Adds AI_PIPELINE_HARDENING_PLAN.md -- a phased, self-contained plan an AI agent can execute end-to-end to harden the Anthropic integration. Six phases: central LLM client + robust parsing -> prompt caching & tool-based structured outputs -> extraction quality (overlap chunking, stateful IDs, adaptive throttle, unified relation vocab) -> quiz quality (Fisher-Yates, debiased correctIndex, difficulty field, dedup) -> R42 retrieval & telemetry (TF-IDF, lookup_topic tool, llm_calls collection) -> eval harness.
Renames src/lib/giteaService.js -> src/lib/githubService.js. The module actually calls api.github.com / raw.githubusercontent.com; the previous name was misleading. Updates the single import site in src/components/admin/KnowledgeGraph.jsx. No behaviour change.
Test plan
npm run lint clean
npm run build succeeds
Admin > Knowledge Graph > Sync Handbook still pulls files from respellion/employee-handbook on GitHub
Plan doc renders correctly on the Gitea repo browser
Generated with Claude Code.
## Summary
- Adds `AI_PIPELINE_HARDENING_PLAN.md` -- a phased, self-contained plan an AI agent can execute end-to-end to harden the Anthropic integration. Six phases: central LLM client + robust parsing -> prompt caching & tool-based structured outputs -> extraction quality (overlap chunking, stateful IDs, adaptive throttle, unified relation vocab) -> quiz quality (Fisher-Yates, debiased correctIndex, difficulty field, dedup) -> R42 retrieval & telemetry (TF-IDF, `lookup_topic` tool, `llm_calls` collection) -> eval harness.
- Renames `src/lib/giteaService.js` -> `src/lib/githubService.js`. The module actually calls `api.github.com` / `raw.githubusercontent.com`; the previous name was misleading. Updates the single import site in `src/components/admin/KnowledgeGraph.jsx`. No behaviour change.
## Test plan
- [ ] `npm run lint` clean
- [ ] `npm run build` succeeds
- [ ] Admin > Knowledge Graph > Sync Handbook still pulls files from `respellion/employee-handbook` on GitHub
- [ ] Plan doc renders correctly on the Gitea repo browser
Generated with Claude Code.
Adds AI_PIPELINE_HARDENING_PLAN.md — a phased, self-contained plan an AI
agent can execute to harden the Anthropic integration (central LLM
client, tool-based structured outputs, prompt caching, retrieval-based
R42 context, eval harness).
Renames src/lib/giteaService.js to src/lib/githubService.js. The module
calls api.github.com and raw.githubusercontent.com; the previous name
was misleading. No behaviour change. Updates the single import site in
src/components/admin/KnowledgeGraph.jsx.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
rve
changed title from docs: AI pipeline hardening plan + rename giteaService ? githubService to docs: AI pipeline hardening plan + rename giteaService -> githubService2026-05-20 10:08:13 +00:00
rve
merged commit d07d15b2a7 into main2026-05-20 10:09:40 +00:00
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Summary
AI_PIPELINE_HARDENING_PLAN.md-- a phased, self-contained plan an AI agent can execute end-to-end to harden the Anthropic integration. Six phases: central LLM client + robust parsing -> prompt caching & tool-based structured outputs -> extraction quality (overlap chunking, stateful IDs, adaptive throttle, unified relation vocab) -> quiz quality (Fisher-Yates, debiased correctIndex, difficulty field, dedup) -> R42 retrieval & telemetry (TF-IDF,lookup_topictool,llm_callscollection) -> eval harness.src/lib/giteaService.js->src/lib/githubService.js. The module actually callsapi.github.com/raw.githubusercontent.com; the previous name was misleading. Updates the single import site insrc/components/admin/KnowledgeGraph.jsx. No behaviour change.Test plan
npm run lintcleannpm run buildsucceedsrespellion/employee-handbookon GitHubGenerated with Claude Code.
docs: AI pipeline hardening plan + rename giteaService ? githubServiceto docs: AI pipeline hardening plan + rename giteaService -> githubService