 config-resolved CMake/MSVC/CTest | pass | <result> | <summary> |
-| Harness Python pytest (when applicable) | pass or N/A | <result> | <summary> |
-| reference verification | pass-for-physics-evaluation | <status> | <summary> |
-| physics evaluation | pass-for-release-agent | <status> | <summary> |
-
-## Known Limitations
-
-| limitation | category | user_impact | disposition |
-| --- | --- | --- | --- |
-| <unsupported Abaqus keyword or solver scope limit> | input | physics | numerical | output | <impact> | documented | deferred | blocker |
-
-## Release Notes Draft
-
-### Feature Summary
-- <user-facing summary>
-
-### Verification Scope
-- <validated analysis, element, material, I/O, reference model scope>
-
-### Main Limitations
-- <known limitation>
-
-### Artifacts
-- <release report, reference bundle, verification report paths>
-
-## Release Verdict
-- verdict: ready-for-release | needs-correction | needs-reference-verification | needs-physics-evaluation | needs-documentation | needs-upstream-decision | blocked
-- reason: <short reason>
-
-## Handoff Recommendation
-
-| target_agent | reason | required_input |
-| --- | --- | --- |
-| Coordinator Agent | <release decision or blocked issue> | <summary> |
-| Correction Agent | <implementation-owned release blocker> | <failure evidence> |
-| Reference Verification Agent | <missing or failed reference comparison> | <artifact and report gap> |
-| Physics Evaluation Agent | <missing or failed physics evaluation> | <reference verification evidence> |
-| Requirement Agent | <requirement or acceptance gap> | <open decision> |
-| I/O Definition Agent | <I/O scope or Abaqus keyword limitation gap> | <contract gap> |
-| Reference Model Agent | <missing required input/CSV or comparison mapping> | <exact gap> |
-
-## No-Change Assertion
-- source_files_modified: false
-- test_files_modified: false
-- cmake_files_modified: false
-- reference_artifacts_modified: false
-- tolerance_policies_modified: false
-- notes: <observed no-change evidence or exception>
-
-## Open Issues
-- <missing evidence, contradictory upstream report, unresolved defect, missing required comparison file/mapping, or documentation gap>
-```
-
-## 상태 값
-
-- `ready-for-release`: required gate가 모두 통과했고, 모든 `must` requirement traceability와 known limitations 문서화가 완료되었다.
-- `needs-correction`: implementation-owned defect 또는 unresolved test/build/reference/physics issue가 있어 Correction Agent가 필요하다.
-- `needs-reference-verification`: reference verification report가 없거나 `pass-for-physics-evaluation`이 아니다.
-- `needs-physics-evaluation`: physics evaluation report가 없거나 `pass-for-release-agent`가 아니다.
-- `needs-documentation`: gate evidence는 통과했지만 release scope, known limitations, release notes, traceability 문서가 불완전하다.
-- `needs-upstream-decision`: requirement, tolerance, required comparison file/mapping, I/O, acceptance evidence가 누락되었거나 상충한다.
-- `blocked`: 사용자 또는 Coordinator Agent 결정 없이는 안전하게 진행할 수 없다.
-
-## 품질 기준
-
-- `ready-for-release`는 Build/Test, Reference Verification, Physics Evaluation gate evidence가 모두 present and passing일 때만 사용할 수 있다.
-- 모든 `must` requirement는 acceptance criterion, test/reference evidence, release scope에 trace되어야 한다.
-- known limitations와 deferred/open issue는 Release Notes Draft에 명확히 기록되어야 한다.
-- missing evidence, contradictory upstream reports, unresolved defects, missing declared comparison
-  files는 적절한 `needs-*` 상태로 분류한다. Canonical naming, README, metadata, provenance와
-  unrequested portfolio는 blocker가 아니다.
-- 이 문서는 FESA 내부 feature release readiness 판정을 위한 것이며, 외부 publish/deploy/package/tag/commit 자동화는 포함하지 않는다.

docs/requirements/README.md

@@ -1,123 +0,0 @@
-# 요구조건 문서 작성 가이드
-
-이 디렉터리는 Requirement Agent가 작성하거나 제안한 기능별 요구조건 문서를 보관하는 위치다.
-
-기본 파일명은 `docs/requirements/<feature-id>.md` 형식을 사용한다. 각 문서는 구현 전에 작성되며, Formulation Agent, I/O Definition Agent, Reference Model Agent, Implementation Planning Agent가 이어받을 수 있는 검증 가능한 baseline이어야 한다.
-
-## Requirement Agent 역할
-
-Requirement Agent는 솔버 기능 요청을 검증 가능한 요구조건으로 바꾼다.
-
-수행한다:
-- 기능 범위, 제외 범위, 입력, 출력, 제약조건을 정의한다.
-- 해석 타입, 대상 요소, 자유도, 재료 모델, 경계조건, 하중 조건을 명확히 한다.
-- 절점 변위, 반력, 요소 내력, 응력 등 검증 물리량을 정한다.
-- tolerance와 reference artifact 요구사항을 기록한다.
-- Requirement Verification Matrix를 작성한다.
-
-수행하지 않는다:
-- C++ 코드를 구현하지 않는다.
-- 유한요소 정식화를 확정하지 않는다.
-- C++ API나 파일 구조를 설계하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- 기능 완료 여부를 승인하지 않는다.
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Requirements
-
-## Metadata
-- feature_id: <feature-id>
-- title: <title>
-- status: draft | needs-user-decision | approved
-- owner_agent: requirement-agent
-- date: <YYYY-MM-DD>
-
-## Purpose
-<기능 목적과 사용자/개발자 관점의 기대 동작>
-
-## In Scope
-- <포함 범위>
-
-## Out Of Scope
-- <제외 범위>
-
-## Analysis Definition
-- analysis_type: linear static | nonlinear static | modal | other
-- elements: <대상 요소>
-- dofs: <자유도>
-- material_model: <재료 모델>
-- boundary_conditions: <경계조건>
-- loads: <하중 조건>
-- coordinate_system: <좌표계>
-- units: <단위계>
-
-## Input Requirements
-- <입력 요구조건>
-
-## Output Requirements
-- <출력 요구조건>
-
-## Verification Quantities
-- nodal_displacement: required | not-applicable
-- reaction: required | not-applicable
-- element_internal_force: required | not-applicable
-- stress: required | not-applicable
-- strain: required | not-applicable
-- energy_or_residual: required | not-applicable
-
-## Tolerance Policy
-- absolute_tolerance: <value or TBD>
-- relative_tolerance: <value or TBD>
-- norm_based_tolerance: <value or TBD>
-
-## Reference Case Requirements
-
-- input_path: <exact existing `.inp` path or N/A>
-- required_csv_paths: <only blocking/warning quantity CSV paths or N/A>
-- blocking_quantities: <components or N/A>
-- warning_only_quantities: <components or N/A>
-- hdf5_projection: <dataset or N/A>
-- source_id_component_matching: <rule or N/A>
-- row_precheck: missing/extra/duplicate/nonfinite fail before tolerance
-- tolerance: <approved formula or N/A>
-
-Canonical naming, README, metadata, provenance, duplicated units/coordinates/step-frame fields,
-CSV schema versions와 비교하지 않는 quantity CSV는 기본 요구조건이 아니다.
-
-## Requirement Verification Matrix
-
-| id | statement | category | rationale | source | priority | verification_method | acceptance_criteria | tolerance | downstream_agents | status |
-| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
-| FESA-REQ-<FEATURE>-001 | The FESA solver shall ... | functional | ... | user | must | reference-comparison | ... | ... | Reference Model Agent; Implementation Planning Agent | draft |
-
-## Open Questions
-- <미확정 값 또는 사용자 결정 필요 사항>
-
-## Downstream Handoff
-
-### Research Agent
-- <조사할 이론, 논문, benchmark, 표준>
-
-### Formulation Agent
-- <정식화 단계에 넘길 해석 타입, 요소, 재료, 자유도, 출력 물리량 계약>
-
-### I/O Definition Agent
-- <입력/출력 schema 요구조건>
-
-### Reference Model Agent
-- <exact input/required CSV paths, blocking/warning quantities, matching, tolerance>
-
-### Implementation Planning Agent
-- <먼저 작성할 테스트와 acceptance criteria>
-```
-
-## 품질 기준
-
-- 모든 `must` 요구조건은 검증 방법과 acceptance criteria를 가져야 한다.
-- 모든 수치 요구조건은 단위, 좌표계, tolerance 또는 `TBD with owner`를 가져야 한다.
-- reference 비교가 필요한 요구조건은 필요한 CSV artifact를 명시해야 한다.
-- "빠르게", "정확하게", "Abaqus처럼" 같은 문장은 검증 가능한 기준으로 바꾸거나 open question으로 남겨야 한다.
-- 구현 방법, 정식화 세부식, C++ API는 이 문서에서 확정하지 않는다.

docs/research/README.md

@@ -1,108 +0,0 @@
-# 연구 브리프 작성 가이드
-
-이 디렉터리는 Research Agent가 작성하거나 제안한 기능별 연구 브리프를 보관하는 위치다.
-
-기본 파일명은 `docs/research/<feature-id>-research.md` 형식을 사용한다. 각 브리프는 Requirement Agent의 요구조건을 입력으로 받아 Formulation Agent, Numerical Review Agent, Reference Model Agent, Implementation Planning Agent가 사용할 수 있는 근거 자료를 제공해야 한다.
-
-## Research Agent 역할
-
-Research Agent는 FEM 이론, benchmark, verification reference, solver manual, 논문 자료를 조사한다.
-
-수행한다:
-- 기능 요구조건과 관련된 이론 자료를 조사한다.
-- 요소별 benchmark, patch test, MMS, MES, convergence study 후보를 찾는다.
-- Abaqus/Nastran 결과와 비교 가능한 공개 benchmark 또는 문헌 해를 정리한다.
-- 자료의 신뢰도, 적용 범위, 한계, 상충 여부를 평가한다.
-- downstream agent가 사용할 수 있도록 출처와 근거를 추적 가능하게 남긴다.
-
-수행하지 않는다:
-- C++ 코드를 구현하지 않는다.
-- 유한요소 정식화를 확정하지 않는다.
-- C++ API나 파일 구조를 설계하지 않는다.
-- Abaqus, Nastran 또는 레퍼런스 솔버를 직접 실행하지 않는다.
-- Abaqus reference CSV 파일을 생성하거나 수정하지 않는다.
-- 기능 완료 여부를 승인하지 않는다.
-
-## Source Reliability Tier
-
-| tier | source type | examples | use |
-| --- | --- | --- | --- |
-| Tier 1 | 공식 표준, 공식 solver manual, 공식 benchmark, 공공기관 자료 | ASME V&V 10, Abaqus Verification Guide, Abaqus Benchmarks Guide, NAFEMS benchmarks, NASA FEMCI, official solver manuals | 우선 근거 |
-| Tier 2 | peer-reviewed paper, reproducible arXiv preprint, textbook | MMS/MES 논문, finite element textbook | 이론/검증 후보 |
-| Tier 3 | vendor example, university course note, technical blog | 공개 강의노트, 기술 블로그 | 보조 근거 |
-| Reject | forum answer, LLM summary, unsourced page, illegal PDF mirror, citation 없는 wiki성 문서 | forum, mirror PDF, 출처 없는 요약 | primary evidence로 사용 금지 |
-
-## 문서 템플릿
-
-```markdown
-# <feature title> Research Brief
-
-## Metadata
-- feature_id: <feature-id>
-- source_requirement: docs/requirements/<feature-id>.md
-- status: draft | needs-user-decision | ready-for-formulation
-- owner_agent: research-agent
-- date: <YYYY-MM-DD>
-
-## Research Questions
-- <Requirement Agent 또는 사용자가 넘긴 조사 질문>
-
-## Source Inventory
-
-| source_type | title | author_or_org | URL_or_DOI | access_date | reliability_tier | notes |
-| --- | --- | --- | --- | --- | --- | --- |
-| standard | ASME V&V 10 | ASME | <URL> | <YYYY-MM-DD> | Tier 1 | VVUQ framework |
-
-## Extracted Facts
-- <정식화에 필요한 사실, benchmark 조건, 검증 물리량, 재료/좌표/단위 가정>
-
-## Candidate Benchmarks
-
-| benchmark_id | source | benchmark_type | physics | target_quantities | artifact_needs | applicability |
-| --- | --- | --- | --- | --- | --- | --- |
-| <id> | <source title> | analytical \| NAFEMS \| Abaqus Verification/Benchmark \| NASA/FEMCI \| paper-derived | <physics> | <targets> | <needed artifacts> | <limits> |
-
-## Verification Relevance
-- code_verification: <관련성>
-- solution_verification: <관련성>
-- validation: <관련성>
-- reference_comparison: <관련성>
-
-## Applicability Limits
-- linear_or_nonlinear: <scope>
-- deformation: small | large | TBD
-- element_type: <scope>
-- material_model: <scope>
-- geometry: <scope>
-- boundary_conditions: <scope>
-- loads: <scope>
-- coordinate_system: <scope>
-- units: <scope>
-
-## Open Issues
-- <근거 부족, 상충 자료, 유료/비공개 자료, 사용자 결정 필요 사항>
-
-## Downstream Handoff
-
-### Formulation Agent
-- <이론 사실, governing assumptions, candidate equations, unresolved formulation questions>
-
-### Numerical Review Agent
-- <수치 위험, convergence expectations, patch test/MMS/MES evidence, source disagreements>
-
-### Reference Model Agent
-- <benchmark candidates, required reference artifacts, target quantities, source limitations>
-
-### Implementation Planning Agent
-- <verification scenarios and testable acceptance evidence>
-```
-
-## 품질 기준
-
-- 모든 외부 근거는 source metadata와 reliability tier를 가져야 한다.
-- verified fact와 inference를 구분해야 한다.
-- benchmark 후보는 target quantity와 applicability limit을 가져야 한다.
-- Abaqus Benchmarks Guide와 Abaqus Verification Guide의 용도를 구분해야 한다.
-- NAFEMS benchmark는 independent standard test와 target value 후보로 기록해야 한다.
-- MMS/MES 자료는 code verification 후보로 기록하되 정식화 확정으로 취급하지 않는다.
-- 사용 불가능하거나 유료/비공개인 자료는 Open Issues에 남겨야 한다.

