modify documents

2026-05-08 17:03:40 +09:00
parent 73f955a8ce
commit a4dcfbdedc
23 changed files with 181 additions and 464 deletions
@@ -8,7 +8,7 @@ nickname_candidates = ["Evaluation Lead", "Skeptical QA", "Quality Analyst"]
 developer_instructions = """
 You are responsible for independent quality evaluation.
-Always read PLAN.md and PROGRESS.md before working. For implementation contract review, also read docs/V1IMPLEMENTATIONPLAN.md and the relevant contract under docs/Sprints/. For Sprint 0 review, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 1 scaffold review, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 path planning review, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 domain records and metadata review, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 MinerU adapter review, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Obsidian Markdown normalization and asset link review, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality checks and report generation review, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, and Python API review, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics and setup documentation review, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation and v1 release gate review, read docs/Sprints/SPRINT9CONTRACT.md. Treat samples/ as local fixture context only; never commit sample files unless the user explicitly requests it.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For implementation contract review, also read docs/V1IMPLEMENTATIONPLAN.md and the relevant contract under docs/Sprints/. For Sprint 0 review, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 1 scaffold review, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 path planning review, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 domain records and metadata review, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 MinerU adapter review, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Obsidian Markdown normalization and asset link review, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality checks and report generation review, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, and Python API review, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics and setup documentation review, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation and v1 release gate review, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 pre-conversion PDF chunking review, read docs/Sprints/SPRINT10CONTRACT.md. Treat samples/ as local fixture context only; never commit sample files unless the user explicitly requests it.
 Before implementation, review proposed sprint contracts from harness-planner-agent or feature-generator-agent. Require concrete done criteria, explicit non-goals, verification steps, and hard failure thresholds before work starts.
@@ -8,7 +8,7 @@ nickname_candidates = ["Feature Builder", "Sprint Builder", "Implementation Driv
 developer_instructions = """
 You are the generator in this project's long-running development harness.
-Only implement code when the user has explicitly requested implementation and a sprint contract exists. Always read PLAN.md, PROGRESS.md, AGENTS.md, PRD.md, ARCHITECTURE.md, docs/V1IMPLEMENTATIONPLAN.md, and the relevant contract under docs/Sprints/ before editing. For Sprint 1 scaffold implementation, read docs/Sprints/SPRINT1CONTRACT.md before creating pyproject.toml, src/, or tests/. For Sprint 2 path planning implementation, read docs/Sprints/SPRINT2CONTRACT.md before creating paths.py, conversion.py, CLI path hooks, or path planning tests. For Sprint 3 domain records and metadata implementation, read docs/Sprints/SPRINT3CONTRACT.md before creating ir.py, metadata.py, report.py handoff types, or metadata tests. For Sprint 4 MinerU adapter implementation, read docs/Sprints/SPRINT4CONTRACT.md before creating mineru_adapter.py, doctor.py availability hooks, or adapter tests. For Sprint 5 Obsidian Markdown normalization implementation, read docs/Sprints/SPRINT5CONTRACT.md before creating markdown.py, quality.py asset-link helpers, or normalization tests. For Sprint 6 quality and report implementation, read docs/Sprints/SPRINT6CONTRACT.md before creating quality.py, report.py, metadata summary helpers, or quality/report tests. For Sprint 7 conversion orchestration, CLI, and Python API implementation, read docs/Sprints/SPRINT7CONTRACT.md before creating conversion.py, changing cli.py, exporting convert_pdf, writing final outputs, or adding conversion/CLI tests. For Sprint 8 doctor and setup documentation implementation, read docs/Sprints/SPRINT8CONTRACT.md before creating doctor.py, changing cli.py doctor behavior, updating README setup docs, adding setup scripts, or adding doctor/CLI tests. For Sprint 9 local fixture evaluation and v1 release gate implementation, read docs/Sprints/SPRINT9CONTRACT.md before creating integration tests, optional MinerU fixture harnesses, fixture manifests, release checklists, or release-gate documentation.
+Only implement code when the user has explicitly requested implementation and a sprint contract exists. Always read PLAN.md, PROGRESS.md, AGENTS.md, PRD.md, ARCHITECTURE.md, docs/V1IMPLEMENTATIONPLAN.md, and the relevant contract under docs/Sprints/ before editing. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For Sprint 1 scaffold implementation, read docs/Sprints/SPRINT1CONTRACT.md before creating pyproject.toml, src/, or tests/. For Sprint 2 path planning implementation, read docs/Sprints/SPRINT2CONTRACT.md before creating paths.py, conversion.py, CLI path hooks, or path planning tests. For Sprint 3 domain records and metadata implementation, read docs/Sprints/SPRINT3CONTRACT.md before creating ir.py, metadata.py, report.py handoff types, or metadata tests. For Sprint 4 MinerU adapter implementation, read docs/Sprints/SPRINT4CONTRACT.md before creating mineru_adapter.py, doctor.py availability hooks, or adapter tests. For Sprint 5 Obsidian Markdown normalization implementation, read docs/Sprints/SPRINT5CONTRACT.md before creating markdown.py, quality.py asset-link helpers, or normalization tests. For Sprint 6 quality and report implementation, read docs/Sprints/SPRINT6CONTRACT.md before creating quality.py, report.py, metadata summary helpers, or quality/report tests. For Sprint 7 conversion orchestration, CLI, and Python API implementation, read docs/Sprints/SPRINT7CONTRACT.md before creating conversion.py, changing cli.py, exporting convert_pdf, writing final outputs, or adding conversion/CLI tests. For Sprint 8 doctor and setup documentation implementation, read docs/Sprints/SPRINT8CONTRACT.md before creating doctor.py, changing cli.py doctor behavior, updating README setup docs, adding setup scripts, or adding doctor/CLI tests. For Sprint 9 local fixture evaluation and v1 release gate implementation, read docs/Sprints/SPRINT9CONTRACT.md before creating integration tests, optional MinerU fixture harnesses, fixture manifests, release checklists, or release-gate documentation. For Sprint 10 pre-conversion PDF chunking implementation, read docs/Sprints/SPRINT10CONTRACT.md before changing pdf_splitter.py, conversion.py chunk orchestration, CLI chunk options, chunk metadata/report behavior, or chunk tests.
 Work one contract at a time. Keep the change surgical, avoid speculative flexibility, and use project-owned boundaries from ARCHITECTURE.md. If the contract is ambiguous, ask the parent agent to negotiate clarification with evaluation-agent before writing code.
@@ -8,7 +8,7 @@ nickname_candidates = ["Harness Planner", "Scope Planner", "Contract Planner"]
 developer_instructions = """
 You are the planner in this project's long-running development harness.
-Always read PLAN.md and PROGRESS.md before working. For substantial work, read PRD.md, ARCHITECTURE.md, docs/V1IMPLEMENTATIONPLAN.md, and the active contract under docs/Sprints/ before expanding the user's request into product context, deliverables, non-goals, dependencies, risks, and a small sequence of implementation chunks. For Sprint 1 planning or refinement, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 path planning refinement, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 domain records and metadata refinement, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 MinerU adapter refinement, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Markdown normalization refinement, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality and report refinement, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, and Python API refinement, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics and setup documentation refinement, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation and v1 release gate refinement, read docs/Sprints/SPRINT9CONTRACT.md.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For substantial work, read PRD.md, ARCHITECTURE.md, docs/V1IMPLEMENTATIONPLAN.md, and the active contract under docs/Sprints/ before expanding the user's request into product context, deliverables, non-goals, dependencies, risks, and a small sequence of implementation chunks. For Sprint 1 planning or refinement, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 path planning refinement, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 domain records and metadata refinement, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 MinerU adapter refinement, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Markdown normalization refinement, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality and report refinement, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, and Python API refinement, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics and setup documentation refinement, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation and v1 release gate refinement, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 pre-conversion PDF chunking refinement, read docs/Sprints/SPRINT10CONTRACT.md.
 Stay focused on what should be built and how success will be judged. Avoid over-specifying low-level implementation details before the feature-generator has inspected the real code. Use domain agents for specialized questions: mineru-integration-agent, obsidian-markdown-agent, metadata-agent, evaluation-agent, local-setup-agent, license-privacy-agent, and requirements-guard-agent.
@@ -8,7 +8,7 @@ nickname_candidates = ["License Guard", "Privacy Reviewer", "Policy Checker"]
 developer_instructions = """
 You are responsible for license and privacy review.
-Always read PLAN.md and PROGRESS.md before working. For v1 license/privacy planning, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 license and privacy verification, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 8 setup documentation, setup helper, model/cache, and strict-local privacy review, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation privacy, no-sample-commit checks, and release gate review, read docs/Sprints/SPRINT9CONTRACT.md. Treat local-only processing as a hard requirement: no uploaded PDFs, page images, extracted text, or model intermediates to remote services.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For v1 license/privacy planning, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 license and privacy verification, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 8 setup documentation, setup helper, model/cache, and strict-local privacy review, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation privacy, no-sample-commit checks, and release gate review, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunking privacy review, read docs/Sprints/SPRINT10CONTRACT.md. Treat local-only processing as a hard requirement: no uploaded PDFs, page images, extracted text, or model intermediates to remote services.
 Review MinerU, model weights, transitive packages, and generated assets for licenses before redistribution. Distinguish personal/research use from redistribution. Record source URLs, license names, and unresolved obligations.
@@ -8,7 +8,7 @@ nickname_candidates = ["Setup Lead", "CUDA Checker", "Environment Guard"]
 developer_instructions = """
 You are responsible for local setup and environment planning.
-Always read PLAN.md and PROGRESS.md before working. For v1 setup planning, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 environment verification, read docs/Sprints/SPRINT0CONTRACT.md; for Sprint 1 scaffold or uv bootstrap planning, read docs/Sprints/SPRINT1CONTRACT.md; for Sprint 4 MinerU availability/version adapter checks, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 6 local math renderability tool-unavailable behavior, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 8 doctor diagnostics, setup documentation, GPU/CUDA/PyTorch checks, uv checks, and model/cache checks, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 optional local MinerU/GPU fixture evaluation gating and doctor preflight handling, read docs/Sprints/SPRINT9CONTRACT.md. Target Windows PowerShell, Python 3.12, uv, NVIDIA GPU execution, and GTX 1070 Ti 8GB constraints.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For v1 setup planning, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 environment verification, read docs/Sprints/SPRINT0CONTRACT.md; for Sprint 1 scaffold or uv bootstrap planning, read docs/Sprints/SPRINT1CONTRACT.md; for Sprint 4 MinerU availability/version adapter checks, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 6 local math renderability tool-unavailable behavior, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 8 doctor diagnostics, setup documentation, GPU/CUDA/PyTorch checks, uv checks, and model/cache checks, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 optional local MinerU/GPU fixture evaluation gating and doctor preflight handling, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunking setup/runtime review, read docs/Sprints/SPRINT10CONTRACT.md. Target Windows PowerShell, Python 3.12, uv, NVIDIA GPU execution, and GTX 1070 Ti 8GB constraints.
 Prefer checks that clearly diagnose missing Python, uv, CUDA, GPU visibility, model cache paths, and MinerU CLI availability. If GPU execution is impossible, require a clear CPU fallback or error message according to project decisions.
@@ -8,7 +8,7 @@ nickname_candidates = ["Metadata Lead", "Report Designer", "Provenance Guard"]
 developer_instructions = """
 You are responsible for metadata and reporting.
-Always read PLAN.md, PROGRESS.md, PRD.md, ARCHITECTURE.md, and docs/V1IMPLEMENTATIONPLAN.md before working. When a metadata/reporting sprint contract exists, read the relevant contract under docs/Sprints/ as well. For Sprint 3 domain records, metadata, and warning model work, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 5 Markdown normalization work that changes warning codes, asset warnings, or table fallback warning semantics, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality checks, metadata summary extensions, and report rendering work, read docs/Sprints/SPRINT6CONTRACT.md before changing quality.py, report.py, metadata.py, or report tests. For Sprint 7 conversion orchestration work that writes metadata JSON, report Markdown, output paths, or asset provenance, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 9 fixture evaluation, metadata assertions, report quality gates, and release checklist work, read docs/Sprints/SPRINT9CONTRACT.md. Maintain provenance for source PDF path, page index, bbox when available, block type, engine, confidence, warnings, asset paths, and output locations.
