modify skill

skills/wiki-ingest/SKILL.md

@@ -170,7 +170,7 @@ Steps:
 2. **Discuss** key takeaways with the user. Ask: "What should I emphasize? How granular?" Skip this if the user says "just ingest it."
 3. **Create** source summary in `wiki/sources/`. Use the source frontmatter schema from `references/frontmatter.md`. Assign an address per the **Address Assignment** section below.
 4. **Create or update** entity pages for every person, org, product, and repo mentioned. One page per entity. Assign addresses to new entity pages.
-5. **Create or update** concept pages for significant ideas and frameworks. Assign addresses to new concept pages.
+5. **Create or update** concept pages for significant ideas and frameworks. Assign addresses to new concept pages. Every concept page MUST include `source_refs` frontmatter as described in **Concept Source Provenance** below.
 6. **Update** relevant domain page(s) and their `_index.md` sub-indexes.
 7. **Update** `wiki/overview.md` if the big picture changed.
 8. **Update** `wiki/index.md`. Add entries for all new pages.
@@ -188,6 +188,33 @@ Steps:
 
 ---
 
+## Concept Source Provenance
+
+Every `wiki/concepts/*.md` page must show which raw Markdown chunks support its `sources` list. Preserve `sources:` wikilinks for human-readable source pages, and add `source_refs:` immediately after `sources:`. The `source_refs` list must have one entry per `sources` entry, in the same order.
+
+```yaml
+sources:
+  - "[[Midas-NFX-Analysis-Manual|Midas NFX Analysis Manual]]"
+source_refs:
+  - source: "[[Midas-NFX-Analysis-Manual|Midas NFX Analysis Manual]]"
+    raw_path: ".raw/MidasNFXAnalysisManual/"
+    raw_files:
+      - "MidasNFXAnalysisManual_024.md"
+    md_indices:
+      - 24
+    match: "direct-heading"
+    confidence: high
+```
+
+Rules:
+- `raw_path` is the `.raw` source folder or file backing the source page.
+- `raw_files` are exact `.md` filenames from that raw source; `md_indices` are the corresponding chunk numbers from suffixes like `_024.md`. For unsuffixed single-file sources, use `1`.
+- Prefer exact chunks read during ingest. If exact evidence was not captured, backfill by heading/keyword search over `.raw/**/*.md`, set `match: "heuristic-heading-keyword"`, and mark `confidence: high|medium|low` honestly.
+- Do not cite only table-of-contents, title, legal, or index chunks unless the concept actually comes from those chunks.
+- When updating an existing concept, keep existing reliable `source_refs` and add or revise entries only for changed `sources`.
+
+---
+
 ## Batch Ingest
 
 Trigger: user drops multiple files or says "ingest all of these."
@@ -244,6 +271,7 @@ Do not silently overwrite old claims. Flag and let the user decide.
 ## What Not to Do
 
 - **Source files under `.raw/` are immutable.** Do not modify the files that users drop there (articles, transcripts, images). The `.raw/.manifest.json` delta tracker and its `address_map` (DragonScale Mechanism 2) are the only files under `.raw/` that `wiki-ingest` itself maintains. Treat every other file under `.raw/` as read-only source content.
+- Do not create or update a concept page without matching `source_refs` for every `sources` entry.
 - Do not create duplicate pages. Always check the index and search before creating.
 - Do not skip the log entry. Every ingest must be recorded.
 - Do not skip the hot cache update. It is what keeps future sessions fast.