docs/superpowers/plans/2026-08-12-independent-reference-validation-policy.md

@@ -14,7 +14,8 @@
 - Do not run Abaqus or another reference solver.
 - Do not modify C++ source, tests, or CMake.
 - Preserve HDF5 as authoritative FESA output.
-- Use the B33 mixed tolerance exactly for MITC4 U; U blocks and UR warns only.
+- Use fixed absolute `1.0e-5` for MITC4 U/UR; U blocks and UR warns only. Keep the
+  separate B33 mixed tolerance unchanged.
 - Retain fixed numerical drilling stabilization but exclude calibration and drilling evidence outputs.
 - Remove `NR-O03` and `NR-O04` from MITC4 implementation readiness.
 - Implementation Planning may create Harness phase files only after user approval of the Step draft.
@@ -38,7 +39,7 @@
 - [ ] Replace canonical/legacy/provenance bundle governance with the minimal reference-case contract.
 - [ ] Separate Abaqus input syntax and external comparison from FESA formulation behavior.
 - [ ] Record the new decision as a superseding ADR while preserving historical decisions.
-- [ ] State the MITC4 B33 tolerance reuse and fixed drilling scope without claiming implementation exists.
+- [ ] State the MITC4 fixed absolute tolerance and fixed drilling scope without claiming implementation exists.
 - [ ] Verify changed policy terms with targeted `rg` checks and `git diff --check`.
 - [ ] Commit with `docs: simplify FESA reference validation policy`.
 
@@ -63,15 +64,15 @@
 ### Task 3: Revise MITC4 requirements and formulation
 
 **Files:**
-- Modify: `docs/requirements/linear-static-mitc4-shell.md`
-- Modify: `docs/research/linear-static-mitc4-shell-research.md`
-- Modify: `docs/formulations/mitc4-shell-formulation.md`
+- Modify: `docs/linear-static-mitc4-shell/requirements.md`
+- Modify: `docs/linear-static-mitc4-shell/research.md`
+- Modify: `docs/linear-static-mitc4-shell/formulation.md`
 
 **Interfaces:**
 - Consumes: Tasks 1-2 policy and approved fixed drilling rule
 - Produces: implementation-ready feature scope without removed calibration gates
 
-- [ ] Correct the observed S4/S4R reference inventory and adopt the B33 U/UR mixed tolerance.
+- [ ] Correct the observed S4/S4R reference inventory and adopt fixed absolute `1.0e-5` for MITC4 U/UR.
 - [ ] Replace drilling candidate/sweep language with the fixed positive rotational-diagonal rule.
 - [ ] Remove drilling calibration, drilling-energy warning, `NR-O03`, `NR-O04`, and expanded portfolio requirements.
 - [ ] Preserve the future nonlinear derivation as documentation-only.
@@ -82,24 +83,24 @@
 ### Task 4: Revise MITC4 I/O and reference-case inventory
 
 **Files:**
-- Modify: `docs/io-definitions/linear-static-mitc4-shell-io.md`
-- Create: `docs/reference-models/linear-static-mitc4-shell-reference-models.md`
+- Modify: `docs/linear-static-mitc4-shell/io.md`
+- Create: `docs/linear-static-mitc4-shell/reference-model.md`
 
 **Interfaces:**
 - Consumes: Task 3 requirements/formulation
-- Produces: exact lightweight S4/S4R input-to-displacement-CSV pairing and HDF5 comparison mapping
+- Produces: exact lightweight full-integration S4 input-to-displacement-CSV pairing and HDF5 comparison mapping; S4R non-reference coverage
 
 - [ ] Remove canonical/legacy/metadata/provenance/schema-version readiness requirements.
 - [ ] Remove mandatory drilling stiffness, ratio, and energy output evidence.
 - [ ] Keep minimal source-node plus U/UR component matching and finite/unique-row checks.
-- [ ] Inventory the exact existing `reference/shell` and `reference/shellR` paths read-only.
+- [ ] Inventory only the exact existing `reference/shell` S4 paths as acceptance inputs; keep `reference/shellR` optional and read-only without consuming it.
 - [ ] Verify no file under `reference/` changed.
 - [ ] Commit with `docs: define lightweight MITC4 reference cases`.
 
 ### Task 5: Rerun MITC4 numerical review under the approved policy
 
 **Files:**
-- Modify: `docs/numerical-reviews/linear-static-mitc4-shell-review.md`
+- Modify: `docs/linear-static-mitc4-shell/numerical-review.md`
 
 **Interfaces:**
 - Consumes: Tasks 1-4 current source-of-truth documents

docs/superpowers/plans/2026-08-16-implementation-agent-terra-model.md

@@ -0,0 +1,70 @@
+# Implementation Agent Terra Model Implementation Plan
+
+> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.
+
+**Goal:** Configure the project-local `implementation-agent` to use `gpt-5.6-terra` while preserving its existing reasoning effort.
+
+**Architecture:** Add one explicit model override to the existing Implementation Agent TOML profile. Do not introduce shared model policy, modify other profiles, or change the agent's instructions.
+
+**Tech Stack:** TOML, Python 3 `tomllib`, Git
+
+## Global Constraints
+
+- Set `model` to exactly `gpt-5.6-terra` only in `.codex/agents/implementation-agent.toml`.
+- Preserve `model_reasoning_effort = "extra high"`.
+- Do not modify other custom agents, user-global Codex configuration, solver production code, or agent instructions.
+- Do not add or modify contract tests or other test files.
+- Verify the change only through TOML parsing, exact-value assertions, and Git diff inspection.
+
+---
+
+### Task 1: Add the Implementation Agent model override
+
+**Files:**
+- Modify: `.codex/agents/implementation-agent.toml`
+- Test: none, per the approved design
+
+**Interfaces:**
+- Consumes: the existing project-local `implementation-agent` TOML profile
+- Produces: `model = "gpt-5.6-terra"` with the existing `model_reasoning_effort = "extra high"`
+
+- [ ] **Step 1: Add the model setting**
+
+Insert the model key between `sandbox_mode` and `model_reasoning_effort` so the profile header is:
+
+```toml
+name = "implementation-agent"
+description = "Implements FESA solver features in C++17/MSVC by following approved TDD-first implementation plans."
+sandbox_mode = "workspace-write"
+model = "gpt-5.6-terra"
+model_reasoning_effort = "extra high"
+```
+
+- [ ] **Step 2: Parse the profile and verify the exact values**
+
+Run:
+
+```powershell
+python -c "import pathlib, tomllib; p = tomllib.loads(pathlib.Path('.codex/agents/implementation-agent.toml').read_text(encoding='utf-8')); assert p['model'] == 'gpt-5.6-terra'; assert p['model_reasoning_effort'] == 'extra high'"
+```
+
+Expected: exit code `0` with no output.
+
+- [ ] **Step 3: Verify the change is surgical**
+
+Run:
+
+```powershell
+git diff --check
+git diff -- .codex/agents/implementation-agent.toml
+```
+
+Expected: no whitespace errors, and the profile diff contains only the added
+`model = "gpt-5.6-terra"` line.
+
+- [ ] **Step 4: Commit the configuration change**
+
+```powershell
+git add -- .codex/agents/implementation-agent.toml
+git commit -m "chore: use Terra for implementation agent"
+```

docs/superpowers/specs/2026-08-08-linear-static-3d-euler-beam-design.md

@@ -7,11 +7,15 @@
 - status: `approved`
 - approved_by: user
 - approved_on: `2026-08-08`
-- amended_on: `2026-08-09`
-- source_formulation: `docs/formulations/3d-isoparametric-euler-beam-formulation.md`
-- reference_baseline: `reference/cantilever beam/` from source commit `2b34d0b`
+- amended_on: `2026-08-18`
+- source_formulation: `docs/linear-static-3d-euler-beam/formulation.md`
+- reference_baseline: exact read-only artifacts under `reference/cantilever beam/`
 - implementation_environment: C++17, MSVC, CMake, CTest, GoogleTest, Intel oneMKL, Intel oneTBB, HDF5
 
+2026-08-18 amendment: the former node-station projection and mixed absolute-floor tolerance are
+superseded by the direct element-endpoint identity and common family-scale row/RMS policy in
+`2026-08-17-common-reference-tolerance-design.md` and ADR-022.
+
 ## 1. 목적
 
 이 설계는 Abaqus `.inp` keyword subset을 읽어 2절점 3차원
@@ -76,13 +80,13 @@ authoritative HDF5 파일 `results.h5`에 기록하는 FESA V0 파이프라인
 
 | Step | Name | Primary output |
 | ---: | --- | --- |
-| 0 | `requirements-baseline` | `docs/requirements/linear-static-3d-euler-beam.md` |
-| 1 | `research-evidence` | `docs/research/linear-static-3d-euler-beam-research.md` |
-| 2 | `formulation-alignment` | `docs/formulations/3d-isoparametric-euler-beam-formulation.md` |
-| 3 | `numerical-review` | `docs/numerical-reviews/linear-static-3d-euler-beam-review.md` |
-| 4 | `io-contract` | `docs/io-definitions/linear-static-3d-euler-beam-io.md` |
-| 5 | `reference-model-contract` | `docs/reference-models/linear-static-3d-euler-beam-reference-models.md` |
-| 6 | `implementation-plan` | `docs/implementation-plans/linear-static-3d-euler-beam.md` |
+| 0 | `requirements-baseline` | `docs/linear-static-3d-euler-beam/requirements.md` |
+| 1 | `research-evidence` | `docs/linear-static-3d-euler-beam/research.md` |
+| 2 | `formulation-alignment` | `docs/linear-static-3d-euler-beam/formulation.md` |
+| 3 | `numerical-review` | `docs/linear-static-3d-euler-beam/numerical-review.md` |
+| 4 | `io-contract` | `docs/linear-static-3d-euler-beam/io.md` |
+| 5 | `reference-model-contract` | `docs/linear-static-3d-euler-beam/reference-model.md` |
+| 6 | `implementation-plan` | `docs/linear-static-3d-euler-beam/implementation-plan.md` |
 