+Always read PLAN.md, PROGRESS.md, PRD.md, ARCHITECTURE.md, and docs/V1IMPLEMENTATIONPLAN.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. When a metadata/reporting sprint contract exists, read the relevant contract under docs/Sprints/ as well. For Sprint 3 domain records, metadata, and warning model work, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 5 Markdown normalization work that changes warning codes, asset warnings, or table fallback warning semantics, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality checks, metadata summary extensions, and report rendering work, read docs/Sprints/SPRINT6CONTRACT.md before changing quality.py, report.py, metadata.py, or report tests. For Sprint 7 conversion orchestration work that writes metadata JSON, report Markdown, output paths, or asset provenance, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 9 fixture evaluation, metadata assertions, report quality gates, and release checklist work, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunk provenance and report context work, read docs/Sprints/SPRINT10CONTRACT.md. Maintain provenance for source PDF path, page index, bbox when available, block type, engine, confidence, warnings, asset paths, output locations, and chunk page ranges when chunking is active.
 Every conversion design must include both machine-readable JSON metadata and a human-readable <stem>.report.md. Reports should be derived from metadata and local checks, not manually duplicated state.
@@ -8,7 +8,7 @@ nickname_candidates = ["MinerU Integrator", "Adapter Planner", "CLI Guard"]
 developer_instructions = """
 You are responsible for the MinerU integration design.
-Always read PLAN.md, PROGRESS.md, ARCHITECTURE.md, PRD.md, and docs/V1IMPLEMENTATIONPLAN.md before proposing integration work. For Sprint 0 output layout or CLI verification, also read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 4 mocked MinerU adapter contract work, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 7 conversion orchestration work that calls the adapter, handles raw output, or preserves no-fallback behavior, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor work that checks MinerU availability, version, local execution, or setup documentation, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 optional local MinerU fixture evaluation, output evidence, and no-fallback release-gate checks, read docs/Sprints/SPRINT9CONTRACT.md. Treat MinerU 3.1.0 as the only engine and direct local CLI execution as the only v1 execution mode.
+Always read PLAN.md, PROGRESS.md, ARCHITECTURE.md, PRD.md, and docs/V1IMPLEMENTATIONPLAN.md before proposing integration work. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For Sprint 0 output layout or CLI verification, also read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 4 mocked MinerU adapter contract work, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 7 conversion orchestration work that calls the adapter, handles raw output, or preserves no-fallback behavior, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor work that checks MinerU availability, version, local execution, or setup documentation, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 optional local MinerU fixture evaluation, output evidence, and no-fallback release-gate checks, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunk PDF staging and pre-conversion orchestration, read docs/Sprints/SPRINT10CONTRACT.md. Treat MinerU 3.1.0 as the only engine and direct local CLI execution as the only v1 execution mode.
 MinerU 3.1.0 may start a temporary local mineru-api process internally when the mineru CLI runs without --api-url. This is allowed. Passing --api-url, using remote APIs, router mode, HTTP client backends, or remote OpenAI-compatible backends is prohibited.
@@ -8,7 +8,7 @@ nickname_candidates = ["Markdown Reviewer", "Math Normalizer", "Obsidian Lead"]
 developer_instructions = """
 You are responsible for Obsidian-friendly Markdown output.
-Always read PLAN.md and PROGRESS.md before working. Read PRD.md, ARCHITECTURE.md, and docs/V1IMPLEMENTATIONPLAN.md when changing output behavior. When a Markdown/output sprint contract exists, read the relevant contract under docs/Sprints/ as well. For Sprint 5 Obsidian Markdown normalization and asset link work, read docs/Sprints/SPRINT5CONTRACT.md before changing markdown.py, quality.py asset-link helpers, or normalization tests. For Sprint 6 math renderability quality checks and render-warning policy, read docs/Sprints/SPRINT6CONTRACT.md before changing quality.py or report-facing math warning tests. For Sprint 7 conversion orchestration work that writes final Markdown, copies assets, or links assets from output Markdown, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 9 fixture evaluation of Obsidian Markdown, math delimiters, table fallback behavior, asset links, and renderability warnings, read docs/Sprints/SPRINT9CONTRACT.md. Preserve the fixed delimiter policy: inline math uses $...$ and display math uses $$...$$.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. Read PRD.md, ARCHITECTURE.md, and docs/V1IMPLEMENTATIONPLAN.md when changing output behavior. When a Markdown/output sprint contract exists, read the relevant contract under docs/Sprints/ as well. For Sprint 5 Obsidian Markdown normalization and asset link work, read docs/Sprints/SPRINT5CONTRACT.md before changing markdown.py, quality.py asset-link helpers, or normalization tests. For Sprint 6 math renderability quality checks and render-warning policy, read docs/Sprints/SPRINT6CONTRACT.md before changing quality.py or report-facing math warning tests. For Sprint 7 conversion orchestration work that writes final Markdown, copies assets, or links assets from output Markdown, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 9 fixture evaluation of Obsidian Markdown, math delimiters, table fallback behavior, asset links, and renderability warnings, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunk output naming and no-merge behavior, read docs/Sprints/SPRINT10CONTRACT.md. Preserve the fixed delimiter policy: inline math uses $...$ and display math uses $$...$$.
 Focus on Markdown normalization, asset path stability, table fallback behavior, readable warnings, and renderability checks. Do not promise perfect LaTeX reconstruction; require metadata warnings for low-confidence or non-renderable math.
@@ -8,7 +8,7 @@ nickname_candidates = ["Requirements Guard", "Doc Auditor", "Consistency Lead"]
 developer_instructions = """
 You are the requirements guard for this repository.
-Always read PLAN.md and PROGRESS.md before working. Then read only the project documents needed for the requested check, including docs/V1IMPLEMENTATIONPLAN.md and relevant contracts under docs/Sprints/ when implementation sequencing or sprint contracts are in scope. For Sprint 1 consistency checks, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 consistency checks, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 consistency checks, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 consistency checks, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Markdown normalization and asset link consistency checks, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality, metadata summary, and report consistency checks, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, Python API, and output-writing consistency checks, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics, setup documentation, strict-local wording, and setup-helper consistency checks, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation, v1 release gate, optional-check gating, and no-sample-commit consistency checks, read docs/Sprints/SPRINT9CONTRACT.md. Prioritize contradictions, outdated decisions, missing acceptance criteria, and text that weakens local-only or MinerU-only constraints.
+Always read PLAN.md and PROGRESS.md before working. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. Then read only the project documents needed for the requested check, including docs/V1IMPLEMENTATIONPLAN.md and relevant contracts under docs/Sprints/ when implementation sequencing or sprint contracts are in scope. For Sprint 1 consistency checks, read docs/Sprints/SPRINT1CONTRACT.md. For Sprint 2 consistency checks, read docs/Sprints/SPRINT2CONTRACT.md. For Sprint 3 consistency checks, read docs/Sprints/SPRINT3CONTRACT.md. For Sprint 4 consistency checks, read docs/Sprints/SPRINT4CONTRACT.md. For Sprint 5 Markdown normalization and asset link consistency checks, read docs/Sprints/SPRINT5CONTRACT.md. For Sprint 6 quality, metadata summary, and report consistency checks, read docs/Sprints/SPRINT6CONTRACT.md. For Sprint 7 conversion orchestration, CLI, Python API, and output-writing consistency checks, read docs/Sprints/SPRINT7CONTRACT.md. For Sprint 8 doctor diagnostics, setup documentation, strict-local wording, and setup-helper consistency checks, read docs/Sprints/SPRINT8CONTRACT.md. For Sprint 9 local fixture evaluation, v1 release gate, optional-check gating, and no-sample-commit consistency checks, read docs/Sprints/SPRINT9CONTRACT.md. For Sprint 10 chunking, CLI/API chunk mode, and chunk provenance consistency checks, read docs/Sprints/SPRINT10CONTRACT.md. Prioritize contradictions, outdated decisions, missing acceptance criteria, and text that weakens local-only or MinerU-only constraints.
 Fixed decisions: Python 3.12, uv, direct local MinerU 3.1.0 CLI execution, CLI-internal temporary local mineru-api allowed, no --api-url or remote API paths, no router mode, no HTTP client backend, no runtime engine selection, Obsidian Markdown output, inline math with $...$, display math with $$...$$, metadata JSON, and human-readable .report.md output.
@@ -8,7 +8,7 @@ nickname_candidates = ["Research Lead", "Source Checker", "MinerU Scout"]
 developer_instructions = """
 You are the project research agent for the local PDF-to-Markdown converter.
-Always read PLAN.md and PROGRESS.md before working. Use PROGRESS.md as the factual state. For v1 implementation research, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 source verification, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 8 setup documentation or doctor facts that may have changed, read docs/Sprints/SPRINT8CONTRACT.md and verify volatile install/model/cache claims against official sources before docs are edited. Prefer official MinerU documentation, MinerU GitHub, primary papers, and official Codex/OpenAI documentation when researching workflow structure. Cite URLs and access dates in any research notes.
+Always read PLAN.md and PROGRESS.md before working. Use PROGRESS.md as the factual current state. Read docs/WORKARCHIVE.md when prior completed sprint context, historical verification, runtime setup evidence, or sample conversion evidence is needed. For v1 implementation research, read docs/V1IMPLEMENTATIONPLAN.md; for Sprint 0 source verification, read docs/Sprints/SPRINT0CONTRACT.md. For Sprint 8 setup documentation or doctor facts that may have changed, read docs/Sprints/SPRINT8CONTRACT.md and verify volatile install/model/cache claims against official sources before docs are edited. For Sprint 10 pypdf or chunking facts that may have changed, read docs/Sprints/SPRINT10CONTRACT.md and verify volatile package facts against official sources before docs are edited. Prefer official MinerU documentation, MinerU GitHub, primary papers, and official Codex/OpenAI documentation when researching workflow structure. Cite URLs and access dates in any research notes.
 Keep MinerU 3.1.0 as the only conversion engine. Do not reintroduce candidate engine comparisons. Record uncertainty explicitly and ask the parent agent for a decision when official sources conflict.
@@ -15,11 +15,12 @@ The user invoked this command with: $ARGUMENTS
 ## Workflow
 1. Read `PLAN.md`, `PROGRESS.md`, `PRD.md`, and `ARCHITECTURE.md`.
-2. Verify any MinerU CLI facts that may have changed before changing docs.
+2. Read `docs/WORKARCHIVE.md` when prior MinerU setup, verification, or sample conversion evidence is needed.
-3. Define the smallest adapter contract for command construction, working directories, outputs, stdout/stderr capture, exit handling, warnings, and provenance.
+3. Verify any MinerU CLI facts that may have changed before changing docs.
-4. Ensure failure behavior is explicit: no silent fallback and no alternate engine route.
+4. Define the smallest adapter contract for command construction, working directories, outputs, stdout/stderr capture, exit handling, warnings, and provenance.
-5. Identify mocked-output tests and optional MinerU-dependent checks.
+5. Ensure failure behavior is explicit: no silent fallback and no alternate engine route.
-6. Update `PLAN.md` only if implementation sequencing changes; update `PROGRESS.md` after the planning work.
+6. Identify mocked-output tests and optional MinerU-dependent checks.
 7. Update `PLAN.md` only if implementation sequencing changes; update `PROGRESS.md` after the planning work.
 ## Guardrails
@@ -15,11 +15,12 @@ The user invoked this command with: $ARGUMENTS
 ## Workflow
 1. Read `PLAN.md`, `PROGRESS.md`, `PRD.md`, and `ARCHITECTURE.md`.
-2. Inspect `samples/` only as local fixture context; do not stage or commit sample files.
+2. Read `docs/WORKARCHIVE.md` for prior sample conversion evidence and historical fixture coverage.
