amirbena / amirbena/code-review-skill
Update the Wiki after Measurement and Long-Term Intelligence lands
- Dominant language
- Python
- Stars
- 7
- Forks
- 2
- Avg merge
- 27m
- Merged PRs (30d)
- 188
Description
## Problem
Tier 4 (#182 → #131 → #130) introduces the project's measurement and long-term-intelligence layer. The Wiki should only be updated once that work has actually landed, documenting real behavior rather than roadmap intent.
## Goal
Synchronize the Wiki with the final implemented measurement, analytics, and repository-scoped learning architecture delivered by Tier 4, using the final implementation and canonical policies as source of truth.
## Scope
- Inspect and update only Wiki pages materially affected: Home, How the Review Model Works, Architecture, Development Workflow, Context and Evidence, Documentation Map, AI Code Review Landscape, and any pages describing observability/evaluation/repository knowledge.
- Explain the final telemetry model (#182): observes execution, never determines findings/severity/decision.
- Document the implemented analytics layer (#131): what signals can be analyzed, what questions they answer, interpretation limitations — without implying a metric proves review quality on its own.
- Document the final learning model (#130): what's learned, how accepted/rejected findings contribute, confidence/decay, auditability, and the boundary preventing learned knowledge from becoming unquestionable truth.
- Clearly distinguish review-time evidence vs. historical measurement vs. learned repository knowledge throughout.
- Update the architecture view to reflect the landed system as layered (evidence → reasoning → findings/decision → telemetry → analytics → learning), using the actual implementation rather than forcing a fixed diagram.
- Revisit the Landscape page only where Tier 4 changes positioning; refresh stale competitor research if needed and record a new snapshot date.
## Non-goals
- Review-engine redesign or Tier 4 implementation work — this issue is documentation-only.
- Speculative Tier 5 capabilities presented as implemented.
- Unrelated Wiki rewrite.
## Dependencies
Blocked by #182 → #131 → #130 (all open — verify still open before starting this work).
## Acceptance criteria
- Wiki accurately reflects the final Tier 4 implementation; canonical policies remain the normative source (no duplicated contracts in the Wiki).
- Telemetry, analytics, and learned knowledge are each clearly separated from review decision logic and from each other.
- Architecture/navigation pages reflect the complete Tier 4 system with clear trust boundaries.
- Landscape positioning updated only where Tier 4 materially changes it.
- No Tier 5 capability presented as already implemented.
Contributor guide
Assessment
This issue has not been assessed yet.