 ### 3.2 C++ TDD 구현
 
@@ -111,10 +115,10 @@ authoritative HDF5 파일 `results.h5`에 기록하는 FESA V0 파이프라인
 
 | Step | Name | Primary output |
 | ---: | --- | --- |
-| 25 | `build-test-verification` | `docs/build-test-reports/linear-static-3d-euler-beam.md` |
-| 26 | `reference-verification` | `docs/reference-verifications/linear-static-3d-euler-beam-reference-verification.md` |
-| 27 | `physics-sanity` | `docs/physics-evaluations/linear-static-3d-euler-beam-physics-evaluation.md` |
-| 28 | `release-readiness` | `docs/releases/linear-static-3d-euler-beam-release.md` |
+| 25 | `build-test-verification` | `docs/linear-static-3d-euler-beam/build-test.md` |
+| 26 | `reference-verification` | `docs/linear-static-3d-euler-beam/reference-comparison.md` |
+| 27 | `physics-sanity` | `docs/linear-static-3d-euler-beam/physics-evaluation.md` |
+| 28 | `release-readiness` | `docs/linear-static-3d-euler-beam/release.md` |
 
 ## 4. 아키텍처와 소유권
 
@@ -173,7 +177,7 @@ V0에서는 velocity, acceleration, temperature, iteration history를 할당하
 ### 4.5 요소 계약
 
 `EulerBeam3D`는
-`docs/formulations/3d-isoparametric-euler-beam-formulation.md`의 부호, DOF 순서,
+`docs/linear-static-3d-euler-beam/formulation.md`의 부호, DOF 순서,
 2점 Gauss rule, transformation을 따른다. 구현 API는 다음 책임을 분리한다.
 
 ```cpp
@@ -288,8 +292,8 @@ node label 또는 node set이며 assembly-level set의 `INSTANCE` parameter를
 Abaqus B31은 transverse shear deformation을 포함하는 Timoshenko beam이고 B33은
 2절점 cubic Euler–Bernoulli beam이다. FESA V0는 `TYPE=B33`만 Euler 요소로 매핑한다.
 `TYPE=B31`은 `unsupported-element-formulation` 오류로 거부한다.
-`reference/cantilever beam/cantilever beam.inp`는 source commit `2b34d0b`에서
-`TYPE=B33`으로 생성된 승인 reference input이다. Reference artifact check는 비교 전에
+`reference/cantilever beam/cantilever beam.inp`는 `TYPE=B33`으로 생성된 승인 reference
+input이다. Reference artifact check는 비교 전에
 이 element type을 다시 확인하며 B31로 되돌아간 입력이나 결과를 허용하지 않는다.
 
 이 결정은 [[Abaqus Structural Element Families]], [[Beam and Frame Finite Elements]],
@@ -370,19 +374,20 @@ Abaqus internal-force CSV는 equilibrium end action이 아니라 section resulta
 | `SM1` | `MY` |
 | `SM2` | `MZ` |
 
-승인된 B33 reference의 `cantilever beam elemental forces.csv`는 element label 없이
-`Frame`, `Part Instance Name`, `Node Label`, `SF1`, `SM1`, `SM2`, `SM3`을 기록한다.
+승인된 B33 reference의 `cantilever beam elemental forces.csv`는 `Frame`,
+`Part Instance Name`, `Element Label`, `Node Label`, `SF1`, `SM1`, `SM2`, `SM3`을
+기록한다.
 따라서 이 파일의 comparison은 다음 규칙을 사용한다.
 
-- Reference row key는 `(Step-1, frame 0, instance name, source node label, component)`다.
+- Reference row key는 `(Step-1, frame 0, instance name, source element label,
+  endpoint source node label, component)`다.
   단일 step invariant에 따라 CSV의 `Increment 1: Step Time = 1.000`을
   `(Step-1, frame 0)`으로 정규화한다.
-- FESA endpoint section resultant를 동일한 section-cut 부호로 변환한 뒤 source node
-  station으로 정규화한다. Interior node의 두 인접 element endpoint 값은 먼저 승인된
-  tolerance 안에서 서로 일치해야 한다. 불일치는 평균으로 숨기지 않고
-  `tolerance-failure`로 보고한다.
-- Interior endpoint가 일치하면 stable internal element ID가 작은 endpoint를 대표 row로
-  선택한다. Boundary node는 하나의 endpoint를 사용한다.
+- 각 B33 element에는 input connectivity의 두 source node label과 일치하는 endpoint row가
+  정확히 두 개 있어야 하며, 각 row를 해당 HDF5 element endpoint에 직접 대응한다.
+  Missing, extra, duplicate 또는 connectivity mismatch는 tolerance 전에 실패한다.
+- Comparator는 같은 source node의 인접 endpoint를 node station으로 축약하거나 대표값을
+  선택하거나 평균하지 않는다.
 - 이 bundle에서 비교하는 mapping은 `SF1 -> N`, `SM1 -> My`, `SM2 -> Mz`,
   `SM3 -> T`다. CSV에 없는 `SF2`와 `SF3`는 reference comparison 대상이 아니며,
   FESA equilibrium end action의 transverse force는 unit/analytical test와 physics sanity로
@@ -496,8 +501,8 @@ GoogleTest target에는 FESA warning policy를 강제하지 않는다.
 - axial, torsion, y/z bending cantilever analytical cases
 - HDF5 schema, identity, component, metadata, atomic finalization
 - CLI `.inp -> results.h5` integration
-- 승인된 B33 CSV header/Frame 정규화, node-station matching, interior endpoint 일치 검사
-- zero 및 near-zero reference row를 포함한 component-scale 혼합 허용오차 검사
+- 승인된 B33 CSV header/Frame 정규화와 element-endpoint direct identity 검사
+- zero 및 near-zero reference row를 포함한 공통 family-scale 상대오차/RMS 검사
 
 ### 11.3 수치 tolerance
 
@@ -506,37 +511,33 @@ GoogleTest target에는 FESA warning policy를 강제하지 않는다.
 | matrix symmetry and Gauss/closed-form comparison | normalized `1e-12` |
 | rigid-mode and linear-system residual | normalized `1e-10` |
 | analytical solution tests | relative `1e-9` |
-| Abaqus B33 reference comparison | component-scale relative `1e-6` |
-| SI displacement and rotation absolute floor | `1e-9` |
-| SI force and moment absolute floor | `1e-3` |
+| Abaqus reference row relative error | `0.05` |
+| near-zero threshold ratio | reference family scale의 `0.01` |
+| reference family scale-relative RMS | `0.01` |
 
-Reference row 판정은 zero-reference에서도 의미가 있도록 component-scale 혼합
-허용오차를 사용한다. 같은 model, step/frame, quantity, component의 Abaqus reference
-rows에 대해 다음 값을 계산한다.
+Reference row 판정은 같은 model, step/frame, quantity, component로 구성한 reference
+family별 공통 규칙을 사용한다. Reference 값으로만 family scale을 계산한다.
 
 ```text
-reference_scale = max(abs(reference_value_i))
-row_tolerance = absolute_floor + 1e-6 * reference_scale
-row_pass = abs(fesa_value_i - reference_value_i) <= row_tolerance
+S = max(abs(reference_value_i))
+near_zero_i = abs(reference_value_i) <= 0.01 * S
+row_pass_i = abs(fesa_value_i - reference_value_i) / abs(reference_value_i) <= 0.05
+             if not near_zero_i
+row_pass_i = abs(fesa_value_i - reference_value_i) <= 0.01 * S
+             if near_zero_i
+relative_rms = sqrt(mean((fesa_value_i - reference_value_i)^2)) / S
+family_pass = all(row_pass_i) and relative_rms <= 0.01
 ```
 
-- Scale은 displacement, rotation, force, moment 및 각 component를 섞지 않는다.
+- Family는 displacement, rotation, force, moment 및 각 component를 섞지 않는다.
 - Scale은 read-only Abaqus reference 값만 사용하며 FESA 결과로 조정하지 않는다.
-- Scale이 0이면 absolute floor만 적용한다.
+- `S == 0`이면 모든 FESA 값이 정확히 0일 때만 통과한다. 하나라도 0이 아니면
+  `zero-reference-scale-nonzero-error`로 실패한다.
 - 모든 row를 개별 판정하며 reference 값을 zero-clamp하거나 row를 제거하지 않는다.
 - Missing/extra row, nonfinite value, schema 또는 identity mismatch는 tolerance 계산 전에
   실패한다.
-- Report는 max absolute error, component-scale normalized error, RMS error, norm error,
-  worst row/component를 기록한다.
-
-현재 B33 `SM1`의 reference scale은 `1.0e7`이고 moment row tolerance는
-`1.0e-3 + 1.0e-6 * 1.0e7 = 10.001`이다. 자유단 reference residue `-1.56e-2`와
-FESA의 이론적 zero 사이 오차 `1.56e-2`는 이 기준을 통과하지만, 예를 들어 `100`의
-자유단 moment 오차는 실패한다.
-
-승인된 cantilever bundle은 SI absolute floor를 사용한다. 이후 SI가 아닌 reference를
-추가하면 quantity별 absolute floor를 해당 단위로 변환하고 변환 근거를 metadata와
-verification report에 기록해야 한다.
+- Report는 family scale, max absolute error, max row relative error, scale-relative RMS,
+  worst row/component와 판정 reason을 기록한다. 독립적인 absolute-error gate는 없다.
 
 ### 11.4 공통 build/test command
 
@@ -561,9 +562,10 @@ CTest discovery 결과는 한 개 이상의 test를 포함해야 한다.
 ## 12. Reference 및 release gate
 
 `reference/cantilever beam/`을 V0의 승인된 Abaqus B33 reference baseline으로 사용한다.
-논리 model ID는 `cantilever-beam-b33`, reference schema ID는
-`abaqus-cae-report-csv-v0`이며 source baseline은 commit `2b34d0b`이다. Input header가
-기록한 generator는 `Abaqus/CAE Learning Edition 2024`다. 모델은 SI 단위계, global
+논리 model ID는 `cantilever-beam-b33`이다. 기존 `abaqus-cae-report-csv-v0`과 source
+commit `2b34d0b` 기록은 2026-08-09 당시 inventory의 역사적 provenance이며, 2026-08-18
+regenerated elemental-force CSV의 현재 readiness gate가 아니다. Input header가 기록한
+generator는 `Abaqus/CAE Learning Edition 2024`다. 모델은 SI 단위계, global
 Cartesian nodal output, beam local section-force output, `Step-1`, increment 1,
 step time 1.0 계약으로 해석한다.
 
@@ -588,8 +590,8 @@ Reference verification은 build/test gate 통과 후 승인 input으로 FESA `re
 
 - nodal displacement: CSV `U1/U2/U3/UR1/UR2/UR3`과 HDF5 displacement
 - nodal reaction: CSV `RF1/RF2/RF3/RM1/RM2/RM3`과 HDF5 reaction
-- section resultant: CSV `SF1/SM1/SM2/SM3`과 node station으로 정규화한 HDF5
-  section resultant
+- section resultant: CSV `SF1/SM1/SM2/SM3`과 element label 및 endpoint node label로
+  직접 대응한 HDF5 section resultant
 - stress: 명시적 N/A
 
 Artifact check는 네 파일의 존재, B33 element type, expected header, 유일한 row key,

docs/superpowers/specs/2026-08-12-independent-reference-validation-policy-design.md

@@ -3,11 +3,16 @@
 ## Status
 
 - date: `2026-08-12`
+- amended_on: `2026-08-18`
 - status: `approved`
 - scope: FESA project policy, agents, skills, and `linear-static-mitc4-shell` upstream contracts
 - implementation_code: out of scope
 - reference_artifact_mutation: prohibited
 
+2026-08-18 amendment: the former MITC4 fixed absolute tolerance is superseded by the common
+family-scale row/RMS policy in `2026-08-17-common-reference-tolerance-design.md` and ADR-022.
+The independent-solver, minimal-reference-case and drilling decisions remain unchanged.
+
 ## Decision
 
 FESA is an independent finite element solver. It accepts an approved subset of the
@@ -71,29 +76,36 @@ duplicated descriptive fields.
 
 ## MITC4 Reference Acceptance
 
-Both source element types map to the same internal `FESA-MITC4` formulation:
+Both source element types map to the same internal `FESA-MITC4` formulation, but the
+2026-08-13 feature-scope amendment uses only the full-integration S4 case for Abaqus
+acceptance comparison:
 
 - `reference/shell/shell.inp` and `reference/shell/shell displacements.csv` cover
   source `S4`;
-- `reference/shellR/shellR.inp` and
-  `reference/shellR/shellR displacements.csv` cover source `S4R`.
+- source `S4R` is covered by parser/common-kernel/deterministic-assembly/HDF5 metadata
+  tests without consuming an Abaqus S4R artifact.
 
 Only matched global `U1/U2/U3` rows are blocking. `UR1/UR2/UR3` rows are compared and
 reported as warning-only evidence. Reactions and stresses may be inspected but are not
 reference pass/fail quantities.
 
-For each case and component `c`:
+The project-wide common policy now evaluates each matched U/UR component family:
 