-3. Define checks for page coverage, reading order, math renderability, delimiter normalization, table handling, asset links, metadata completeness, and warning counts.
+3. Inspect `samples/` only as local fixture context; do not stage or commit sample files.
-4. Define `.json` metadata and `.report.md` expectations from the same source data.
+4. Define checks for page coverage, reading order, math renderability, delimiter normalization, table handling, asset links, metadata completeness, and warning counts.
-5. Separate fast mocked checks from optional MinerU/model/GPU-dependent checks.
+5. Define `.json` metadata and `.report.md` expectations from the same source data.
-6. Update `PROGRESS.md` with the planned coverage and remaining sample gaps.
+6. Separate fast mocked checks from optional MinerU/model/GPU-dependent checks.
 7. Update `PROGRESS.md` with the planned coverage and remaining sample gaps.
 ## Guardrails
@@ -15,10 +15,11 @@ The user invoked this command with: $ARGUMENTS
 ## Workflow
 1. Read `PLAN.md` and `PROGRESS.md`.
-2. Read the requested document scope, defaulting to `AGENTS.md`, `PRD.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md`.
+2. Read `docs/WORKARCHIVE.md` when reviewing completed-work history, prior verification, or sample conversion evidence.
-3. Check for contradictions against fixed decisions: MinerU 3.1.0 only, local-only, direct CLI execution, CLI-internal temporary local `mineru-api` allowed, no `--api-url` or remote API path, Python 3.12, uv, Obsidian Markdown, metadata JSON, and `.report.md`.
+3. Read the requested document scope, defaulting to `AGENTS.md`, `PRD.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md`.
-4. Report findings first with file and line references.
+4. Check for contradictions against fixed decisions: MinerU 3.1.0 only, local-only, direct CLI execution, CLI-internal temporary local `mineru-api` allowed, no `--api-url` or remote API path, Python 3.12, uv, Obsidian Markdown, metadata JSON, and `.report.md`.
-5. If edits are requested, make only surgical documentation changes and update `PROGRESS.md`.
+5. Report findings first with file and line references.
 6. If edits are requested, make only surgical documentation changes and update `PROGRESS.md`.
 ## Guardrails
@@ -15,11 +15,12 @@ The user invoked this command with: $ARGUMENTS
 ## Workflow
 1. Read `PLAN.md`, `PROGRESS.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md`.
-2. Use official MinerU documentation, the MinerU GitHub repository, primary papers, and official dependency documentation.
+2. Read `docs/WORKARCHIVE.md` when prior MinerU setup, verification, or sample conversion evidence is needed.
-3. Verify facts that can change: install commands, supported Python/CUDA versions, CLI flags, output formats, model download behavior, and licenses.
+3. Use official MinerU documentation, the MinerU GitHub repository, primary papers, and official dependency documentation.
-4. Record sources with URLs and access dates when updating docs.
+4. Verify facts that can change: install commands, supported Python/CUDA versions, CLI flags, output formats, model download behavior, and licenses.
-5. Keep findings scoped to MinerU 3.1.0; do not add candidate-engine comparisons.
+5. Record sources with URLs and access dates when updating docs.
-6. Update `PROGRESS.md` with what was verified and what remains uncertain.
+6. Keep findings scoped to MinerU 3.1.0; do not add candidate-engine comparisons.
 7. Update `PROGRESS.md` with what was verified and what remains uncertain.
 ## Guardrails
@@ -16,12 +16,14 @@ The user invoked this command with: $ARGUMENTS
 1. Read `PLAN.md` and `PROGRESS.md`.
 2. State the current goal, the next action, and any blocker that matters for the task.
-3. Read only the additional source documents needed for the requested work.
+3. Read `docs/WORKARCHIVE.md` when historical completed-work context, previous verification, or sample conversion evidence is needed.
-4. If subagents are useful and the user explicitly asked for delegated agent work, choose the smallest set of `.codex/agents/*.toml` roles that covers the task.
+4. Read only the additional source documents needed for the requested work.
-5. For substantial implementation work, use the harness sequence: `harness-planner-agent` drafts the plan and contract, `feature-generator-agent` implements one agreed chunk, and `evaluation-agent` reviews the contract and completed work.
+5. If subagents are useful and the user explicitly asked for delegated agent work, choose the smallest set of `.codex/agents/*.toml` roles that covers the task.
-6. Do not implement converter code unless the user explicitly requests implementation.
+6. For substantial implementation work, use the harness sequence: `harness-planner-agent` drafts the plan and contract, `feature-generator-agent` implements one agreed chunk, and `evaluation-agent` reviews the contract and completed work.
-7. After meaningful changes, update `PROGRESS.md`; update `PLAN.md` only when sequencing, decisions, ownership, or blockers change.
+7. Do not implement converter code unless the user explicitly requests implementation.
-8. Run the smallest useful verification, check git status, and commit project changes while excluding `samples/`.
+8. After meaningful changes, update `PROGRESS.md`; update `PLAN.md` only when sequencing, decisions, ownership, or blockers change.
 9. Archive completed work in `docs/WORKARCHIVE.md` when it no longer needs to stay in `PROGRESS.md`.
 10. Run the smallest useful verification, check git status, and commit project changes while excluding `samples/`.
 ## Guardrails
@@ -12,11 +12,12 @@ Use this skill to turn local sample PDFs into a small, repeatable quality plan.
 ## Workflow
 1. Read `PLAN.md` and `PROGRESS.md` first.
-2. Inspect `samples/` only enough to understand fixture categories and filenames.
+2. Read `docs/WORKARCHIVE.md` when prior fixture coverage, verification, or sample conversion evidence is needed.
-3. Map each fixture to risks: math, tables, multi-column reading order, figures/assets, Korean filenames, and metadata coverage.
+3. Inspect `samples/` only enough to understand fixture categories and filenames.
-4. Separate fast checks using mocked MinerU outputs from optional checks that require MinerU models, GPU, or long execution.
+4. Map each fixture to risks: math, tables, multi-column reading order, figures/assets, Korean filenames, and metadata coverage.
-5. Define metrics for both JSON metadata and `<stem>.report.md`.
+5. Separate fast checks using mocked MinerU outputs from optional checks that require MinerU models, GPU, or long execution.
-6. Update `PROGRESS.md` with fixture coverage and gaps.
+6. Define metrics for both JSON metadata and `<stem>.report.md`.
 7. Update `PROGRESS.md` with fixture coverage and gaps.
 ## Guardrails
@@ -12,11 +12,12 @@ Use this skill when Markdown output quality matters more than raw text extractio
 ## Workflow
 1. Read `PLAN.md` and `PROGRESS.md` first.
-2. Read `PRD.md` and `ARCHITECTURE.md` when output behavior, metadata, or reporting is affected.
+2. Read `docs/WORKARCHIVE.md` when prior Markdown output, MathJax, or sample conversion evidence is needed.
-3. Preserve project delimiter policy: inline math uses `$...$`; display math uses `$$...$$`.
+3. Read `PRD.md` and `ARCHITECTURE.md` when output behavior, metadata, or reporting is affected.
-4. Check asset links, table fallback behavior, heading/list interactions, and page boundary markers against Obsidian rendering assumptions.
+4. Preserve project delimiter policy: inline math uses `$...$`; display math uses `$$...$$`.
-5. Define warnings for low-confidence math, non-renderable LaTeX, broken asset links, table degradation, and reading-order uncertainty.
+5. Check asset links, table fallback behavior, heading/list interactions, and page boundary markers against Obsidian rendering assumptions.
-6. Ensure `.report.md` content is derived from metadata, not separate manual state.
+6. Define warnings for low-confidence math, non-renderable LaTeX, broken asset links, table degradation, and reading-order uncertainty.
 7. Ensure `.report.md` content is derived from metadata, not separate manual state.
 ## Checks
@@ -12,11 +12,12 @@ Use this skill to verify MinerU 3.1.0 facts before changing project docs or plan
 ## Workflow
 1. Read `PLAN.md` and `PROGRESS.md` first.
-2. Read `PRD.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md` when the change affects product or architecture decisions.
+2. Read `docs/WORKARCHIVE.md` when prior MinerU setup, verification, or sample conversion evidence is needed.
-3. Prefer official MinerU documentation, the MinerU GitHub repository, release notes, primary papers, and official dependency docs.
+3. Read `PRD.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md` when the change affects product or architecture decisions.
-4. Verify time-sensitive facts with web research before updating docs.
+4. Prefer official MinerU documentation, the MinerU GitHub repository, release notes, primary papers, and official dependency docs.
-5. Record source URLs and access dates in durable docs when the finding affects future implementation.
+5. Verify time-sensitive facts with web research before updating docs.
-6. Update `PROGRESS.md` with the verified fact, unresolved uncertainty, and next action.
+6. Record source URLs and access dates in durable docs when the finding affects future implementation.
 7. Update `PROGRESS.md` with the verified fact, unresolved uncertainty, and next action.
 ## Constraints
@@ -80,6 +80,7 @@ Strong success criteria let you loop independently. Weak criteria ("make it work
 - `ARCHITECTURE.md`: system layers, MinerU adapter contract, intermediate representation, metadata schema, and local-only enforcement.
 - `docs/KNOWLEDGEBASE.md`: research basis and implementation background.
 - `docs/V1IMPLEMENTATIONPLAN.md`: v1 implementation sequence, sprint contracts, verification gates, and agent ownership.
 - `docs/WORKARCHIVE.md`: archived completed work, historical sprint outcomes, setup results, verification history, and sample conversion evidence.
 - `docs/Sprints/*.md`: active and historical sprint contracts.
 - `.codex/agents/*.toml`: project-scoped custom subagent roles.
 - `.codex/commands/*.md`: reusable project prompt commands.
@@ -92,6 +93,7 @@ At the start of every task:
 - Read `PLAN.md` and `PROGRESS.md` before deciding what to do.
 - Read only the other source documents needed for the task.
 - Read `docs/WORKARCHIVE.md` when the task needs historical completed-work context, previous verification results, or sample conversion evidence.
 - Use `.codex/agents`, `.codex/commands`, and `.codex/skills` when the user explicitly asks for agent delegation, reusable workflows, or specialized project guidance.
 - State the relevant current goal, next action, and blocker if one exists.
 - If `PLAN.md` and `PROGRESS.md` conflict, trust `PROGRESS.md` for what has happened and update `PLAN.md` when making the next change.
@@ -102,6 +104,7 @@ Use `PLAN.md` and `PROGRESS.md` to coordinate work across agents.
 - Update `PLAN.md` when planned work, ownership, sequencing, open questions, or decisions change.
 - Update `PROGRESS.md` after meaningful work, verification, blockers, or next actions change.
 - Move completed work from `PROGRESS.md` to `docs/WORKARCHIVE.md` when it is no longer needed for current coordination.
 - Keep entries short and factual.
 - Do not use these files as scratchpads or long research notes.
 - Do not mark work complete until it has been verified.
@@ -4,7 +4,7 @@ This file is the shared work plan for agents. Read it before starting work, then
 ## Current Goal
-CUDA-enabled PyTorch and MinerU 3.1.0 runtime setup is complete in the project `.venv`. Sprint 10 pre-conversion PDF chunking is implemented; next work is optional real local sample validation only if requested.
+Completed work history is archived in `docs/WORKARCHIVE.md`. CUDA-enabled PyTorch and MinerU 3.1.0 runtime setup is complete in the project `.venv`; Sprint 10 pre-conversion PDF chunking is implemented; next work is optional real local sample validation only if requested.
 ## Active Constraints
@@ -34,6 +34,7 @@ CUDA-enabled PyTorch and MinerU 3.1.0 runtime setup is complete in the project `
 11. Use `evaluation-agent` as the independent contract reviewer and QA evaluator before and after each implementation chunk.
 12. Follow `docs/V1IMPLEMENTATIONPLAN.md` for the v1 implementation sprint sequence.
 13. Use `docs/Sprints/SPRINT10CONTRACT.md` for the implemented long-PDF pre-conversion chunking sprint.
 14. Use `docs/WORKARCHIVE.md` for completed sprint history, prior verification, runtime setup evidence, and sample conversion evidence.
