[PR #1075] [MERGED] fix(metadata): stop schema changes churning generated skill evidence #2624

New issue

Closed

opened 2026-10-17 21:08:20 +00:00 by bsakaguchi · 0 comments

bsakaguchi commented

2026-10-17 21:08:20 +00:00

Owner

📋 Pull Request Information

Original PR: https://github.com/morluto/rea/pull/1075
Author: @N0zoM1z0
Created: 10/8/2026
Status: ✅ Merged
Merged: 10/8/2026
Merged by: @N0zoM1z0

Base: main ← Head: fix/generated-metadata-churn

📝 Commits (2)

8e8a0d5 fix(metadata): stop schema changes churning generated skill evidence
b2e693f fix(metadata): reconcile generated facts with current main

📊 Changes

11 files changed (+98 additions, -18 deletions)

View changed files

📝 CONTRIBUTING.md (+9 -0)
📝 docs/product-catalog.json (+0 -3)
📝 docs/verification/managed-conformance-manifest.json (+2 -2)
📝 scripts/generate-skill-metadata.mjs (+6 -7)
📝 scripts/lib/product-catalog.mjs (+1 -1)
📝 skills/reverse-engineer-anything/SKILL.md (+0 -1)
📝 src/application/Doctor.fixture.ts (+2 -1)
📝 src/application/Doctor.test.ts (+1 -0)
📝 src/application/Doctor.ts (+7 -2)
📝 tests/boundary/cli/productCatalog.test.ts (+5 -0)
📝 tests/boundary/filesystem/setupSkill.test.ts (+65 -1)

📄 Description

Schema-only bug fixes currently rewrite the same global digest in three tracked files: the product catalog, skill frontmatter, and the managed conformance manifest. Independent fixes therefore conflict even when their documented facts and skill instructions are unchanged.

Keep full schema-sensitive runtime identity in doctor and binary_session. Remove those aggregate hashes from the checked-in facts projection and skill frontmatter; retain provider identity and the manifest's exact skill-file commitment. Doctor now compares every installed managed instruction/reference file with the packaged canonical bundle, including missing files and local modifications, and still reports observed legacy metadata.

Verification: 41 focused doctor, product-catalog, skill-installation, and MCP identity tests passed; the expanded installation regression passed after integrating current main. Two actual compiled-schema changes (a length bound and description) each changed runtime identity while leaving all three generated files byte-identical. npm run check, npm run docs:check, and npm pack --dry-run passed. npm run verify:package passed under an external owned TMPDIR: 139 tools, packaged CLI/MCP, setup backups/idempotence, and canonical skill/reference installation verified. No real Hopper/Ghidra behavior changed.

_{🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.}

## 📋 Pull Request Information **Original PR:** https://github.com/morluto/rea/pull/1075 **Author:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Created:** 10/8/2026 **Status:** ✅ Merged **Merged:** 10/8/2026 **Merged by:** [@N0zoM1z0](https://github.com/N0zoM1z0) **Base:** `main` ← **Head:** `fix/generated-metadata-churn` --- ### 📝 Commits (2) - [`8e8a0d5`](https://github.com/morluto/rea/commit/8e8a0d50715ea4791002c106745aeb8297d89bcc) fix(metadata): stop schema changes churning generated skill evidence - [`b2e693f`](https://github.com/morluto/rea/commit/b2e693f3cf71743473dbac15438382c3215b2593) fix(metadata): reconcile generated facts with current main ### 📊 Changes **11 files changed** (+98 additions, -18 deletions) <details> <summary>View changed files</summary> 📝 `CONTRIBUTING.md` (+9 -0) 📝 `docs/product-catalog.json` (+0 -3) 📝 `docs/verification/managed-conformance-manifest.json` (+2 -2) 📝 `scripts/generate-skill-metadata.mjs` (+6 -7) 📝 `scripts/lib/product-catalog.mjs` (+1 -1) 📝 `skills/reverse-engineer-anything/SKILL.md` (+0 -1) 📝 `src/application/Doctor.fixture.ts` (+2 -1) 📝 `src/application/Doctor.test.ts` (+1 -0) 📝 `src/application/Doctor.ts` (+7 -2) 📝 `tests/boundary/cli/productCatalog.test.ts` (+5 -0) 📝 `tests/boundary/filesystem/setupSkill.test.ts` (+65 -1) </details> ### 📄 Description Schema-only bug fixes currently rewrite the same global digest in three tracked files: the product catalog, skill frontmatter, and the managed conformance manifest. Independent fixes therefore conflict even when their documented facts and skill instructions are unchanged. Keep full schema-sensitive runtime identity in doctor and `binary_session`. Remove those aggregate hashes from the checked-in facts projection and skill frontmatter; retain provider identity and the manifest's exact skill-file commitment. Doctor now compares every installed managed instruction/reference file with the packaged canonical bundle, including missing files and local modifications, and still reports observed legacy metadata. Verification: 41 focused doctor, product-catalog, skill-installation, and MCP identity tests passed; the expanded installation regression passed after integrating current main. Two actual compiled-schema changes (a length bound and description) each changed runtime identity while leaving all three generated files byte-identical. `npm run check`, `npm run docs:check`, and `npm pack --dry-run` passed. `npm run verify:package` passed under an external owned TMPDIR: 139 tools, packaged CLI/MCP, setup backups/idempotence, and canonical skill/reference installation verified. No real Hopper/Ghidra behavior changed. --- <sub>🔄 This issue represents a GitHub Pull Request. It cannot be merged through Gitea due to API limitations.</sub>

bsakaguchi

2026-10-17 21:08:20 +00:00