 ```text
-reference_scale_c = max_i(abs(abaqus_value_i,c))
-tolerance_c = 1e-9 + 1e-6 * reference_scale_c
+S = max(abs(reference_value_i))
+near_zero_i = abs(reference_value_i) <= 0.01 * S
+ordinary_row_pass_i = abs(error_i) / abs(reference_value_i) <= 0.05
+near_zero_row_pass_i = abs(error_i) <= 0.01 * S
+relative_rms = sqrt(mean(error_i^2)) / S <= 0.01
 ```
 
-The `1e-9` absolute floor is expressed in the model's user-consistent length unit for
-translation and as a dimensionless rotation value for warning-only UR. Values are not
-zero-clamped. Missing, duplicate, unmatched, or nonfinite required rows fail before
-numeric tolerance evaluation. U tolerance failure fails reference verification; UR
-tolerance exceedance emits a deterministic warning only.
+The scale is computed from read-only reference values in the same model, step/frame,
+quantity and component family. If `S == 0`, only exact-zero FESA values pass; otherwise
+the reason is `zero-reference-scale-nonzero-error`. There is no independent absolute-error
+gate. Values are not zero-clamped. Missing, duplicate, unmatched, or nonfinite required
+rows fail before numeric tolerance evaluation. U tolerance failure fails reference
+verification; UR tolerance exceedance emits a deterministic warning only. B33 uses the
+same row and family-RMS thresholds.
 
 No additional locking, convergence, distorted, curved, director-angle, or invalid
 geometry reference portfolio is required for this MITC4 implementation acceptance.
@@ -144,7 +156,7 @@ For MITC4:
 
 - fixed drilling stabilization closes `NR-O01` and removes `NR-O02` by scope;
 - `NR-O03` and `NR-O04` are removed from the approved verification scope;
-- the B33 mixed tolerance closes the blocking part of `NR-O05`;
+- the common family-scale row/RMS policy closes the blocking part of `NR-O05`;
 - missing heavyweight reference metadata is `not-required-by-policy`;
 - confirmed formulation defects remain blocking.
 

docs/superpowers/specs/2026-08-15-agent-skill-workflow-consolidation-design.md

@@ -0,0 +1,276 @@
+# FESA Agent/Skill 개발 프로세스 통합 설계
+
+## Metadata
+
+- date: 2026-08-15
+- status: approved-design
+- scope: FESA 개발 workflow, custom agent, project-local skill, agent 산출물 배치
+
+## 목표
+
+FESA 기능 개발 workflow를 요구사항 단위로 단순화한다. 중복된 agent와 skill을 실제로
+병합하고, `coordinator-agent`를 유일한 main agent로 두며, 모든 단계별 산출물을
+`docs/<feature-id>/`에 모은다.
+
+이 변경은 개발 운영 계약만 바꾼다. Solver C++ 구현, CMake/CTest target, Harness executor,
+hook 동작, Abaqus reference artifact와 승인된 FEM 기능 계약은 변경하지 않는다.
+
+## 핵심 결정
+
+1. 기존 reference model 계약 준비 단계는 Numerical Review 단계에 병합한다.
+2. 기존 Build/Test와 Reference Comparison 단계는 C++ Implementation 단계에 병합한다.
+3. 병합된 단계의 agent와 skill도 각각 하나로 병합한다.
+4. Agent 수를 줄이되 build/test와 reference comparison 증거 문서는 별도로 유지한다.
+5. 모든 agent 산출물은 `docs/<feature-id>/` 아래에 둔다.
+6. `coordinator-agent`는 main agent이고 나머지 agent는 Coordinator가 호출하는 sub-agent다.
+7. `correction-agent`는 정규 단계가 아니라 반복되거나 원인이 불명확한 실패를 다루는
+   재작업 sub-agent로 유지한다.
+
+## 단순화된 개발 프로세스
+
+| 단계 | 담당 sub-agent | 필수 skill | 산출물 |
+| --- | --- | --- | --- |
+| 1. 요구조건 | `requirement-agent` | `fesa-requirements-baseline` | `requirements.md` |
+| 2. 연구 | `research-agent` | `fesa-research-evidence`, 필요 시 `fem-theory-query` | `research.md` |
+| 3. 정식화 | `formulation-agent` | `fesa-formulation-spec` | `formulation.md` |
+| 4. 수치 검토 + reference model 계약 | `numerical-review-agent` | 확장된 `fesa-numerical-review` | `numerical-review.md`, `reference-model.md` |
+| 5. I/O 정의 | `io-definition-agent` | `fesa-io-contract` | `io.md` |
+| 6. 구현 계획 + C++ 구현 + build/test + reference comparison | `implementation-planning-agent`, `implementation-agent` | 확장된 `fesa-cpp-msvc-tdd`와 계획 시 project-local `harness` | `implementation-plan.md`, `implementation-report.md`, `build-test.md`, `reference-comparison.md` |
+| 7. 물리 검토 | `physics-evaluation-agent` | `fesa-physics-sanity` | `physics-evaluation.md` |
+| 8. 배포 준비 | `release-agent` | `fesa-release-readiness` | `release.md` |
+
+`coordinator-agent`는 위 단계 밖에서 전체 workflow와 handoff를 관리하고
+`coordination.md`를 소유한다. `correction-agent`는 Coordinator가 재작업을 요청할 때만
+`corrections.md`에 원인, 수정, 재검증 결과를 추가한다.
+
+## Agent 통합
+
+### Numerical Review 통합
+
+`reference-model-agent`의 책임을 `numerical-review-agent`에 흡수하고
+`.codex/agents/reference-model-agent.toml`을 제거한다.
+
+병합된 Numerical Review Agent는 다음 두 판단을 함께 소유한다.
+
+- 정식화의 차원, 부호, 좌표 변환, Jacobian, 적분, locking, rigid-body mode와 검증 준비도
+- 기존 reference input/CSV의 존재, 비교량, blocking/warning 정책, source identity/component,
+  row precheck와 승인된 tolerance
+
+기존 Reference Model Agent가 I/O 문서를 입력으로 요구하던 순환 의존성은 제거한다.
+`reference-model.md`는 exact input/CSV, 비교량, source identity/component와 tolerance를
+확정한다. 다음 I/O 단계가 이를 소비해 `io.md`에서 최종 HDF5 dataset projection과 CSV
+column mapping을 정의한다.
+
+### Implementation 통합
+
+`build-test-executor-agent`와 `reference-verification-agent`의 책임을
+`implementation-agent`에 흡수하고 다음 파일을 제거한다.
+
+- `.codex/agents/build-test-executor-agent.toml`
+- `.codex/agents/reference-verification-agent.toml`
+
+병합된 Implementation Agent는 한 단계 안에서 다음 순서를 완료한다.
+
+```text
+RED -> observed failure -> minimal GREEN -> focused VERIFY
+    -> full MSVC x64 Debug build/CTest -> reference artifact check
+    -> blocking/warning reference comparison -> report
+```
+
+하나의 agent가 수행하더라도 증거는 다음 세 문서로 분리한다.
+
+- `implementation-report.md`: 구현 범위, RED/GREEN/VERIFY, 코드 변경과 traceability
+- `build-test.md`: 환경, 명령, exit code, 실패 분류와 CTest 결과
+- `reference-comparison.md`: artifact inventory, row identity, tolerance와 quantity별 결과
+
+Implementation 단계의 최종 성공 상태는 `pass-for-physics-evaluation`이다. Compile, link,
+ordinary test 또는 명확한 implementation-owned mismatch는 Implementation Agent가 먼저
+수정한다. 같은 failure classification이 반복되거나 원인이 불명확하면 Coordinator가
+`correction-agent`를 호출한다.
+
+## Skill 통합
+
+### Numerical Review skill
+
+`fesa-reference-models`의 reference inventory, comparison scope, identity, row precheck,
+tolerance와 handoff 계약을 `fesa-numerical-review`에 흡수한다. 다음 디렉터리를 제거한다.
+
+- `.codex/skills/fesa-reference-models/`
+
+확장된 `fesa-numerical-review`는 `numerical-review.md`와 `reference-model.md`를 모두
+산출하고 I/O Definition Agent에 두 문서를 handoff한다.
+
+### C++ TDD skill
+
+`fesa-reference-comparison`의 `ARTIFACT CHECK -> COMPARE -> CLASSIFY -> REPORT`, deterministic
+row matching, nonfinite/누락/추가/중복 row rejection과 tolerance 적용 계약을
+`fesa-cpp-msvc-tdd`에 흡수한다. 다음 디렉터리를 제거한다.
+
+- `.codex/skills/fesa-reference-comparison/`
+
+확장된 `fesa-cpp-msvc-tdd`는 구현 계획, TDD 구현, MSVC build/CTest, failure correction과
+reference comparison을 하나의 구현 단계 절차로 정의한다. Project-local `harness`의 계획
+승인, phase materialization, executor와 hook 계약은 변경하지 않는다.
+
+통합 후 FESA workflow skill은 다음 8개다.
+
+- `fesa-requirements-baseline`
+- `fesa-research-evidence`
+- `fesa-formulation-spec`
+- `fesa-numerical-review`
+- `fesa-io-contract`
+- `fesa-cpp-msvc-tdd`
+- `fesa-physics-sanity`
+- `fesa-release-readiness`
+
+`fem-theory-query`와 project-local `harness`, `review`는 별도 보조 skill로 유지한다.
+
+## Agent 계층과 Coordinator Orchestration
+
+통합 후 `.codex/agents/`에는 main agent 1개와 sub-agent 10개가 남는다.
+
+- main: `coordinator-agent`
+- sub-agents: `requirement-agent`, `research-agent`, `formulation-agent`,
+  `numerical-review-agent`, `io-definition-agent`, `implementation-planning-agent`,
+  `implementation-agent`, `correction-agent`, `physics-evaluation-agent`, `release-agent`
+
+Coordinator는 다음 순서로 동작한다.
+
+```text
+INTAKE -> STATE AUDIT -> WORKLIST UPDATE -> SUB-AGENT DISPATCH
+       -> EVIDENCE CHECK -> GATE DECISION -> STATUS REPORT
+```
+
+Coordinator는 한 번에 다음 유효 단계의 owner만 호출한다. Sub-agent는 전달받은 단계와
+산출물만 처리하고 peer agent를 직접 호출하거나 다음 단계로 진행하지 않는다. 완료 후
+산출물 경로, status, 핵심 evidence, blocker를 Coordinator에 반환한다.
+
+`docs/<feature-id>/coordination.md`는 다음 정보를 중심으로 작성한다.
+
+- feature scope와 현재 목표
+- 8단계 worklist
+- 단계별 owner, 상태, 산출물 경로와 완료 evidence
+- 현재 실행 중인 sub-agent와 다음 handoff
+- blocker, 사용자 결정과 failure classification
+- correction attempt count와 반복 실패 stop condition
+- gate transition decision log와 최종 workflow closure
+
+Worklist 항목 상태는 `pending | in-progress | passed | needs-rework | blocked`만 사용한다.
+동일 failure classification이 두 번 반복되면 자동 재작업을 중단하고 전체 workflow를
+`needs-user-decision` 또는 `blocked`로 전환한다.
+
+## Gate 계약
+
+1. Requirements gate: `requirements.md`가 승인된 범위, acceptance criteria, 검증량과
+   tolerance를 가진다.
+2. Research gate: `research.md`가 필요한 이론과 검증 evidence를 제공한다.
+3. Formulation gate: `formulation.md`가 구현 가능한 수치 계약을 제공한다.
+4. Numerical/reference gate: `numerical-review.md`와 `reference-model.md`가 모두 통과한다.
+5. I/O gate: `io.md`가 reference identity를 authoritative HDF5 dataset과 연결한다.