 ## Open Questions
@@ -43,6 +44,7 @@ CUDA-enabled PyTorch and MinerU 3.1.0 runtime setup is complete in the project `
 - Use `PLAN.md` for intended work and ownership.
 - Use `PROGRESS.md` for completed work, current status, blockers, and next actions.
 - Use `docs/WORKARCHIVE.md` for archived completed work and historical handoff details.
 - MinerU default local CLI execution is the only v1 execution mode.
 - MinerU 3.1.0 may launch a temporary local `mineru-api` internally when `mineru` CLI runs without `--api-url`.
 - Strict-local mode forbids `--api-url`, remote APIs, router mode, HTTP client backends, and remote OpenAI-compatible backends.
@@ -81,7 +83,6 @@ CUDA-enabled PyTorch and MinerU 3.1.0 runtime setup is complete in the project `
 - The MinerU adapter maps CUDA device requests to local subprocess environment variables instead of adding speculative MinerU CLI flags.
 - GTX 1070 Ti local runtime uses PyTorch `2.6.0+cu126` and `torchvision 0.21.0+cu126` installed after `uv sync`, followed by `mineru[core]==3.1.0`.
 - MinerU models are downloaded with `mineru-models-download -s huggingface -m all`, and runtime model loading uses `MINERU_MODEL_SOURCE=local`.
 - Sprint 10 should use `pypdf` for local 20-page PDF chunk creation if implementation is approved.
 - Sprint 10 uses `pypdf` for local PDF page chunk planning and temporary chunk PDF writing.
 - Sprint 10 converts chunk PDFs independently and does not merge generated Markdown outputs.
 - Chunking is opt-in through `--chunk-pages`; if the option is present without a value, the CLI uses 20 pages per chunk.
@@ -1,250 +1,40 @@
 # PROGRESS.md
-This file records actual progress for agents. Read it before starting work, then update it after meaningful changes.
+This file records current progress for agents. Read it before starting work, then update it after meaningful changes. Completed historical work is archived in `docs/WORKARCHIVE.md`.
 ## Current Status
- Project direction is documented.
+- Project direction is documented in `PRD.md`, `ARCHITECTURE.md`, `AGENTS.md`, and `docs/KNOWLEDGEBASE.md`.
 - MinerU 3.1.0 is fixed as the only conversion engine.
- `PRD.md`, `ARCHITECTURE.md`, `AGENTS.md`, and `docs/KNOWLEDGEBASE.md` exist.
+- The converter currently includes path planning, project-owned records, metadata, direct local MinerU adapter boundary, Obsidian Markdown normalization, local quality checks, report rendering, conversion orchestration, `pdf2md convert`, `pdf2md doctor`, local MathJax render checking, release-gate tests, and opt-in pre-conversion PDF chunking.
 - `docs/V1IMPLEMENTATIONPLAN.md` defines the v1 implementation sequence.
 - `docs/Sprints/` contains completed sprint contracts through Sprint 10.
 - `docs/WORKARCHIVE.md` contains completed sprint history, historical verification results, runtime setup notes, and sample conversion evidence.
 - `samples/` exists locally and is untracked by git.
- Converter implementation exists through Sprint 9 path planning, project-owned records, metadata, mocked direct local MinerU adapter boundary, Obsidian Markdown normalization, local quality checks, report content rendering, conversion orchestration, public conversion API, `pdf2md convert`, `pdf2md doctor`, fast mocked integration tests, optional local MinerU fixture evaluation, and the v1 release checklist.
+- `outputs/` is ignored and contains local generated conversion outputs.
 - Default conversion now requests `cuda:0`; the MinerU adapter sets local GPU-related environment for the MinerU subprocess.
 - Project-local Codex workflow assets now live under `.codex/`.
 - `docs/V1IMPLEMENTATIONPLAN.md` now defines the v1 implementation sequence.
 - `docs/Sprints/SPRINT0CONTRACT.md` now defines the Sprint 0 contract.
 - `docs/Sprints/SPRINT1CONTRACT.md` now defines the Sprint 1 scaffold contract.
 - `docs/Sprints/SPRINT2CONTRACT.md` now defines the Sprint 2 path planning contract.
 - `docs/Sprints/SPRINT3CONTRACT.md` now defines the Sprint 3 domain records and metadata contract.
 - `docs/Sprints/SPRINT4CONTRACT.md` now defines the Sprint 4 mocked MinerU adapter contract.
 - `docs/Sprints/SPRINT5CONTRACT.md` now defines the Sprint 5 Obsidian Markdown normalization and asset link contract.
 - `docs/Sprints/SPRINT6CONTRACT.md` now defines the Sprint 6 quality checks and report generation contract.
 - `docs/Sprints/SPRINT7CONTRACT.md` now defines the Sprint 7 conversion orchestration, CLI, and Python API contract.
 - `docs/Sprints/SPRINT8CONTRACT.md` now defines the Sprint 8 doctor and setup documentation contract.
 - `docs/Sprints/SPRINT9CONTRACT.md` now defines the Sprint 9 local fixture evaluation and v1 release gate contract.
 - Relevant `.codex/agents/*.toml` files now reference the v1 plan and sprint contract paths directly.
 - Sprint 10 is implemented with opt-in pre-conversion PDF chunking, temporary chunk PDF cleanup, chunk metadata/report context, and mocked tests.
 - Sprint 0 source, environment, license, privacy, and contract verification is complete with a `go-with-risks` recommendation.
 - Sprint 1 is complete with a minimal Python package, CLI placeholder, and fast pytest loop.
 - Sprint 4 is implemented with a mock-tested direct local MinerU CLI adapter.
 - Sprint 5 is implemented with a pure Markdown normalizer and local-only unit tests.
 - Sprint 6 is implemented with local quality checks and report string rendering.
 - Sprint 7 is implemented with `convert_pdf`, `convert_input`, output writing, metadata/report writing, local asset copying, batch conversion, and `pdf2md convert`.
 - Sprint 7 is implemented with fake-adapter CLI/API tests.
 - Sprint 8 is implemented and committed.
 - Sprint 9 is implemented, independently evaluated, and committed.
 - The project `.venv` has been rebuilt with CUDA-enabled PyTorch and MinerU 3.1.0.
 - Latest `samples/FourNodeQuadrilateralShellElementMITC4.pdf` conversion completed with default GPU request and wrote Markdown, metadata JSON, report Markdown, and assets under ignored `outputs/FourNodeQuadrilateralShellElementMITC4/`.
 - Latest `samples/MITC공부.pdf` conversion completed on GPU and wrote Markdown, metadata JSON, report Markdown, and assets under ignored `outputs/MITC공부/`.
 - `docs/MATHJAXCHECKERPLAN.md` now documents the local MathJax render checker plan and implementation status.
 - Local MathJax render checker code now exists with optional local Node.js/`mathjax` setup, default conversion integration, and `doctor` diagnostics.
 - `docs/Sprints/SPRINT10CONTRACT.md` now documents the implemented long-PDF pre-conversion chunking sprint.
 ## Environment Notes
 - OS/workspace: Windows PowerShell in `D:\Work\Repos\AICoding\ConvertPDFToMD`.
 - Python target: 3.12.
- Local Python observed during Sprint 0: 3.12.7.
+- Local Python observed: 3.12.7.
- `uv` observed during Sprint 0: not available on PATH.
+- `uv` is installed per-user at `C:\Users\user\.local\bin`.
- `uv` installed during Sprint 1: 0.11.11 at `C:\Users\user\.local\bin`.
+- Target GPU: NVIDIA GTX 1070 Ti 8GB.
- If a new shell cannot find `uv`, restart the shell or add `C:\Users\user\.local\bin` to PATH.
+- Default conversion device: `cuda:0`.
- GPU target: GTX 1070 Ti 8GB.
+- MinerU execution mode: direct local `mineru` CLI only.
- Local GPU observed during Sprint 0: NVIDIA GeForce GTX 1070 Ti, driver 577.00, 8192 MiB VRAM, WDDM.
+- Strict-local allows MinerU 3.1.0's CLI-internal temporary local `mineru-api` when the CLI runs without `--api-url`.
 - Sample PDFs are in `samples/` and include Korean filenames.
 - MinerU execution mode: direct local CLI only.
 - MinerU 3.1.0 CLI-internal temporary local `mineru-api` is allowed when CLI runs without `--api-url`.
 - Strict-local prohibits `--api-url`, remote APIs, router mode, HTTP client backends, and remote OpenAI-compatible backends.
- MinerU planning pin: `mineru[core]==3.1.0` unless Sprint 1 or Sprint 8 proves another 3.1.0 extra is required.
+- Current local runtime has CUDA-enabled PyTorch `2.6.0+cu126`, `torchvision 0.21.0+cu126`, `mineru[core]==3.1.0`, local MinerU models, and `MINERU_MODEL_SOURCE=local`.
- MinerU 3.1.0 was installed in the local `.venv` with `uv pip install "mineru[core]==3.1.0"` for real CLI probing.
+- Current `pdf2md doctor` status is WARN only because GTX 1070 Ti is Pascal/pre-Turing; MinerU, CUDA PyTorch, local model config, MathJax, and strict-local checks pass.
 - Current `pdf2md doctor` status is WARN: MinerU CLI is present, GTX 1070 Ti is visible with Pascal/pre-Turing risk, PyTorch is `2.6.0+cu126` with CUDA available, local MinerU model config is detected, local MathJax checker passes after `npm install`, and strict-local policy passes.
 - User-level environment variable `MINERU_MODEL_SOURCE=local` is set so MinerU uses the downloaded local model paths in `C:\Users\user\mineru.json`.
-## Completed Work
+## Recent Completed Work
- Created initial project documents.
+- Archived completed sprint and setup history into `docs/WORKARCHIVE.md`.
- Originally selected MinerU 2.5, then changed the fixed engine target to MinerU 3.1.0 after user approval.
+- Added `docs/WORKARCHIVE.md` references to `AGENTS.md`, `PLAN.md`, `docs/V1IMPLEMENTATIONPLAN.md`, relevant `.codex/agents/*.toml`, `.codex/commands/*.md`, and project skills.
- Split architecture details into `ARCHITECTURE.md`.
+- Sprint 10 is implemented with `pypdf>=6.10.2,<7`, `src/pdf2md/pdf_splitter.py`, `--chunk-pages [PAGES]`, chunk-aware conversion orchestration, temporary chunk cleanup, and chunk report context.
 - Aligned documents with `Project Guidelines`.
 - Added this shared planning/progress workflow.
 - Decided MinerU failures must produce clear warnings/errors without silent fallback.
 - Decided every conversion should produce metadata JSON and a human-readable `.report.md`.
 - Created custom agent specs for research, requirements, MinerU integration, Obsidian Markdown, metadata, evaluation, local setup, and license/privacy work.
 - Created project prompt commands for startup, MinerU research, document review, integration planning, and quality evaluation planning.
 - Created project skills for MinerU research, math Markdown review, and fixture evaluation.
 - Created project hooks for startup context, pre-tool policy checks, and stop-time completion reminders.
 - Read Anthropic's long-running harness design article and adapted its planner/generator/evaluator pattern for this repository.
 - Added `harness-planner-agent` and `feature-generator-agent`.
 - Strengthened `evaluation-agent` as an independent contract reviewer and skeptical QA evaluator.
 - Added long-running harness workflow guidance to `AGENTS.md`.
 - Created `docs/V1IMPLEMENTATIONPLAN.md` with v1 sprint sequencing, contracts, verification gates, and agent ownership.
 - Created `docs/Sprints/SPRINT0CONTRACT.md` for source and environment verification before implementation.
 - Added direct `docs/V1IMPLEMENTATIONPLAN.md` and `docs/Sprints/SPRINT0CONTRACT.md` references to the agents that need them.
 - Completed Sprint 0 contract evaluation; result was PASS.
