PR Review & Experience Learning
Performs evidence-driven GitHub pull request reviews, assesses merge readiness, and distills accepted code feedback into reusable project rules.
The PR Review & Experience Learning skill equips Claude Code to conduct rigorous, read-only code reviews and re-reviews on GitHub pull requests using repository guidelines. Beyond surfacing prioritized findings (P0–P3) and evaluating delta changes between review rounds, it analyzes accepted reviewer feedback to distill lasting, repository-scoped engineering standards. It enforces safe evidence collection, treats PR contents as untrusted data, verifies validation trails, and facilitates continuous engineering learning without blindly auto-approving or modifying contributor branches.
主要功能
01Continuous feedback learning to synthesize accepted review comments into project rules
02Explicit gating requiring clear confirmation when architectural intent or acceptance criteria diverge
03Delta-based re-reviews tracking specific thread resolutions and preventing reverted fixes
04Safe, read-only evidence collection treating pull request inputs as untrusted data
05Evidence-driven PR review and re-review with strict P0–P3 prioritized finding categorizations
0620 GitHub stars
使用场景
01Conducting in-depth initial reviews or delta re-reviews on complex GitHub pull requests
02Assessing merge risk and validation coverage before finalizing a pull request landing decision
03Extracting recurring engineering rules and architectural standards from accepted review comments