+6. Implementation gate: 구현 계획, RED/GREEN evidence, MSVC build/CTest와 blocking reference
+   comparison이 모두 통과한다.
+7. Physics gate: `physics-evaluation.md`가 `pass-for-release-agent`다.
+8. Release gate: `release.md`가 `ready-for-release`이고 Coordinator가 closure를 기록한다.
+
+## 요구사항 단위 산출물 구조
+
+```text
+docs/<feature-id>/
+├── coordination.md
+├── requirements.md
+├── research.md
+├── formulation.md
+├── numerical-review.md
+├── reference-model.md
+├── io.md
+├── implementation-plan.md
+├── implementation-report.md
+├── build-test.md
+├── reference-comparison.md
+├── corrections.md
+├── physics-evaluation.md
+└── release.md
+```
+
+산출물이 필요하지 않았거나 기존 workflow에서 생성되지 않은 경우 빈 placeholder 문서를
+만들지 않는다. `corrections.md`는 실제 correction이 발생할 때만 생성한다.
+
+## 기존 문서 마이그레이션
+
+현재 agent별 폴더의 기능 문서를 다음 두 요구사항 폴더로 이동한다.
+
+- `docs/linear-static-3d-euler-beam/`
+- `docs/linear-static-mitc4-shell/`
+
+파일 내용과 당시 `owner_agent`, status, numerical verdict는 역사적 evidence이므로 바꾸지
+않는다. 내부 문서 경로만 새 위치로 갱신한다. 삭제된 agent 이름이 기존 evidence의
+소유자로 기록된 경우도 그대로 보존한다.
+
+기존 `docs/requirements/`, `docs/research/`, `docs/formulations/`,
+`docs/numerical-reviews/`, `docs/io-definitions/`, `docs/reference-models/`,
+`docs/implementation-plans/`, `docs/build-test-reports/`, `docs/corrections/`,
+`docs/reference-verifications/`, `docs/physics-evaluations/`, `docs/releases/`,
+`docs/coordination/`의 기능 문서와 필요한 내부 링크를 옮긴 뒤 agent별 README는 제거한다.
+공통 운영 계약은 `AGENTS.md`, `docs/SOLVER_AGENT_DESIGN.md`,
+`docs/SOLVER_SKILL_DESIGN.md`에 통합한다.
+
+기존 두 기능은 `ready-for-release` evidence를 가지고 있으므로 새 `coordination.md`에서
+`completed`로 inventory한다. 누락된 과거 산출물을 새 문서로 꾸미거나 새 workflow를
+소급 실행하지 않는다.
+
+다음 항목은 이동하거나 변경하지 않는다.
+
+- `docs/PRD.md`, `docs/ARCHITECTURE.md`, `docs/ADR.md`
+- `docs/reference-papers/`와 기타 공통 기술 문서
+- `reference/` 아래의 Abaqus input/CSV artifact
+- `phases/`의 실행 상태와 Harness 실행 결과 본문. 이동된 문서를 가리키는 path reference만
+  기계적으로 갱신할 수 있다.
+- Solver C++ source, tests, CMake/CTest configuration
+
+## 문서와 참조 갱신
+
+현재 실행 계약에서는 제거된 agent/skill과 기존 agent별 산출물 경로를 사용하지 않는다.
+다음 파일군을 새 workflow와 `docs/<feature-id>/` 경로로 갱신한다.
+
+- `AGENTS.md`
+- `.codex/agents/*.toml`
+- `.codex/skills/*/SKILL.md`와 대상 `agents/openai.yaml`
+- `docs/SOLVER_AGENT_DESIGN.md`
+- `docs/SOLVER_SKILL_DESIGN.md`
+- 이동한 기능 문서의 내부 링크
+- 기존 경로를 입력으로 사용하는 관련 계획, spec, phase 문서
+
+역사적 산출물 본문에 기록된 과거 agent 이름과 status는 stale reference 검사 대상에서
+제외하지만, 새 입력 경로를 가리키는 링크는 유효해야 한다.
+
+## 검증 전략
+
+`tests/test_agent_skill_workflow_contract.py`를 추가해 다음 정적 계약을 검증한다.
+
+- 남은 agent 파일 집합이 main 1개와 sub-agent 10개인지 확인
+- 제거 대상 agent 3개와 skill 2개가 존재하지 않는지 확인
+- Coordinator가 main 역할과 `docs/<feature-id>/coordination.md` worklist를 정의하는지 확인
+- 모든 다른 agent가 sub-agent 역할과 Coordinator handoff를 정의하는지 확인
+- 병합 대상 skill의 핵심 workflow와 output contract가 destination skill에 존재하는지 확인
+- 현재 실행 계약에 제거된 agent/skill 이름이나 기존 agent별 출력 경로가 없는지 확인
+- 모든 TOML이 parse되고 각 skill의 frontmatter와 UI metadata가 유효한지 확인
+
+검증 명령은 다음과 같다.
+
+```powershell
+uv run --with pytest python -m pytest -v -rs
+python <skill-creator>/scripts/quick_validate.py .codex/skills/<remaining-skill>
+git diff --check
+```
+
+`quick_validate.py`는 남아 있는 project-local skill 각각에 실행한다. 이번 변경은 C++
+production과 CMake/CTest를 수정하지 않으므로 MSVC build와 CTest는 필수 검증 범위가 아니다.
+
+## 완료 기준
+
+- 8단계 workflow가 `AGENTS.md`, agent config와 설계 문서에서 동일하다.
+- 삭제 대상 agent와 skill이 제거되고 모든 책임과 quality gate가 destination에 보존된다.
+- `coordinator-agent`가 main agent로서 sub-agent dispatch와 worklist 갱신을 소유한다.
+- 모든 새 산출물 경로가 `docs/<feature-id>/` 규칙을 따른다.
+- 기존 기능 문서가 새 폴더에 모이고 역사적 verdict가 보존된다.
+- 자동 계약 테스트, 전체 Python pytest, skill validation과 `git diff --check`가 통과한다.

docs/superpowers/specs/2026-08-16-cpp-object-oriented-modular-refactoring-design.md

@@ -0,0 +1,355 @@
+# FESA C++ Object-Oriented Modular Refactoring Design
+
+## 상태
+
+- 설계 대화 승인: 2026-08-16
+- 서면 spec 리뷰: 승인 완료 (2026-08-16)
+- 구현 상태: not-started
+
+## 목적
+
+현재 B33 Euler beam, MITC4 shell과 linear static solver의 수치 및 외부 동작을
+유지하면서 C++ production code를 객체 책임 중심으로 재구성한다. 이번 리팩터링은
+Google C++ Style Guide 기반의 일관된 코드 스타일, 중복 제거, production Doxygen 문서화,
+명시적인 추상 경계와 응집된 모듈을 제공해야 한다.
+
+효율성은 runtime 성능 향상이 아니라 다음 유지보수 특성을 의미한다.
+
+- 새 element, element property, material, analysis, load 또는 boundary condition을 추가할
+  때 기존 concrete 구현을 수정하는 범위를 줄인다.
+- 하나의 의미를 여러 translation unit에서 다시 구현하지 않는다.
+- parser semantic data, numerical kernel, assembly, recovery와 output 책임을 구분한다.
+- 수치식과 I/O 계약을 사람이 직접 대조할 수 있는 단순한 dependency direction을 유지한다.
+
+## 범위
+
+### 포함
+
+- production 및 test C++ 전체의 Google-style naming과 formatting 전환
+- production header의 `.hpp`에서 `.h`로 전환과 header guard 적용
+- production 함수와 class의 Doxygen 문서화
+- `ElementDefinition`, `Element`, `ElementProperty`, `Material`, `Analysis`, `Load`,
+  `BoundaryCondition` 추상 경계
+- B33, MITC4, isotropic linear elasticity, beam/shell property, linear static analysis,
+  concentrated nodal load와 prescribed displacement의 concrete 구현 연결
+- `Vector3`, source-target resolution, DOF invariant validation과 dense-BLAS adapter의
+  중복 제거
+- `domain_mapper.cpp`, `hdf5_results_writer.cpp`, `result_recovery.cpp`의 책임별 분할
+- style, Doxygen, MSVC/CTest와 reference comparison 검증
+- `docs/CODINGSTYLE.md` 유지관리 문서와 Implementation Agent 필수 참조 연결
+
+### 제외
+
+- MITC3, solid hexa/tetra 또는 다른 element 구현
+- density, plasticity, anisotropic material 동작 구현
+- dynamic, eigenvalue, response spectrum 또는 random vibration analysis 구현
+- distributed load, body force 또는 MPC 구현
+- 승인된 formulation, sign, units, coordinate, HDF5 schema, reference artifact 또는
+  tolerance 변경
+- runtime 성능 최적화 또는 parallel reduction policy 변경
+- general plugin registry, global static registration 또는 shared ownership framework
+
+## 근거와 제약
+
+현재 production code는 semantic record와 numerical kernel을 이미 구분하지만 Domain은
+element, material, property, load와 boundary를 concrete vector로 각각 소유한다.
+`SparseAssembler`, `DofManager`와 `ResultRecovery`는 B33/MITC4 concrete storage를 직접
+알아야 한다. B33과 MITC4라는 두 실제 element가 있으므로 element abstraction은 현재
+구현으로 검증할 수 있다.
+
+반면 아직 구현되지 않은 plastic integration, dynamic state, MPC enforcement의 메서드를
+base class에 미리 추가할 근거는 없다. 추상 class는 현재 concrete 구현이 실제로 공유하는
+계약만 제공하고 future capability는 해당 기능의 requirements/formulation/ADR이 승인될 때
+추가한다.
+
+다음 기존 계약은 리팩터링보다 우선한다.
+
+- Domain은 semantic definition을 단독 소유하고 parsing 이후 불변으로 취급한다.
+- AnalysisModel은 Domain을 복사하지 않는 non-owning stable-index view다.
+- DofManager는 DOF와 equation numbering 및 sparse pattern을 단독 소유한다.
+- assembly worker는 global CSR storage를 직접 수정하지 않는다.
+- stiffness partition과 factorization은 load assembly보다 먼저 수행한다.
+- reaction과 free-equilibrium evidence는 full residual `K*d-F`에서 구한다.
+- result recovery와 final HDF5는 candidate validation 후 commit한다.
+- B33 및 MITC4 reference identity와 tolerance는 변경하지 않는다.
+
+## 추상 계층과 소유권
+
+```text
+Domain
+├─ ElementDefinition*
+│  ├─ EulerBeam3DDefinition
+│  └─ Mitc4ShellDefinition
+├─ ElementProperty*
+│  ├─ GeneralBeamSection
+│  └─ ShellSection
+├─ Material*
+│  └─ IsotropicLinearElasticMaterial
+└─ StepDefinition
+   ├─ Load*
+   │  └─ ConcentratedNodalLoad
+   └─ BoundaryCondition*
+      └─ PrescribedDisplacementBoundaryCondition
+
+Analysis
+└─ LinearStaticAnalysis
+
+Element
+├─ EulerBeam3D
+└─ Mitc4Shell
+```
+
+Domain은 각 base type을 `std::unique_ptr`로 단독 소유한다. Public access는 const이며
+collection의 vector position은 기존 stable `EntityIndex` 의미를 유지한다. AnalysisModel과
+후속 solver object는 raw ownership을 획득하지 않고 Domain 수명 안에서 index 또는 const
+reference만 사용한다. Copy를 지원하기 위한 speculative `Clone()`과 `std::shared_ptr`는
+추가하지 않는다.
+
+`ElementDefinition`과 `Element`는 서로 다른 책임이다.
+
+- `ElementDefinition`은 source identity, source element type, node connectivity와
+  property/material identity를 제공하는 semantic model이다.
+- `Element`는 active DOF layout, stiffness contribution, transformation과 result recovery를
+  제공하는 numerical kernel이다.
+- `ElementFactory`는 definition, property와 material compatibility를 검증한 뒤 concrete
+  kernel을 만든다.
+- 잘못된 조합은 `dynamic_cast` 실패나 undefined behavior로 넘기지 않고 기존
+  `Status`/`Result<T>` diagnostic으로 fail-closed 처리한다.
+
+Element result는 모든 element에 의미 없는 field를 추가한 비대한 base record가 아니다.
+공통 step/frame/source identity를 가진 backend-neutral result bundle이 beam 및 shell row를
+각자의 명확한 record로 보관한다. ResultRecovery와 ResultsWriter는 stable row identity를
+보존하며 서로 다른 result location을 평균하거나 합치지 않는다.
+
+## Material과 Element Property
+
+`Material` base는 identity, source location과 수명 계약만 제공한다. 현재 concrete type은
+물리 의미를 드러내도록 `IsotropicLinearElasticMaterial`로 명명한다. 현재 element factory가
+필요로 하는 isotropic elastic capability만 노출한다.
+
+다음 future concern은 이번 interface에 빈 메서드나 optional field로 미리 넣지 않는다.
+
+- density와 inertia contribution