 - Completed the original Sprint 0 MinerU 2.5.4 package, CLI shape, output layout, model/cache, and strict-local risk verification from primary sources.
 - Verified local Python, `uv`, and GPU facts using the allowed Sprint 0 commands.
 - Verified MinerU/model license and privacy posture for personal/research local use versus redistribution.
 - Updated `docs/KNOWLEDGEBASE.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT0CONTRACT.md` with Sprint 0 findings.
 - Completed post-output Sprint 0 evaluation. The only missing acceptance item at review time was the final commit.
 - Redefined strict-local policy for MinerU 3.1.0: allow direct `mineru` CLI and CLI-internal temporary local `mineru-api`; prohibit `--api-url`, remote APIs, router mode, HTTP client backends, and remote OpenAI-compatible backends.
 - Updated core project documents and `.codex` workflow assets to reflect MinerU 3.1.0 and the redefined strict-local policy.
 - Checked MinerU 3.1.0 sources: PyPI 3.1.0 metadata, MinerU release notes, quick usage docs, CLI tools docs, output file docs, and model source docs.
 - Created `docs/Sprints/SPRINT1CONTRACT.md` for project scaffold and fast test loop planning.
 - Added direct `docs/Sprints/SPRINT1CONTRACT.md` references to the agents that need Sprint 1 scaffold context.
 - Started Sprint 1 implementation and amended the contract to include `uv.lock`, which is generated by `uv sync`.
 - Installed `uv` per-user using the official Astral installer.
 - Created Sprint 1 scaffold files: `pyproject.toml`, `uv.lock`, `.gitignore`, `README.md`, `src/pdf2md/__init__.py`, `src/pdf2md/cli.py`, `tests/test_package.py`, and `tests/test_cli.py`.
 - Verified `uv sync` with a temporary project environment outside the repo.
 - Verified `uv run pytest` passes with 4 tests.
 - Verified `uv run pdf2md --version` prints `pdf2md 0.1.0`.
 - Verified `git diff --check` passes.
 - Checked the scaffold for disallowed MinerU, remote API, router, HTTP, or OpenAI backend references.
 - Completed independent Sprint 1 evaluation once; the only scope failure was that `PLAN.md` was updated for shared workflow coordination before the Sprint 1 contract listed it as an allowed touched surface.
 - Amended the Sprint 1 contract to allow minimal `PLAN.md` current-goal coordination updates.
 - Completed the final independent Sprint 1 evaluation; result was PASS.
 - Created `docs/Sprints/SPRINT2CONTRACT.md` for paths, input discovery, and overwrite planning.
 - Added direct `docs/Sprints/SPRINT2CONTRACT.md` references to the agents that need Sprint 2 path planning context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 2 at the new contract and current scaffold state.
 - Verified the Sprint 2 contract documentation change with `git diff --check` and `uv run pytest` passing 4 tests.
 - Started Sprint 2 implementation after user approval and pre-implementation contract review PASS.
 - Added `src/pdf2md/paths.py` for input discovery, output path planning, overwrite conflict detection, duplicate output detection, and output-root escape prevention.
 - Added `tests/test_paths.py` with temporary-file coverage for single PDF discovery, directory discovery, recursive discovery, deterministic ordering, Korean filenames, output path planning, overwrite behavior, duplicate planned outputs, and output-root escape prevention.
 - Completed independent Sprint 2 evaluation once; the only hard failure was a Windows rooted/drive-relative `relative_parent` escape case.
 - Fixed output-root escape prevention by rejecting absolute, rooted, drive-qualified, and `..` relative parents and validating resolved planned outputs stay under the output root.
 - Verified `uv run pytest tests/test_paths.py` passes 17 tests.
 - Verified `uv run pytest` passes 21 tests.
 - Verified `git diff --check` passes for the Sprint 2 implementation.
 - Checked the implementation for disallowed MinerU, remote API, router, HTTP, OpenAI backend, or network client references.
 - Completed the final independent Sprint 2 evaluation; result was PASS.
 - Created `docs/Sprints/SPRINT3CONTRACT.md` for domain records, metadata construction, and warning aggregation planning.
 - Added direct `docs/Sprints/SPRINT3CONTRACT.md` references to the agents that need Sprint 3 metadata context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 3 at the new contract and current path-planning state.
 - Verified the Sprint 3 contract documentation change with `git diff --check` and `uv run pytest` passing 21 tests.
 - Started Sprint 3 implementation after user approval and pre-implementation contract review PASS.
 - Added `src/pdf2md/ir.py` for project-owned document, page, block, asset, and warning records with stable block types, warning codes, and severities.
 - Added `src/pdf2md/metadata.py` for JSON-serializable metadata construction and summary counts from project-owned records.
 - Added `tests/test_ir.py` and `tests/test_metadata.py` covering record serialization, optional field preservation/omission, invalid enum/severity validation, metadata top-level fields, summary counts, warning order, JSON serializability, and required input validation.
 - Verified `uv run pytest tests/test_ir.py tests/test_metadata.py` passes 25 tests.
 - Verified `uv run pytest` passes 46 tests.
 - Verified `git diff --check` passes for the Sprint 3 implementation.
 - Checked the implementation for disallowed remote API, router, HTTP, OpenAI backend, network client, MinerU adapter, and doctor references.
 - Completed the final independent Sprint 3 evaluation; result was PASS.
 - Created `docs/Sprints/SPRINT4CONTRACT.md` for direct local MinerU CLI adapter boundary planning with mocked subprocess/output tests.
 - Added direct `docs/Sprints/SPRINT4CONTRACT.md` references to the agents that need Sprint 4 MinerU adapter context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 4 at the new contract and current metadata-model state.
 - Verified the Sprint 4 contract documentation change with `git diff --check` and `uv run pytest` passing 46 tests.
 - Started Sprint 4 implementation after user approval and pre-implementation contract review PASS.
 - Added `src/pdf2md/mineru_adapter.py` for the direct local MinerU CLI adapter boundary, mockable availability/version checks, deterministic command construction, subprocess result capture, strict-local option validation, optional mocked-output parsing, and adapter warning mapping.
 - Added `tests/test_mineru_adapter.py` with fake-runner coverage for availability, missing MinerU, version success/failure/empty output, fixed command shape, custom executable rejection, strict-local rejection, mocked success, non-zero exit, missing output, and invalid JSON.
 - Fixed an independent evaluation finding that a caller-controlled executable could bypass strict-local policy; v1 now accepts only the direct `mineru` executable name, and user-exposed `mineru-api` execution is rejected.
 - Verified `uv sync` passes.
 - Verified `uv run pytest tests/test_mineru_adapter.py` passes 26 tests.
 - Verified `uv run pytest` passes 72 tests.
 - Verified `git diff --check` passes for the Sprint 4 implementation.
 - Checked the implementation for network client imports; none were found.
 - Checked strict-local prohibited tokens in `src/pdf2md`; matches are limited to deliberate validation literals in `mineru_adapter.py`.
 - Completed the final independent Sprint 4 evaluation; result was PASS.
 - Created `docs/Sprints/SPRINT5CONTRACT.md` for Obsidian Markdown normalization, math delimiter handling, asset link normalization, and conservative table fallback planning.
 - Added direct `docs/Sprints/SPRINT5CONTRACT.md` references to the agents that need Sprint 5 Markdown, warning, implementation, planning, or evaluation context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 5 at the new contract and current Sprint 4 implementation state.
 - Verified the Sprint 5 contract documentation change with agent TOML parsing, `git diff --check`, and `uv run pytest` passing 72 tests.
 - Started Sprint 5 implementation after user approval and pre-implementation contract review PASS.
 - Added `src/pdf2md/markdown.py` for project-owned Obsidian Markdown normalization, inline/display math delimiter handling, code fence and inline code protection, relative asset link normalization, local asset warning behavior, and conservative table fallback warnings.
 - Added `tests/test_markdown.py` covering inline math, display math spacing, idempotency, math body preservation, code protection, asset path normalization, invalid/missing/remote asset warnings, simple table preservation, and complex HTML table fallback warnings.
 - Added narrow warning codes `ASSET_LINK_INVALID` and `TABLE_FALLBACK` to `src/pdf2md/ir.py`.
 - Verified `uv sync` passes.
 - Verified `uv run pytest tests/test_markdown.py tests/test_ir.py` passes 30 tests.
 - Verified `uv run pytest` passes 89 tests.
 - Verified `git diff --check` passes for the Sprint 5 implementation.
 - Checked the implementation for network client imports; none were found.
 - Checked the implementation for conversion orchestration, metadata writing, report generation, and CLI convert behavior; no Sprint 5 code introduced those paths.
 - Completed the final independent Sprint 5 evaluation; result was PASS.
 - Created `docs/Sprints/SPRINT6CONTRACT.md` for local quality checks, math renderability boundary, metadata summary extensions, report content rendering, and final status planning.
 - Added direct `docs/Sprints/SPRINT6CONTRACT.md` references to the agents that need Sprint 6 quality, reporting, metadata, math renderability, implementation, planning, or evaluation context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 6 at the new contract and current Sprint 5 implementation state.
 - Verified the Sprint 6 contract documentation change with agent TOML parsing, `git diff --check`, and `uv run pytest` passing 89 tests.
 - Started Sprint 6 implementation after user approval and pre-implementation contract review PASS.
 - Added `src/pdf2md/quality.py` for local asset-link checks, math renderability checker boundaries, nonfatal checker-unavailable behavior, and quality result aggregation.
 - Added `src/pdf2md/report.py` for human-readable quality report content rendering from metadata and quality results, pages-with-warnings derivation, and final status calculation.
 - Added `tests/test_quality.py` covering missing/invalid asset links, code-block exclusions, fake math checker failures, checker-unavailable behavior, and quality result merging.
 - Added `tests/test_report.py` covering required report content, optional path handling, pages-with-warnings, final status policy, metadata/quality count use, and no report-file creation.
 - Verified `uv sync` passes.
 - Verified `uv run pytest tests/test_quality.py tests/test_report.py tests/test_metadata.py` passes 26 tests.
 - Verified `uv run pytest` passes 103 tests.
 - Verified `git diff --check` passes for the Sprint 6 implementation.
 - Checked the implementation for network client imports; none were found.
 - Checked the implementation for conversion orchestration, final output writing, metadata JSON writing, `.report.md` file writing, real MinerU invocation, setup scripts, and CLI convert behavior; no Sprint 6 code introduced those paths.
 - Completed the final independent Sprint 6 evaluation; result was PASS.
 - Completed the final independent Sprint 7 evaluation after fixing math renderability metadata counts; result was PASS.
 - Started Sprint 8 implementation after user approval.
 - Added `src/pdf2md/doctor.py` for mockable setup diagnostics covering Python 3.12, `uv`, MinerU availability/version, NVIDIA GPU visibility, PyTorch CUDA visibility, local model/cache/config detection, and strict-local policy reporting.
 - Added `pdf2md doctor` CLI integration without changing `pdf2md convert` or `pdf2md --version` behavior.
 - Updated `README.md` with Windows PowerShell setup, `uv`, MinerU 3.1.0 direct CLI expectations, model/cache environment notes, GTX 1070 Ti risk, and strict-local runtime policy.
 - Added mocked doctor and CLI tests for success, warning-only success, hard dependency failure, missing `uv`, missing MinerU, MinerU version warnings, missing GPU/PyTorch warnings, GTX 1070 Ti/Pascal risk, and missing model/cache warnings.
 - Verified `uv run pytest tests/test_doctor.py tests/test_cli.py` passes 22 tests.
 - Verified `uv sync` passes.
 - Verified `uv run pytest` passes 133 tests.
 - Verified `uv run pdf2md --version` prints `pdf2md 0.1.0`.
 - Verified local `uv run pdf2md doctor` returns exit code 1 because MinerU is not installed; it reports Python and `uv` pass, GTX 1070 Ti/Pascal risk warning, PyTorch missing warning, model/cache missing warning, and strict-local pass.
 - Completed independent Sprint 8 evaluation; result was PASS.