+- anisotropic elastic constitutive data
+- plastic history state와 return mapping
+- temperature 또는 rate dependency
+
+이 concern은 각 기능이 승인될 때 별도의 capability 또는 구성 객체로 추가한다. 같은
+원칙으로 `ElementProperty`는 identity를 제공하고 `GeneralBeamSection`과 `ShellSection`은
+각자 필요한 기하 property를 소유한다. Solid property를 예상해 비어 있는 thickness/area
+accessor를 base에 추가하지 않는다.
+
+## Analysis 계층
+
+현재 `Analysis` base의 8개 protected hook은 linear-static lifecycle에 특화되어 있다.
+이를 모든 future procedure에 강제하지 않는다.
+
+```cpp
+class Analysis {
+ public:
+  virtual ~Analysis() = default;
+  virtual Status Run(const AnalysisRequest& request) = 0;
+};
+```
+
+현재 승인 순서는 `LinearStaticAnalysis::Run()`의 private 단계로 유지한다.
+
+```text
+initialize
+-> build analysis model
+-> build DOF map and sparse pattern
+-> assemble and partition stiffness
+-> factorize Kff
+-> assemble loads and effective RHS
+-> substitute and reconstruct
+-> recover and write results
+```
+
+Dynamic, eigenvalue와 stochastic procedure는 추가될 때 별도 state, equation, solver와 output
+lifecycle을 정의한다. 기존 linear-static hook 사이에 condition이나 unused future state를
+추가하지 않는다. 이 책임 변경은 구현 전에 ADR-007을 대체하거나 개정하는 ADR로 기록한다.
+
+## Load와 Boundary Condition
+
+`Load` concrete object는 자신의 semantic target과 magnitude를 소유하고 ordered full-DOF
+contribution을 생성한다. `LoadAssembler`는 active source order로 contribution을 모아 기존
+fixed accumulation order로 global vector에 반영한다. Polymorphic load가 global vector를
+직접 병렬 갱신하지 않는다.
+
+현재 concrete load는 `ConcentratedNodalLoad`다. Future distributed load와 body force는
+element-local contribution을 생성할 수 있지만 stable global reduction은 계속 assembler가
+소유한다.
+
+`BoundaryCondition`은 enforcement algorithm을 직접 수행하지 않고 constraint definition을
+생성한다. 현재 concrete type은 nonzero 값을 포함하는
+`PrescribedDisplacementBoundaryCondition`이다. `EssentialConstraintPolicy`가 기존 stable
+elimination과 full/reduced reconstruction을 수행한다.
+
+Future MPC는 별도 constraint equation과 enforcement policy를 요구한다. Prescribed
+displacement, MPC, penalty와 Lagrange multiplier를 하나의 bool/enum branch가 누적된 class로
+합치지 않는다.
+
+## 공통 수학과 중복 제거
+
+### Vector3
+
+좌표, local axis, shell director와 cross-product는 고정 크기 `Vector3` 값 class를 사용한다.
+동적 크기와 MKL-backed storage를 소유하는 기존 `Vector`와 역할을 섞지 않는다.
+
+`Vector3`는 현재 반복되는 다음 연산을 한 번만 정의한다.
+
+- component access
+- addition, subtraction과 scalar multiplication
+- `Dot()`
+- `Cross()`
+- `Norm()`
+- `Normalized()`
+- `IsFinite()`
+
+Normalization failure policy는 호출 위치에서 기존 scale-aware diagnostic을 유지한다.
+`Vector3`가 임의 tolerance, zero clamp 또는 solver diagnostic을 소유하지 않는다.
+
+### 다른 공통 책임
+
+- `SourceTargetResolver`: source label, instance와 set target을 stable identity로 해석한다.
+- `DofManager::ValidateInvariants()`: full/free/constrained ordering과 equation mapping을 owner가
+  한 번 검증한다.
+- private dense-BLAS adapter: Matrix와 Vector의 MKL integer conversion 및 copy operation을
+  공유한다. Vendor type은 public header에 노출하지 않는다.
+- ASCII utility: case-insensitive name comparison과 positive source-label parsing을 공유한다.
+
+중복 제거는 같은 의미와 failure policy가 반복될 때만 적용한다. 이름만 비슷하지만 units,
+identity 또는 diagnostic owner가 다른 계산을 하나로 합치지 않는다. State가 없는 helper를
+static-only class로 포장하지 않고 internal namespace/module을 사용한다.
+
+## 모듈 구조
+
+```text
+include/fesa/
+├─ analysis/
+│  ├─ analysis.h
+│  └─ linear_static_analysis.h
+├─ elements/
+│  ├─ element.h
+│  ├─ element_definition.h
+│  ├─ element_factory.h
+│  ├─ euler_beam_3d.h
+│  └─ mitc4_shell.h
+├─ properties/
+│  ├─ element_property.h
+│  ├─ general_beam_section.h
+│  └─ shell_section.h
+├─ materials/
+│  ├─ material.h
+│  └─ isotropic_linear_elastic_material.h
+├─ loads/
+│  ├─ load.h
+│  └─ concentrated_nodal_load.h
+├─ constraints/
+│  ├─ boundary_condition.h
+│  ├─ prescribed_displacement.h
+│  └─ essential_constraint_policy.h
+├─ math/
+│  ├─ vector.h
+│  ├─ vector3.h
+│  ├─ matrix.h
+│  └─ sparse_matrix.h
+└─ model/
+   ├─ domain.h
+   ├─ analysis_model.h
+   └─ source_target_resolver.h
+```
+
+`model_types.hpp`의 unrelated record는 각 owner module로 이동한다. Top-level orchestration
+file은 다음과 같이 분리한다.
+
+- Abaqus mapping: topology, material/property, step/load/BC와 final Domain assembly
+- HDF5 output: RAII/primitives, model dataset, result dataset, self-check와 atomic finalization
+- Result recovery: global equilibrium, beam recovery, shell recovery와 atomic state commit
+
+Public header와 implementation dependency direction을 역전하지 않는다. MKL, TBB, HDF5와
+Win32 type은 기존 adapter/private implementation 경계 안에 남는다.
+
+## 코드 스타일과 문서화
+
+`docs/CODINGSTYLE.md`를 FESA C++ style의 project-local source of truth로 사용한다. Google
+C++ Style Guide가 baseline이고 FESA 계약이 우선한다.
+
+주요 결정은 다음과 같다.
+
+- C++17/MSVC 호환을 유지한다. Google guide의 현재 C++20 language target은 적용하지 않는다.
+- 함수와 accessor를 포함한 production API는 PascalCase로 전면 전환한다.
+- type은 PascalCase, 변수는 snake_case, constant/enumerator는 `kPascalCase`, class member는
+  trailing underscore를 사용한다.
+- Header는 `.h`와 full-path Google header guard를 사용한다.
+- Source는 기존 FESA/CMake 관례인 `.cpp`를 유지하는 project exception으로 둔다.
+- Formatting은 `BasedOnStyle: Google`, 2-space indentation과 80-column limit를 사용한다.
+- Production public/protected declaration에는 Doxygen contract를 기록한다.
+- Production internal function은 definition에 목적과 비자명한 수치/순서 의미를 기록한다.
+- Test code에는 Doxygen coverage를 요구하지 않는다.
+
+Repository는 `.clang-format`, selected C++17-compatible `.clang-tidy`, `Doxyfile`과 optional
+CMake docs target을 제공한다. Generated HTML은 source control에 넣지 않는다.
+
+Implementation Agent의 profile은 구현 전에 `docs/CODINGSTYLE.md`를 mandatory global input으로
+읽도록 변경한다. Agent workflow contract test는 해당 profile이 문서를 직접 참조하는지
+검증한다.
+
+## 오류 처리
+
+- 모든 polymorphic base는 public virtual destructor를 갖는다.
+- Factory는 null object를 성공 결과로 반환하지 않는다.
+- Element/property/material incompatibility는 structured model diagnostic으로 거부한다.
+- Public solver 경계는 기존 `Status`/`Result<T>`를 사용한다.
+- Backend exception은 현재 failure category와 atomicity contract를 유지해 번역한다.
+- Unknown future kind를 silent fallback이나 default concrete type으로 바꾸지 않는다.
+- Failed candidate는 Domain, AnalysisState 또는 final HDF5를 부분 변경하지 않는다.
+
+## 단계적 마이그레이션
+
+1. 기존 unit/integration/reference 및 HDF5 contract baseline을 기록하고 architecture ADR을
+   갱신한다.
+2. `.clang-format`, header rename/guard와 PascalCase를 module slice별 mechanical change로
+   적용한다.
+3. `Vector3`, ASCII utility, SourceTargetResolver, DOF invariant validation과 private BLAS
+   adapter를 도입한다.
+4. Domain semantic hierarchy와 current concrete material/property/load/boundary type을
+   연결한다.
+5. Element runtime hierarchy와 factory를 DofManager, SparseAssembler와 ResultRecovery에
+   연결한다.
+6. Ordered load contribution과 essential constraint policy를 연결한다.
+7. Minimal Analysis base와 LinearStaticAnalysis-owned lifecycle로 전환한다.
+8. Mapper, HDF5 writer와 recovery를 책임별로 분할하고 Doxygen/style coverage를 완료한다.
+
+Mechanical formatting, API rename와 semantic restructuring을 같은 review unit에 섞지 않는다.
+각 slice는 buildable하고 독립 검증 가능해야 한다.
+
+## TDD와 검증
+
+각 production change는 관련 C++ test와 같은 Step에서 `RED -> GREEN -> VERIFY`를 수행한다.
+
+- abstract base와 concrete polymorphic use를 검증하는 compile-time/unit test
+- factory success와 incompatible property/material rejection test
+- base interface를 통한 B33/MITC4 stiffness 및 recovery test
+- stable element, load와 boundary source-order test
+- Domain ownership, AnalysisModel lifetime와 stable identity test
+- Vector3 arithmetic, finite and normalization-boundary test
+- 기존 parser/I/O, HDF5 schema와 atomicity test
+- B33 및 MITC4 integration/reference comparison
+- repeated execution의 sparse structure, result row와 diagnostic order test
+- Doxygen warning, formatting과 selected lint check
+- full MSVC x64 Debug `/W4 /WX` build와 CTest
+
+수치 산술 순서를 의도적으로 변경하지 않은 slice는 가능한 한 exact equality를 요구한다.
+Feature-approved reference tolerance는 최종 external comparison에만 그대로 적용한다.
+
+## 완료 기준
+
+- 승인된 abstraction과 current concrete implementation이 base interface를 통해 연결된다.
+- DofManager, assembler와 recovery에 B33/MITC4 type branch 또는 duplicate geometry helper가
+  남지 않는다.
+- Production 및 test C++가 `docs/CODINGSTYLE.md`의 naming/formatting 규칙을 만족한다.
+- Production API와 non-obvious internal function에 요구된 Doxygen가 존재한다.
+- Implementation Agent profile이 `docs/CODINGSTYLE.md`를 mandatory input으로 참조한다.
+- 전체 MSVC x64 Debug build/CTest와 B33/MITC4 reference comparison이 통과한다.
+- HDF5 schema, stable identity, diagnostic, tolerance와 reference artifact에 변경이 없다.

docs/superpowers/specs/2026-08-16-implementation-agent-terra-model-design.md

@@ -0,0 +1,38 @@
+# Implementation Agent Terra 모델 지정 설계
+
+## Metadata
+
+- date: 2026-08-16
+- status: approved-design
+- scope: project-local `implementation-agent` 모델 선택
+
+## 목표
+
+FESA의 `implementation-agent`가 명시적으로 `gpt-5.6-terra`를 사용하도록 설정한다.