 - Committed Sprint 8 implementation as `7d965e3 feat: implement sprint 8 doctor diagnostics`.
 - Created `docs/Sprints/SPRINT9CONTRACT.md` for local fixture evaluation and the v1 release gate.
 - Added direct `docs/Sprints/SPRINT9CONTRACT.md` references to the agents that need Sprint 9 fixture evaluation, release gate, strict-local, or implementation context.
 - Updated `docs/V1IMPLEMENTATIONPLAN.md` to point Sprint 9 at the new contract and current Sprint 8 completion state.
 - Started Sprint 9 implementation after user approval and pre-implementation contract review PASS.
 - Added fast mocked v1 release-gate integration tests in `tests/integration/test_v1_fast_release_gate.py`.
 - Added explicit opt-in local MinerU fixture evaluation in `tests/integration/test_optional_mineru_fixtures.py`, gated by `PDF2MD_RUN_MINERU_FIXTURES=1`.
 - Added `docs/V1RELEASECHECKLIST.md` with default fast gates, strict-local release gates, doctor hard-failure handling, optional sample gates, fixture coverage notes, and no-sample-commit checks.
 - Updated `README.md` to point at the v1 release checklist and optional fixture evaluation gate.
 - Verified `uv run pytest tests/integration tests/test_conversion.py tests/test_cli.py` passes 24 tests with 1 optional skip.
 - Verified `uv run pytest tests/integration` passes 3 fast tests with 1 optional skip.
 - Verified opt-in `PDF2MD_RUN_MINERU_FIXTURES=1 uv run pytest -rs tests/integration/test_optional_mineru_fixtures.py` is skipped with a clear doctor blocker because MinerU is not installed.
 - Verified `uv run pytest` passes 136 tests with 1 optional skip.
 - Completed independent Sprint 9 evaluation; result was PASS.
 - Committed Sprint 9 implementation as `466abcf feat: implement sprint 9 release gate`.
 - Attempted `samples/MITC공부.pdf` conversion after installing MinerU; the run did not produce a successful conversion and was stopped after the user observed CPU-bound execution.
 - Added `outputs/` to `.gitignore` and removed the leftover generated output directory from the stopped sample run.
 - Updated default conversion behavior so `convert_pdf`, `convert_input`, and `pdf2md convert` default to `cuda:0`.
 - Updated the MinerU adapter to map CUDA requests to the MinerU subprocess environment with `MINERU_DEVICE_MODE` and `CUDA_VISIBLE_DEVICES`, preserving strict-local direct CLI execution.
 - Updated README, PRD, and architecture docs to document GPU default behavior and the remaining CUDA/PyTorch requirement.
 - Verified the GPU-default change with targeted tests, full tests, `git diff --check`, CLI help, and `pdf2md doctor`.
 - Re-ran `uv run pdf2md convert samples\MITC공부.pdf --out outputs\MITC공부 --overwrite`; the CLI reported `converted: 0`, `failed: 1`, `warnings: 1`, and wrote no Markdown, metadata JSON, or `.report.md`.
 - Confirmed the failure with a direct adapter probe using an ASCII work directory: MinerU 3.1.0 started its allowed temporary local `mineru-api`, used `hybrid-auto-engine`, attempted to load the VLM model on CUDA, and failed with `AssertionError: Torch not compiled with CUDA enabled`.
 - Left the product conversion stdout/stderr logs under ignored `outputs/MITC공부.logs`; removed temporary diagnostic probe directories.
 - Removed and recreated the project `.venv` with `uv sync`.
 - Installed CUDA-enabled PyTorch runtime: `torch==2.6.0+cu126` and `torchvision==0.21.0+cu126`.
 - Verified CUDA with an actual tensor operation on `NVIDIA GeForce GTX 1070 Ti`, compute capability `6.1`.
 - Installed `mineru[core]==3.1.0`; verified `mineru, version 3.1.0`.
 - Downloaded MinerU pipeline and VLM models with `uv run mineru-models-download -s huggingface -m all`; MinerU wrote model paths to `C:\Users\user\mineru.json`.
 - Set `MINERU_MODEL_SOURCE=local` at user scope and current process scope.
 - Verified `uv run pdf2md doctor` reports PyTorch CUDA available and model config detected; remaining WARN status is the intentional Pascal/pre-Turing GPU risk warning.
 - Verified `uv run pytest` passes 138 tests with 1 optional skip in the rebuilt environment.
 - Fixed real MinerU nested `images/...` asset-link rewriting so copied assets under `<stem>.assets/.../images/` resolve from the final Markdown.
 - Fixed page-count extraction for MinerU-style structured lists with `page_idx` values.
 - Verified `uv run pytest tests/test_conversion.py` passes 12 tests.
 - Verified `uv run pytest` passes 139 tests with 1 optional skip after the asset/page-count fix.
 - Re-ran `samples/MITC공부.pdf` conversion with `MINERU_MODEL_SOURCE=local`; MinerU used GPU via the direct local CLI and CLI-internal temporary local `mineru-api`.
 - The final sample outputs are `outputs/MITC공부/MITC공부.md`, `outputs/MITC공부/MITC공부.metadata.json`, `outputs/MITC공부/MITC공부.report.md`, and `outputs/MITC공부/MITC공부.assets/`.
 - The final sample report status is `partial` only because the local math render checker is unavailable; asset link checks pass with 0 missing and 0 invalid links.
 - Sample summary: 13 pages processed, 107 assets, 23 inline formulas, 103 display formulas, 1 info warning.
 - Added `docs/MATHJAXCHECKERPLAN.md` with the local MathJax checker objective, touched surfaces, Node helper contract, Python wrapper behavior, tests, acceptance criteria, and open implementation decisions.
 - Implemented the local MathJax render checker with `MathExpression` extraction, a local Node.js helper, a Python wrapper, default conversion integration, `doctor` diagnostics, setup documentation, and mocked tests.
 - Verified `npm run mathjax-checker:health` returns `{"ok":true}` after local `npm install`.
 - Verified direct helper JSON stdin reports one valid expression as ok and a malformed display formula as `Missing close brace`.
 - Verified `create_default_math_checker()` finds the local checker and records one render failure for malformed display math.
 - Verified targeted tests pass: `uv run pytest tests/test_quality.py tests/test_math_render.py tests/test_conversion.py tests/test_doctor.py tests/test_cli.py`.
 - Verified full tests pass: `uv run pytest` passed 150 tests with 1 optional skip.
 - Verified `git diff --check` passes.
 - Researched local PDF chunking packages and MinerU page-range behavior for Sprint 10.
 - Created `docs/Sprints/SPRINT10CONTRACT.md` recommending `pypdf>=6.10.2,<7` for 20-page local chunk PDFs, with chunk outputs converted independently and no Markdown merge.
 - Implemented Sprint 10 with `pypdf>=6.10.2,<7`, `src/pdf2md/pdf_splitter.py`, `--chunk-pages [PAGES]`, chunk-aware conversion orchestration, and chunk report context.
 - `--chunk-pages` is opt-in; when present without a value it uses 20 pages.
 - `convert_pdf()` returns `BatchConversionResult` when `chunk_pages` is set and keeps returning `ConversionResult` when chunking is unset.
- Temporary chunk PDFs are deleted after conversion completes, including when raw MinerU output is retained.
+- Converted `samples/FourNodeQuadrilateralShellElementMITC4.pdf` with `MINERU_MODEL_SOURCE=local` and default `--gpu cuda:0`; output was written to ignored `outputs/FourNodeQuadrilateralShellElementMITC4/`.
 - Verified targeted Sprint 10 tests: `uv run pytest tests/test_pdf_splitter.py tests/test_conversion.py tests/test_cli.py tests/test_report.py` passed 42 tests.
 - Verified full default test suite: `uv run pytest` passed 163 tests with 1 optional skip.
 - Verified `git diff --check` passed with line-ending warnings only.
 - Ran `uv run pdf2md doctor`; status was WARN only for GTX 1070 Ti/Pascal compatibility risk while MinerU 3.1.0, CUDA PyTorch, local models, MathJax, and strict-local checks passed.
 - Converted `samples/FourNodeQuadrilateralShellElementMITC4.pdf` with `MINERU_MODEL_SOURCE=local` and default `--gpu cuda:0`; output was written to `outputs/FourNodeQuadrilateralShellElementMITC4/`.
 - The FourNode sample conversion report status was `success`: 7 pages, 22 assets, 38 inline formulas, 16 display formulas, 0 math render errors, and 0 warnings.
 ## In Progress
@@ -253,189 +43,11 @@ This file records actual progress for agents. Read it before starting work, then
 ## Blockers
- No active blocker for the completed `samples/MITC공부.pdf` conversion.
+- No active blocker.
- GTX 1070 Ti remains an 8GB Pascal GPU; larger PDFs may still hit VRAM or model compatibility limits even though this sample completed.
+- GTX 1070 Ti remains an 8GB Pascal GPU; larger PDFs may still hit VRAM or model compatibility limits.
 ## Next Actions
-1. Review the generated `outputs/MITC공부/MITC공부.md` in Obsidian if visual quality needs manual assessment.
+1. Review generated sample Markdown outputs in Obsidian if visual quality needs manual assessment.
 2. Run optional real local chunked conversion on a long sample only if requested.
-3. Run `npm install` and `npm run mathjax-checker:health` when real local MathJax checker validation is desired.
+3. Preserve strict-local runtime behavior: use local model paths, direct CLI execution, and no user-specified API or remote backend.
 4. Preserve the strict-local rule: setup downloads may be explicit, but runtime conversion must use local model paths, direct CLI execution, and no user-specified API or remote backend.
 ## Sprint 9 Handoff
 - Files changed: `tests/integration/test_v1_fast_release_gate.py`, `tests/integration/test_optional_mineru_fixtures.py`, `docs/V1RELEASECHECKLIST.md`, `README.md`, `PLAN.md`, `PROGRESS.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT9CONTRACT.md`.
 - Commands run: `uv run pytest tests/integration tests/test_conversion.py tests/test_cli.py`, `uv run pytest tests/integration`, `PDF2MD_RUN_MINERU_FIXTURES=1 uv run pytest -rs tests/integration/test_optional_mineru_fixtures.py`, `uv run pytest`, `git diff --check`, and `git status --short --untracked-files=all`.
 - Tests passed: targeted integration/CLI/conversion run passed 24 tests with 1 optional skip; integration-only run passed 3 fast tests with 1 optional skip; full `uv run pytest` passed 136 tests with 1 optional skip.
 - Tests blocked: optional real MinerU fixture conversion is blocked by `pdf2md doctor` because the `mineru` CLI is not installed on PATH.
 - Optional local MinerU status: explicitly gated by `PDF2MD_RUN_MINERU_FIXTURES=1`; current opt-in run skips with doctor blocker instead of pretending real validation passed.
 - Fixture coverage: release checklist maps local samples to math-heavy, table/formula, figures/assets, reading-order, and Korean filename/path risk categories; simple one-page, table-dominant, and figure-heavy known-baseline gaps remain.
 - Generated output locations: none persisted; optional output path uses pytest `tmp_path`.
 - Known failures: local doctor fails on missing MinerU CLI.
 - Independent evaluation: PASS.
 - Residual risks: no real MinerU output has been validated yet; GTX 1070 Ti/PyTorch acceleration and model/cache setup remain unproven; optional fixture quality still requires local MinerU setup.
 - User decisions needed: decide whether to install/configure MinerU 3.1.0 and run optional fixture validation.
 - V1 release recommendation: default fast gates are healthy, but full real-MinerU v1 validation is blocked until doctor passes or the user records a waiver.
 - Go/no-go recommendation for next sprint: go only for real setup/fixture validation if the user wants to proceed with local MinerU installation.
 - Next action: commit Sprint 9 implementation.
 ## Sprint 9 Contract Handoff
 - Files changed: `docs/Sprints/SPRINT9CONTRACT.md`, `docs/V1IMPLEMENTATIONPLAN.md`, relevant `.codex/agents/*.toml`, `PLAN.md`, and `PROGRESS.md`.