+기존 `model_reasoning_effort = "extra high"` 설정은 그대로 유지한다.
+
+## 설계
+
+`.codex/agents/implementation-agent.toml`에 다음 모델 설정만 추가한다.
+
+```toml
+model = "gpt-5.6-terra"
+```
+
+이 설정은 프로젝트 로컬 `implementation-agent` 프로필에만 적용된다. 다른 custom agent,
+사용자 전역 Codex 설정, agent 지시문과 solver production 코드는 변경하지 않는다.
+
+## 검증
+
+새 계약 테스트나 테스트 파일 변경은 추가하지 않는다. 변경 후 다음 항목만 확인한다.
+
+1. `implementation-agent.toml`이 유효한 TOML로 파싱된다.
+2. `model` 값이 정확히 `gpt-5.6-terra`이다.
+3. `model_reasoning_effort` 값이 기존의 `extra high`로 유지된다.
+4. Git diff에 설계된 설정 외의 구현 변경이 없다.
+
+## 완료 조건
+
+- `implementation-agent`에만 `gpt-5.6-terra` 모델 override가 존재한다.
+- 기존 reasoning effort와 agent 동작 계약은 변경되지 않는다.
+- 새 계약 테스트는 추가되지 않는다.

docs/superpowers/specs/2026-08-17-common-reference-tolerance-design.md

@@ -0,0 +1,227 @@
+# Common Reference Tolerance Policy Design
+
+## Status
+
+- date: `2026-08-17`
+- status: `implemented-and-verified`
+- verified_on: `2026-08-18`
+- scope: 모든 FESA 외부 reference comparison
+- production_solver_behavior: 변경 없음
+- reference_artifact_mutation: 금지
+
+현재 운영 상수, 검증 순서, report schema와 변경 관리는 `docs/TOLERANCE.md`를 source of
+truth로 사용한다. 이 문서는 승인 결정의 설계 근거와 대안 검토를 보존한다.
+
+## 목적
+
+B33, MITC4 및 이후 추가되는 기능이 서로 다른 수치 tolerance 식을 만들지 않도록 공통
+reference comparison 규칙을 정의한다. 이 정책은 Abaqus CSV와 FESA HDF5에서 결정적으로
+대응된 행을 비교하는 외부 reference gate에만 적용한다. Element formulation, analytical
+solution, matrix symmetry, residual 및 physics sanity tolerance는 이 정책의 범위가 아니다.
+
+## 결정
+
+모든 reference comparison은 다음 세 개의 무차원 상수를 사용한다.
+
+```text
+near_zero_ratio = 0.01
+relative_tolerance = 0.05
+relative_rms_tolerance = 0.01
+```
+
+독립적인 absolute-error gate는 두지 않는다. Absolute error는 reference 값이 0에 가까워
+행별 상대오차를 안정적으로 계산할 수 없는 경우의 대체 판정과 진단에만 사용한다.
+
+## Comparison Family와 Scale
+
+Tolerance scale은 개별 component가 아니라 동일한 물리 차원의 component family마다
+계산한다. 하나의 family는 최소한 다음 identity가 모두 같은 행으로 제한한다.
+
+- model/reference case
+- step 및 frame
+- logical quantity
+- unit dimension
+- coordinate system
+- blocking 또는 warning-only behavior
+
+서로 다른 quantity나 단위 차원의 값은 scale을 공유하지 않는다. Family scale은 대응된
+reference 값만 사용해 계산한다.
+
+\[
+S_g = \max_{i \in g}|R_i|
+\]
+
+여기서 \(g\)는 comparison family, \(R_i\)는 원본 reference 값이다. Reference 값은 scale
+계산 전후에 clamp, rewrite 또는 생략하지 않는다.
+
+현재 기능의 family mapping은 다음과 같다.
+
+| feature | logical quantity / dimension | components | behavior |
+| --- | --- | --- | --- |
+| B33 | displacement / length | `UX`, `UY`, `UZ` | blocking |
+| B33 | displacement / rotation | `URX`, `URY`, `URZ` | blocking |
+| B33 | reaction / force | `RF1`, `RF2`, `RF3` | blocking |
+| B33 | reaction / moment | `RM1`, `RM2`, `RM3` | blocking |
+| B33 | section resultant / force | `N` | blocking |
+| B33 | section resultant / moment | `T`, `My`, `Mz` | blocking |
+| MITC4 | displacement / length | `U1`, `U2`, `U3` | blocking |
+| MITC4 | displacement / rotation | `UR1`, `UR2`, `UR3` | warning-only |
+
+새 quantity는 해당 기능의 requirements와 reference-model 계약에서 logical quantity,
+dimension, components 및 behavior를 명시해 family에 배치한다. 물리 차원이 같다는 이유만으로
+서로 다른 logical quantity를 자동으로 합치지 않는다.
+
+### B33 element-endpoint identity amendment
+
+2026-08-18에 사용자가 다시 생성한
+`reference/cantilever beam/cantilever beam elemental forces.csv`는 `Element Label`과
+`Node Label`을 모두 포함하고 B33 요소마다 두 endpoint 행을 제공한다. Section-resultant
+comparison은 다음 key로 각 CSV 행을 HDF5 `[element, endpoint, component]` 행에 직접
+대응시킨다.
+
+```text
+(model_id, step/frame, instance_name, source_element_label,
+ source_node_label, quantity, component)
+```
+
+Component mapping은 `SF-SF1 -> N`, `SM-SM1 -> My`, `SM-SM2 -> Mz`,
+`SM-SM3 -> T`다. Comparator는 이 CSV를 source-node station으로 collapse하거나 인접
+endpoint를 평균하지 않는다. Missing, extra, duplicate 또는 connectivity-mismatched
+element-endpoint identity는 tolerance 전에 실패한다. Production
+`ResultRecovery::NormalizeSectionResultantsToNodeStations` 계약과 그 단위 테스트는 이 외부
+reference identity 변경의 범위가 아니다.
+
+## 행별 판정
+
+각 대응 행에 대해 다음 absolute error를 계산한다.
+
+\[
+E_i = |F_i-R_i|
+\]
+
+Family scale이 양수이면 near-zero band는 다음과 같다.
+
+\[
+Z_g = 0.01S_g
+\]
+
+행은 다음 두 분기 중 정확히 하나로 판정한다.
+
+1. \(|R_i| \le Z_g\)이면 near-zero 행이다. 이 행은 \(E_i \le Z_g\)일 때 통과한다.
+2. \(|R_i| > Z_g\)이면 일반 상대오차 행이다. 이 행은
+   \(E_i/|R_i| \le 0.05\)일 때 통과한다.
+
+따라서 absolute error는 모든 행에 적용되는 별도 acceptance gate가 아니다. Near-zero
+분기는 0 또는 0에 가까운 reference 행에서 정의되지 않거나 과도하게 증폭되는 raw relative
+error를 대체한다.
+
+### Zero-scale family
+
+\(S_g=0\)이면 해당 family의 모든 reference 값이 정확히 0이다. 사용자가 독립 absolute
+floor를 제외했으므로 이 경우에는 다음 fail-closed 규칙을 사용한다.
+
+- 모든 FESA 값도 정확히 0이면 모든 행과 family relative RMS가 통과한다.
+- 하나라도 0이 아닌 FESA 값이 있으면 해당 행과 family가 실패한다.
+
+Comparator는 이 경우 infinity 또는 NaN을 report에 기록하지 않고
+`zero-reference-scale-nonzero-error` 진단을 기록한다.
+
+## Scale-relative RMS
+
+Family 전체의 평균적인 오차 수준은 scale-relative RMS로 판정한다.
+
+\[
+\operatorname{relative\_rms}_g =
+\frac{\sqrt{\frac{1}{N_g}\sum_{i \in g} E_i^2}}{S_g}
+\]
+
+\[
+\operatorname{relative\_rms}_g \le 0.01
+\]
+
+이 값은 raw row-relative error의 RMS가 아니다. Error RMS를 같은 family의 reference 최대
+scale로 정규화한 무차원 값이다. 행별 gate가 국부적인 큰 오차를 검출하고, relative RMS
+gate가 family 전체의 평균 오차를 검출한다. 두 gate를 모두 통과해야 한다.
+
+## 판정 순서
+
+1. 필요한 artifact와 schema를 확인한다.
+2. Source identity 및 component를 결정적으로 대응시킨다.
+3. Missing, extra, duplicate, identity-mismatched 또는 nonfinite required row를 tolerance 전에
+   실패시킨다.
+4. Comparison family와 reference-only scale을 계산한다.
+5. 모든 행에 near-zero 또는 일반 상대오차 판정을 적용한다.
+6. Family scale-relative RMS를 계산한다.
+7. Blocking family는 모든 행과 RMS가 통과해야 reference gate를 통과한다.
+8. Warning-only family의 동일한 실패는 deterministic warning을 생성하지만 전체 reference
+   verdict를 실패시키지 않는다.
+
+Tolerance 정책은 feature별 blocking/warning-only quantity 선택을 변경하지 않는다.
+
+## Report Contract
+
+Comparator report는 최소한 다음 항목을 보존한다.
+
+- family identity와 component 목록
+- reference scale와 near-zero band
+- 전체 행 수와 near-zero 행 수
+- 각 행의 원본 FESA/reference 값, absolute error, 적용된 분기 및 판정
+- 일반 상대오차 행의 relative error
+- family scale-relative RMS와 판정
+- maximum absolute error와 worst row를 진단 정보로 기록하되 독립 gate로 사용하지 않음
+- blocking 실패와 warning-only exceedance의 구분
+
+## 현재 B33/MITC4 Evidence
+
+MSVC Debug build와 214개 CTest를 통과한 현재 comparator가 다음 ledger를 생성했다.
+Reference artifact는 수정하지 않았으며 B33은 regenerated 20-row element-endpoint identity를
+직접 비교한다.
+
+- `.harness/build/reference/cantilever-beam-b33/comparison.json`
+- `.harness/build/reference/mitc4-shell-s4-comparison/comparison.json`
+
+| case | row gate | maximum family relative RMS | RMS gate | verdict |
+| --- | --- | ---: | --- | --- |
+| B33 direct endpoint | 모든 212 blocking 행 통과 | 약 `1.17e-8` | 통과 | 통과 |
+| MITC4 U | 모든 blocking 행 통과 | 약 `1.03e-3` | 통과 | 통과 |
+| MITC4 UR | 모든 warning-only 행 통과 | 약 `2.98e-3` | 통과 | warning 없음 |
+
+개별 component scale을 사용하면 MITC4 `U1/U2`의 최대 reference가 약 `3.12e-23`이어서
+near-zero 행 판정과 relative RMS가 실패한다. Translation family의 `U1/U2/U3`가 공통
+scale을 사용하면 물리적으로 zero-like인 in-plane residue를 실제 translational response와
+같은 차원에서 판정할 수 있다.
+
+## 검토한 대안
+
+### Component별 scale
+
+규칙은 단순하지만 component 전체가 zero-like이면 수치 residue 자체가 scale이 된다. 현재
+MITC4 `U1/U2`가 실패하므로 채택하지 않는다.
+
+### Reference RMS를 분모로 한 relative L2 norm
+
+`sqrt(sum(error^2)/sum(reference^2))`는 일반적인 상대 norm이지만 zero-dominant component나
+family에서 분모가 불안정하거나 정의되지 않는다. Near-zero 행 규칙과 동일한 scale 의미를
+공유하지 않으므로 공통 정책으로 채택하지 않는다.
+
+### Near-zero 보정 row-relative error의 RMS
+
+각 행의 분모를 `max(abs(reference), near_zero_band)`로 바꿔 RMS를 계산할 수 있다. 그러나
+행별 near-zero 판정을 다시 집계해 같은 오차를 중복 평가하고 해석이 복잡해지므로 채택하지
+않는다.
+
+## 구현 및 계약 정렬 범위
+
+이 문서의 승인은 목표 tolerance 정책을 고정하지만 기존 comparator와 기능 문서를 즉시
+변경하지 않는다. 후속 implementation plan은 TDD로 다음을 함께 정렬해야 한다.
+
+- 공통 comparison policy와 metric 구현
+- B33 및 MITC4 comparator의 공통 정책 사용
+- positive, boundary, zero-scale, near-zero, nonfinite 및 row-identity 테스트
+- B33와 MITC4 requirements, numerical review, reference model, I/O 및 reference comparison
+  문서의 tolerance 부분
+- `AGENTS.md`와 `docs/ADR.md`의 기존 B33/MITC4 tolerance 결정을 새 공통 정책으로 대체
+
+기존 reference input과 CSV는 생성, rename, rewrite 또는 보정하지 않는다. Production solver
+formulation, HDF5 output schema, component mapping, blocking/warning-only 분류 및 physics gate도
+이 tolerance 변경으로 수정하지 않는다.

include/fesa/analysis/analysis.h

@@ -0,0 +1,30 @@
+#ifndef FESA_ANALYSIS_ANALYSIS_H_
+#define FESA_ANALYSIS_ANALYSIS_H_
+
+#include <filesystem>
+
+#include "fesa/core/status.h"
+
+namespace fesa {
+
+/// @brief Carries input and authoritative output paths for one analysis run.
+struct AnalysisRequest {
+  std::filesystem::path input_path;
+  std::filesystem::path output_path;
+};
+
+/// @brief Defines the minimal execution contract shared by analysis procedures.
+class Analysis {
+ public:
+  virtual ~Analysis() = default;