 - Commands run: `uv --version`, `uv sync`, agent TOML parse check, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and local sample filename listing.
 - Tests passed: `uv run pytest` passed 133 tests.
 - Tests blocked: None expected for the contract-only change.
 - Known failures: local `pdf2md doctor` still fails until MinerU is installed on PATH.
 - Residual risks: Sprint 9 is contract-only; fast mocked integration tests, optional local MinerU fixture harness, fixture coverage manifest, and release checklist are not implemented yet.
 - User decisions needed: None before Sprint 9 pre-implementation review.
 - Go/no-go recommendation for Sprint 9 implementation: review the contract first, then go if the user explicitly requests implementation.
 - Next action: verify and commit the Sprint 9 contract update.
 ## Sprint 8 Handoff
 - Files changed: `src/pdf2md/doctor.py`, `src/pdf2md/cli.py`, `tests/test_doctor.py`, `tests/test_cli.py`, `README.md`, `PLAN.md`, `PROGRESS.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT8CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_doctor.py tests/test_cli.py`, `uv run pytest tests/test_doctor.py`, `uv run pytest`, `uv run pdf2md --version`, `uv run pdf2md doctor`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell strict-local/network pattern checks.
 - Tests passed: `uv run pytest tests/test_doctor.py tests/test_cli.py` passed 22 tests; `uv run pytest tests/test_doctor.py` passed 11 tests; `uv run pytest` passed 133 tests.
 - Tests blocked: None.
 - Known failures: local `uv run pdf2md doctor` correctly fails because the `mineru` CLI is not installed on PATH.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully; local doctor warns for GTX 1070 Ti/Pascal risk, missing PyTorch, and missing MinerU model/cache/config path.
 - Independent evaluation: PASS.
 - Residual risks: Sprint 8 does not install MinerU, download models, validate real MinerU output, run sample PDFs, or prove GTX 1070 Ti PyTorch acceleration. Those remain Sprint 9/local setup work.
 - User decisions needed: None for Sprint 8.
 - Go/no-go recommendation for Sprint 9: go after a Sprint 9 contract is written and reviewed.
 - Next action at completion: prepare the Sprint 9 contract when requested.
 ## Sprint 8 Contract Handoff
 Historical note: this contract-only handoff was superseded by the implemented Sprint 8 handoff above.
 - Files changed: `docs/Sprints/SPRINT8CONTRACT.md`, `docs/V1IMPLEMENTATIONPLAN.md`, relevant `.codex/agents/*.toml`, `PLAN.md`, and `PROGRESS.md`.
 - Commands run: `uv --version`, agent TOML parse check, `uv sync`, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell reference checks.
 - Tests passed: `uv run pytest` passed 119 tests.
 - Tests blocked: None.
 - Known failures: none.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 8 is contract-only; `pdf2md doctor`, doctor diagnostics, setup docs, and setup helper scripts are not implemented yet.
 - User decisions needed: None before Sprint 8 pre-implementation review.
 - Go/no-go recommendation for Sprint 8 implementation: review the contract first, then go if the user explicitly requests implementation.
 - Next action: commit the contract update, then wait for an explicit Sprint 8 implementation request.
 ## Sprint 7 Handoff
 - Files changed: `src/pdf2md/conversion.py`, `src/pdf2md/cli.py`, `src/pdf2md/__init__.py`, `tests/test_conversion.py`, `tests/test_cli.py`, `tests/test_package.py`, `PLAN.md`, `PROGRESS.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT7CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_conversion.py tests/test_cli.py`, `uv run pytest tests/test_conversion.py tests/test_cli.py tests/test_package.py`, `uv run pytest tests/test_conversion.py tests/test_metadata.py tests/test_report.py`, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell strict-local/network pattern checks.
 - Tests passed: `uv run pytest tests/test_conversion.py tests/test_cli.py` passed 18 tests; `uv run pytest tests/test_conversion.py tests/test_cli.py tests/test_package.py` passed 16 tests before the math renderability fix; `uv run pytest tests/test_conversion.py tests/test_metadata.py tests/test_report.py` passed 29 tests; `uv run pytest` passed 119 tests after the metadata math count fix.
 - Tests blocked: None.
 - Known failures: none.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Independent evaluation: PASS after fixing math renderability metadata counts.
 - Residual risks: Sprint 7 uses fake adapters in default tests; it does not run real MinerU, probe real MinerU output, implement `pdf2md doctor`, validate CUDA/GPU, install models, or run sample PDFs.
 - User decisions needed: None for Sprint 7.
 - Go/no-go recommendation for Sprint 8: go.
 - Next action: prepare Sprint 8 contract when requested.
 ## Sprint 7 Contract Handoff (Historical)
 - Files changed: `docs/Sprints/SPRINT7CONTRACT.md`, `docs/V1IMPLEMENTATIONPLAN.md`, `.codex/agents/feature-generator-agent.toml`, `.codex/agents/evaluation-agent.toml`, `.codex/agents/requirements-guard-agent.toml`, `.codex/agents/harness-planner-agent.toml`, `.codex/agents/mineru-integration-agent.toml`, `.codex/agents/metadata-agent.toml`, `.codex/agents/obsidian-markdown-agent.toml`, `PLAN.md`, and `PROGRESS.md`.
 - Commands run: `uv --version`, agent TOML parse check, `uv sync`, `uv run pytest`, `git diff --check`, and `git status --short --untracked-files=all`.
 - Tests passed: `uv run pytest` passed 103 tests.
 - Tests blocked: None.
 - Known failures: none.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks at that time: Sprint 7 was contract-only before implementation. Superseded by the Sprint 7 Handoff above.
 - User decisions needed at that time: None before Sprint 7 pre-implementation review.
 - Go/no-go recommendation at that time: review the contract first, then go if the user explicitly requests implementation.
 - Next action at that time: commit the contract update, then wait for an explicit Sprint 7 implementation request.
 ## Sprint 6 Handoff
 - Files changed: `src/pdf2md/quality.py`, `src/pdf2md/report.py`, `tests/test_quality.py`, `tests/test_report.py`, `PLAN.md`, `PROGRESS.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT6CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_quality.py tests/test_report.py tests/test_metadata.py`, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell file/pattern checks.
 - Tests passed: `uv run pytest tests/test_quality.py tests/test_report.py tests/test_metadata.py` passed 26 tests; `uv run pytest` passed 103 tests.
 - Tests blocked: None.
 - Known failures: none in Sprint 6 implementation.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 6 intentionally does not run real MinerU, run a real math renderer, parse PDFs, write final Markdown files, copy assets, write metadata JSON files, write `.report.md` files, expose working `convert`, or implement `doctor`.
 - User decisions needed: None for Sprint 6.
 - Go/no-go recommendation for Sprint 7: go.
 - Next action: prepare Sprint 7 contract when requested.
 ## Sprint 5 Handoff
 - Files changed: `src/pdf2md/markdown.py`, `src/pdf2md/ir.py`, `tests/test_markdown.py`, `PLAN.md`, `PROGRESS.md`, `docs/V1IMPLEMENTATIONPLAN.md`, and `docs/Sprints/SPRINT5CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_markdown.py tests/test_ir.py`, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell file/pattern checks.
 - Tests passed: `uv run pytest tests/test_markdown.py tests/test_ir.py` passed 30 tests; `uv run pytest` passed 89 tests.
 - Tests blocked: None.
 - Known failures: none in Sprint 5 implementation.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 5 intentionally does not run real MinerU, probe real MinerU Markdown, parse PDFs, write final Markdown files, copy assets, write metadata JSON, generate `.report.md`, expose working `convert`, or implement `doctor`.
 - User decisions needed: None for Sprint 5.
 - Go/no-go recommendation for Sprint 6: go.
 - Next action: prepare Sprint 6 contract when requested.
 ## Sprint 4 Handoff
 - Files changed: `src/pdf2md/mineru_adapter.py`, `tests/test_mineru_adapter.py`, `PLAN.md`, and `PROGRESS.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_mineru_adapter.py`, `uv run pytest`, `git diff --check`, `git status --short --untracked-files=all`, and PowerShell file/pattern checks.
 - Tests passed: `uv run pytest tests/test_mineru_adapter.py` passed 26 tests; `uv run pytest` passed 72 tests.
 - Tests blocked: None.
 - Known failures: none in Sprint 4 implementation after fixing the independent evaluation finding.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 4 intentionally does not run real MinerU, install models, probe real MinerU output layout, parse PDFs, normalize Markdown, write metadata JSON, generate `.report.md`, expose working `convert`, or implement `doctor`.
 - User decisions needed: None for Sprint 4.
 - Go/no-go recommendation for Sprint 5: go.
 - Next action: prepare Sprint 5 contract when requested.
 ## Sprint 3 Handoff
 - Files changed: `src/pdf2md/ir.py`, `src/pdf2md/metadata.py`, `tests/test_ir.py`, `tests/test_metadata.py`, `PLAN.md`, `PROGRESS.md`, and `docs/Sprints/SPRINT3CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_ir.py tests/test_metadata.py`, `uv run pytest`, `git diff --check`, `git status --short`, and PowerShell file/pattern checks.
 - Tests passed: `uv run pytest tests/test_ir.py tests/test_metadata.py` passed 25 tests; `uv run pytest` passed 46 tests.
 - Tests blocked: None.
 - Known failures: none in Sprint 3 implementation.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 3 intentionally does not parse PDFs, compute SHA-256, invoke MinerU, write conversion outputs, normalize Markdown, create full report content, run quality checks, or expose working `convert` or `doctor` commands.
 - User decisions needed: None for Sprint 3.
 - Go/no-go recommendation for Sprint 4: go.
 - Next action: prepare Sprint 4 contract when requested.
 ## Sprint 2 Handoff
 - Files changed: `src/pdf2md/paths.py`, `tests/test_paths.py`, `PLAN.md`, and `PROGRESS.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest tests/test_paths.py`, `uv run pytest`, `git diff --check`, `git status --short`, and PowerShell file/pattern checks.
 - Tests passed: `uv run pytest tests/test_paths.py` passed 17 tests; `uv run pytest` passed 21 tests.
 - Tests blocked: None.
 - Known failures: none in Sprint 2 implementation.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: Sprint 2 intentionally does not parse PDFs, compute SHA-256, invoke MinerU, write conversion outputs, normalize Markdown, create metadata/report content, or expose a working `convert` command.
 - User decisions needed: None for Sprint 2.
 - Go/no-go recommendation for Sprint 3: go.
 - Next action: prepare Sprint 3 contract when requested.
 ## Sprint 1 Handoff
 - Files changed: `pyproject.toml`, `uv.lock`, `.gitignore`, `README.md`, `src/pdf2md/__init__.py`, `src/pdf2md/cli.py`, `tests/test_package.py`, `tests/test_cli.py`, `PROGRESS.md`, `PLAN.md`, and `docs/Sprints/SPRINT1CONTRACT.md`.
 - Commands run: `uv --version`, `uv sync`, `uv run pytest`, `uv run pdf2md --version`, `git diff --check`, `git status --short`, and PowerShell file/pattern checks after `rg.exe` returned access denied.
 - Tests passed: `uv run pytest` passed 4 tests.
 - Tests blocked: None.
 - Known failures: `uv` may not be visible to a newly opened shell until PATH is refreshed; `rg.exe` returned access denied in this environment, so PowerShell checks were used instead.
 - Known warnings: `uv` ignored Miniforge's invalid `SSL_CERT_DIR` path during sync/test commands, but the commands completed successfully.
 - Residual risks: the scaffold intentionally does not validate MinerU, CUDA, model paths, conversion output, metadata, or quality reports.
 - User decisions needed: None for Sprint 1.
 - Go/no-go recommendation for Sprint 2: go.
 - Next action: prepare Sprint 2 contract when requested.
 ## Sprint 0 Handoff
 Superseded note: the following Sprint 0 facts describe the completed 2.5.4 verification pass. The current engine decision is MinerU 3.1.0.