+
+  /// @brief Executes one procedure for the supplied input and output paths.
+  /// @param request Input and authoritative output paths for this run.
+  /// @return The concrete procedure result without changing its failure
+  /// category.
+  virtual Status Run(const AnalysisRequest& request) = 0;
+};
+
+}  // namespace fesa
+
+#endif  // FESA_ANALYSIS_ANALYSIS_H_

include/fesa/analysis/analysis_model.h

@@ -0,0 +1,63 @@
+#ifndef FESA_ANALYSIS_ANALYSIS_MODEL_H_
+#define FESA_ANALYSIS_ANALYSIS_MODEL_H_
+
+#include <vector>
+
+#include "fesa/model/domain.h"
+
+namespace fesa {
+
+/// @brief Provides the active-step view into a non-owned Domain.
+/// @note The referenced Domain must outlive this object and retains all
+/// semantic ownership.
+class AnalysisModel {
+ public:
+  /// @brief Creates the sole active-step view for a valid Domain.
+  /// @param domain Domain that remains alive for the returned view's lifetime.
+  /// @return A stable view or an input-cardinality failure.
+  static Result<AnalysisModel> Create(const Domain& domain);
+
+  /// @brief Returns the non-owned Domain backing this view.
+  const Domain& GetDomain() const noexcept;
+
+  /// @brief Returns the sole active static step.
+  const StepDefinition& Step() const noexcept;
+
+  /// @brief Returns active element-definition indices in stable Domain order.
+  const std::vector<EntityIndex>& ActiveElements() const noexcept;
+
+  /// @brief Returns active B33 indices in their concrete compatibility view.
+  const std::vector<EntityIndex>& ActiveBeamElements() const noexcept;
+
+  /// @brief Returns reachable material indices in stable internal order.
+  const std::vector<EntityIndex>& ActiveMaterials() const noexcept;
+
+  /// @brief Returns reachable property indices in stable internal order.
+  const std::vector<EntityIndex>& ActiveProperties() const noexcept;
+
+  /// @brief Returns reachable beam-section indices in stable internal order.
+  const std::vector<EntityIndex>& ActiveSections() const noexcept;
+
+  /// @brief Returns boundary-condition indices in source order.
+  const std::vector<EntityIndex>& ActiveBoundaryConditions() const noexcept;
+
+  /// @brief Returns concentrated-load indices in source order.
+  const std::vector<EntityIndex>& ActiveLoads() const noexcept;
+
+ private:
+  /// @brief Builds stable indices without copying the referenced Domain.
+  explicit AnalysisModel(const Domain& domain);
+
+  const Domain* domain_;
+  std::vector<EntityIndex> active_elements_;
+  std::vector<EntityIndex> active_beam_elements_;
+  std::vector<EntityIndex> active_materials_;
+  std::vector<EntityIndex> active_properties_;
+  std::vector<EntityIndex> active_sections_;
+  std::vector<EntityIndex> active_boundary_conditions_;
+  std::vector<EntityIndex> active_loads_;
+};
+
+}  // namespace fesa
+
+#endif  // FESA_ANALYSIS_ANALYSIS_MODEL_H_

include/fesa/analysis/analysis_model.hpp

@@ -1,34 +0,0 @@
-#pragma once
-
-#include "fesa/model/domain.hpp"
-
-#include <vector>
-
-namespace fesa {
-
-// Provides the sole active-step view while the referenced Domain retains all
-// semantic ownership and must outlive this object.
-class AnalysisModel {
-public:
-    static Result<AnalysisModel> create(const Domain& domain);
-
-    const Domain& domain() const noexcept;
-    const StaticStepDefinition& step() const noexcept;
-    const std::vector<EntityIndex>& activeElements() const noexcept;
-    const std::vector<EntityIndex>& activeMaterials() const noexcept;
-    const std::vector<EntityIndex>& activeSections() const noexcept;
-    const std::vector<EntityIndex>& activeBoundaryConditions() const noexcept;
-    const std::vector<EntityIndex>& activeLoads() const noexcept;
-
-private:
-    explicit AnalysisModel(const Domain& domain);
-
-    const Domain* domain_;
-    std::vector<EntityIndex> activeElements_;
-    std::vector<EntityIndex> activeMaterials_;
-    std::vector<EntityIndex> activeSections_;
-    std::vector<EntityIndex> activeBoundaryConditions_;
-    std::vector<EntityIndex> activeLoads_;
-};
-
-} // namespace fesa

include/fesa/analysis/analysis_state.h

@@ -0,0 +1,102 @@
+#ifndef FESA_ANALYSIS_ANALYSIS_STATE_H_
+#define FESA_ANALYSIS_ANALYSIS_STATE_H_
+
+#include <array>
+#include <cstddef>
+#include <vector>
+
+#include "fesa/core/status.h"
+#include "fesa/fem/dof_manager.h"
+#include "fesa/math/vector.h"
+#include "fesa/results/result_records.h"
+
+namespace fesa {
+
+/// @brief Owns mutable quantities required by the V0 linear-static frame.
+class AnalysisState {
+ public:
+  /// @brief Allocates zeroed full-DOF vectors for a DOF manager.
+  /// @param dofs Owner of the full-DOF dimension used by every state vector.
+  /// @param identity Stable step and frame identity for this state.
+  static AnalysisState Create(const DofManager& dofs,
+                              StepFrameIdentity identity);
+
+  /// @brief Returns mutable full-space displacement.
+  Vector& Displacement() noexcept;
+  /// @brief Returns full-space displacement.
+  const Vector& Displacement() const noexcept;
+  /// @brief Returns mutable full-space external force.
+  Vector& ExternalForce() noexcept;
+  /// @brief Returns full-space external force.
+  const Vector& ExternalForce() const noexcept;
+  /// @brief Returns mutable full-space internal force.
+  Vector& InternalForce() noexcept;
+  /// @brief Returns full-space internal force.
+  const Vector& InternalForce() const noexcept;
+  /// @brief Returns mutable full residual K*d-F.
+  Vector& Residual() noexcept;
+  /// @brief Returns full residual K*d-F.
+  const Vector& Residual() const noexcept;
+  /// @brief Returns mutable full-index reaction and free residual evidence.
+  Vector& Reaction() noexcept;
+  /// @brief Returns full-index reaction and free residual evidence.
+  const Vector& Reaction() const noexcept;
+  /// @brief Returns the stable step and frame identity.
+  const StepFrameIdentity& Identity() const noexcept;
+  /// @brief Returns mutable beam endpoint result rows.
+  std::vector<EndpointResultRow>& EndpointResults() noexcept;
+  /// @brief Returns beam endpoint result rows.
+  const std::vector<EndpointResultRow>& EndpointResults() const noexcept;
+  /// @brief Returns mutable beam Gauss result rows.
+  std::vector<GaussResultRow>& GaussResults() noexcept;
+  /// @brief Returns beam Gauss result rows.
+  const std::vector<GaussResultRow>& GaussResults() const noexcept;
+  /// @brief Returns mutable beam axial-stress rows.
+  std::vector<StressS11Row>& StressResults() noexcept;
+  /// @brief Returns beam axial-stress rows.
+  const std::vector<StressS11Row>& StressResults() const noexcept;
+
+  /// @brief Validates and atomically replaces all shell recovery evidence.
+  /// @param expected_element_order Unique shell indices in stable order.
+  /// @param candidate Complete shell rows, energy, and equilibrium evidence.
+  /// @return Success only after the complete candidate is validated and
+  /// committed; failure preserves the prior shell state.
+  Status CommitShellResults(
+      const std::vector<EntityIndex>& expected_element_order,
+      ShellStateCandidate candidate);
+
+  /// @brief Returns shell rows in stable element and location order.
+  const std::vector<ShellResultRow>& ShellResults() const noexcept;
+  /// @brief Returns physical shell strain energy without drilling energy.
+  double PhysicalStrainEnergy() const noexcept;
+  /// @brief Returns global force and moment equilibrium components.
+  const std::array<double, 6>& Equilibrium() const noexcept;
+  /// @brief Returns normalized shell verification metrics.
+  const std::array<double, 3>& VerificationMetrics() const noexcept;
+
+ private:
+  /// @brief Allocates state storage for one stable full-DOF dimension.
+  AnalysisState(std::size_t full_dof_count, StepFrameIdentity identity);
+
+  StepFrameIdentity identity_;
+  Vector displacement_;
+  Vector external_force_;
+  Vector internal_force_;
+  Vector residual_;
+  // Reactions retain full-index space so free residual components remain
+  // visible.
+  Vector reaction_;
+  // Recovery appends rows in stable element/location order.
+  std::vector<EndpointResultRow> endpoint_results_;
+  std::vector<GaussResultRow> gauss_results_;
+  std::vector<StressS11Row> stress_results_;
+  // Shell recovery is replaced only through validated candidate commit.
+  std::vector<ShellResultRow> shell_results_;
+  double physical_strain_energy_{0.0};
+  std::array<double, 6> equilibrium_{};
+  std::array<double, 3> verification_metrics_{};
+};
+
+}  // namespace fesa
+
+#endif  // FESA_ANALYSIS_ANALYSIS_STATE_H_