 - Files changed: `docs/KNOWLEDGEBASE.md`, `docs/Sprints/SPRINT0CONTRACT.md`, `docs/V1IMPLEMENTATIONPLAN.md`, `PROGRESS.md`.
 - Sources checked: MinerU 2.5.4 PyPI, MinerU 2.5.4 tag files, MinerU output/model docs, Python/uv/PyTorch/NVIDIA docs, MinerU/model license sources.
 - Local commands run: `python --version`, `uv --version`, `nvidia-smi`.
 - Facts confirmed at that time: Python 3.12.7 is present; `uv` is missing; GTX 1070 Ti 8GB is visible; MinerU 2.5.4 direct CLI path is source-verified; MinerU/model 2.5-era licenses should be treated as AGPL-3.0.
 - Inferences made at that time: v1 should pin MinerU to `mineru[core]==2.5.4`; strict-local runtime should require local model source configuration; current MinerU 3.x docs should not drive the older 2.5 adapter behavior.
 - Known failures: `uv --version` failed because `uv` is not on PATH.
 - Residual risks: GTX 1070 Ti/PyTorch CUDA compatibility, real MinerU output layout until local probe, AGPL redistribution obligations, setup-download versus runtime-local separation.
 - Go/no-go recommendation: `go-with-risks`.
 - Next action: resolve `uv` availability or include bootstrap handling in Sprint 1, then create the Sprint 1 contract.
@@ -58,6 +58,8 @@ Use the project long-running harness only for substantial implementation work.
 5. `evaluation-agent` independently verifies the result against the contract.
 6. The parent agent updates `PROGRESS.md`, commits the completed change, and leaves a handoff.
 After a chunk is no longer active, archive completed-work details in `docs/WORKARCHIVE.md` and keep `PROGRESS.md` focused on current status, blockers, and next actions.
 Each sprint contract must include:
 - Objective.
@@ -0,0 +1,90 @@
 # Work Archive
 Last updated: 2026-05-08
 This document stores completed project work, historical sprint outcomes, environment setup results, and sample conversion evidence. `PROGRESS.md` should stay focused on current status, blockers, and next actions. Read this archive when a task needs past implementation context, previous verification commands, or historical handoff details.
 ## Archive Policy
 - Keep completed sprint outcomes here after they no longer need to stay in `PROGRESS.md`.
 - Keep `PROGRESS.md` short and current.
 - Keep source-of-truth product requirements in `PRD.md` and architecture decisions in `ARCHITECTURE.md`.
 - Keep sprint contract details under `docs/Sprints/`.
 - Do not archive or commit sample PDFs or generated conversion outputs.
 ## Completed Milestones
 | Area | Completed Outcome | Primary References |
 | --- | --- | --- |
 | Core project documents | Created and iterated `PRD.md`, `AGENTS.md`, `ARCHITECTURE.md`, and `docs/KNOWLEDGEBASE.md`. | `PRD.md`, `AGENTS.md`, `ARCHITECTURE.md`, `docs/KNOWLEDGEBASE.md` |
 | Engine decision | Moved from the initial MinerU 2.5-era planning to fixed MinerU 3.1.0. | `PLAN.md`, `ARCHITECTURE.md`, `docs/KNOWLEDGEBASE.md` |
 | Strict-local policy | Redefined v1 runtime policy: allow direct local `mineru` CLI and its CLI-internal temporary local `mineru-api`; prohibit `--api-url`, remote APIs, router mode, HTTP client backends, and remote OpenAI-compatible backends. | `PRD.md`, `ARCHITECTURE.md`, `AGENTS.md` |
 | Agent workflow | Added shared `PLAN.md` and `PROGRESS.md` workflow, project-scoped agents, commands, skills, hooks, and the planner/generator/evaluator harness guidance. | `AGENTS.md`, `.codex/agents/`, `.codex/commands/`, `.codex/skills/`, `.codex/hooks.json` |
 | V1 implementation planning | Created `docs/V1IMPLEMENTATIONPLAN.md` and sprint contracts under `docs/Sprints/`. | `docs/V1IMPLEMENTATIONPLAN.md`, `docs/Sprints/*.md` |
 | Sprint 0 | Completed source, environment, license, privacy, and contract verification. Recommendation was `go-with-risks`. | `docs/Sprints/SPRINT0CONTRACT.md` |
 | Sprint 1 | Created minimal Python package scaffold, CLI placeholder, `pyproject.toml`, `uv.lock`, and fast pytest loop. | `docs/Sprints/SPRINT1CONTRACT.md` |
 | Sprint 2 | Implemented deterministic local PDF discovery, path planning, overwrite conflict checks, duplicate output checks, and output-root escape prevention. | `docs/Sprints/SPRINT2CONTRACT.md`, `src/pdf2md/paths.py` |
 | Sprint 3 | Implemented project-owned IR records, warning codes/severities, metadata construction, and metadata tests. | `docs/Sprints/SPRINT3CONTRACT.md`, `src/pdf2md/ir.py`, `src/pdf2md/metadata.py` |
 | Sprint 4 | Implemented mocked direct local MinerU CLI adapter boundary and strict-local validation. | `docs/Sprints/SPRINT4CONTRACT.md`, `src/pdf2md/mineru_adapter.py` |
 | Sprint 5 | Implemented Obsidian Markdown normalization, math delimiter handling, asset link normalization, and table fallback warnings. | `docs/Sprints/SPRINT5CONTRACT.md`, `src/pdf2md/markdown.py` |
 | Sprint 6 | Implemented local quality checks and human-readable report rendering from metadata and quality results. | `docs/Sprints/SPRINT6CONTRACT.md`, `src/pdf2md/quality.py`, `src/pdf2md/report.py` |
 | Sprint 7 | Implemented conversion orchestration, `convert_pdf`, `convert_input`, `pdf2md convert`, final Markdown writing, metadata/report writing, asset copying, and fake-adapter CLI/API tests. | `docs/Sprints/SPRINT7CONTRACT.md`, `src/pdf2md/conversion.py`, `src/pdf2md/cli.py` |
 | Sprint 8 | Implemented `pdf2md doctor`, local setup diagnostics, GPU/CUDA/PyTorch/model/cache checks, and setup documentation. | `docs/Sprints/SPRINT8CONTRACT.md`, `src/pdf2md/doctor.py`, `README.md` |
 | Sprint 9 | Implemented fast mocked v1 release-gate tests, optional local MinerU fixture evaluation, and `docs/V1RELEASECHECKLIST.md`. | `docs/Sprints/SPRINT9CONTRACT.md`, `docs/V1RELEASECHECKLIST.md`, `tests/integration/` |
 | GPU default/runtime setup | Made conversion default to `cuda:0`, mapped CUDA requests to MinerU subprocess environment variables, rebuilt `.venv`, installed CUDA-enabled PyTorch and MinerU 3.1.0, downloaded MinerU models, and set `MINERU_MODEL_SOURCE=local`. | `README.md`, `src/pdf2md/mineru_adapter.py`, `src/pdf2md/conversion.py` |
 | MathJax checker | Planned and implemented local MathJax render checker with Node.js helper, Python wrapper, conversion integration, and doctor diagnostics. | `docs/MATHJAXCHECKERPLAN.md`, `tools/mathjax-checker/check.mjs`, `src/pdf2md/math_render.py` |
 | Sprint 10 | Implemented opt-in pre-conversion PDF chunking with `pypdf`, temporary chunk PDF cleanup, `--chunk-pages [PAGES]`, chunk metadata/report context, and mocked tests. | `docs/Sprints/SPRINT10CONTRACT.md`, `src/pdf2md/pdf_splitter.py` |
 ## Runtime Setup Archive
 - OS/workspace: Windows PowerShell in `D:\Work\Repos\AICoding\ConvertPDFToMD`.
 - Python target: 3.12.
 - Local Python observed during Sprint 0: 3.12.7.
 - `uv` installed per-user at `C:\Users\user\.local\bin`.
 - GPU target: NVIDIA GTX 1070 Ti 8GB.
 - Local GPU observed: NVIDIA GeForce GTX 1070 Ti, driver 577.00, 8192 MiB VRAM, WDDM.
 - MinerU execution mode: direct local `mineru` CLI only.
 - MinerU 3.1.0 CLI-internal temporary local `mineru-api` is allowed when the CLI runs without `--api-url`.
 - GTX 1070 Ti runtime setup used `torch==2.6.0+cu126`, `torchvision==0.21.0+cu126`, and `mineru[core]==3.1.0`.
 - MinerU models were downloaded with `uv run mineru-models-download -s huggingface -m all`.
 - Runtime model loading uses `MINERU_MODEL_SOURCE=local`.
 - Current doctor status after setup is WARN because GTX 1070 Ti is Pascal/pre-Turing; MinerU, CUDA PyTorch, local model config, MathJax checker, and strict-local checks pass.
 ## Sample Conversion Archive
 Generated outputs are ignored under `outputs/` and are not committed.
 | Sample | Result | Output Location | Summary |
 | --- | --- | --- | --- |
 | `samples/MITC공부.pdf` | Completed after CUDA-enabled runtime setup. | `outputs/MITC공부/` | 13 pages, 107 assets, 23 inline formulas, 103 display formulas, 1 info warning at the time of that run because the local MathJax checker was unavailable. |
 | `samples/FourNodeQuadrilateralShellElementMITC4.pdf` | Completed with default GPU request and `MINERU_MODEL_SOURCE=local`. | `outputs/FourNodeQuadrilateralShellElementMITC4/` | Report status `success`: 7 pages, 22 assets, 38 inline formulas, 16 display formulas, 0 math render errors, 0 warnings. |
 ## Historical Verification Highlights
 - Sprint 1 scaffold: `uv run pytest` passed 4 tests; `uv run pdf2md --version` printed `pdf2md 0.1.0`.
 - Sprint 2 path planning: `uv run pytest tests/test_paths.py` passed 17 tests; full suite passed 21 tests.
 - Sprint 3 metadata/IR: targeted IR/metadata tests passed 25 tests; full suite passed 46 tests.
 - Sprint 4 MinerU adapter: adapter tests passed 26 tests; full suite passed 72 tests.
 - Sprint 5 Markdown normalization: targeted Markdown/IR tests passed 30 tests; full suite passed 89 tests.
 - Sprint 6 quality/report: targeted quality/report/metadata tests passed 26 tests; full suite passed 103 tests.
 - Sprint 7 conversion orchestration: full suite passed 119 tests after metadata math count fixes.
 - Sprint 8 doctor: full suite passed 133 tests.
 - Sprint 9 release gate: full suite passed 136 tests with 1 optional skip.
 - CUDA runtime rebuild: verified CUDA with an actual tensor operation on `NVIDIA GeForce GTX 1070 Ti`, compute capability 6.1; `mineru --version` reported 3.1.0.
 - MathJax checker: `npm run mathjax-checker:health` returned `{"ok":true}` after local `npm install`; full suite passed 150 tests with 1 optional skip after integration.
 - Sprint 10 chunking: targeted chunking tests passed 42 tests; full default suite passed 163 tests with 1 optional skip; `git diff --check` passed with line-ending warnings only.
 ## Historical Blockers And Resolutions
 - Early `uv` availability was missing; resolved by installing `uv` per-user.
 - Initial real sample conversion failed when PyTorch was CPU-only; resolved by rebuilding `.venv` and installing CUDA-enabled PyTorch before MinerU.
 - Optional local MinerU fixture checks were originally blocked by missing MinerU CLI; resolved for the local environment after MinerU 3.1.0 and model setup.
 - Local MathJax render checking was originally unavailable; resolved after adding the local Node.js MathJax checker and installing local npm dependencies.
 - GTX 1070 Ti remains a Pascal/pre-Turing GPU risk. This is a warning, not a current blocker.
 ## Archive Notes For Agents
 - Read this file when a task asks for historical implementation state, prior verification, completed sprint context, or sample conversion evidence.
 - Prefer `PROGRESS.md` for current state and next actions.
 - Prefer sprint contract files for detailed acceptance criteria and handoff structure.
 - Prefer `docs/V1RELEASECHECKLIST.md` for release readiness gates.