rule-extractor

To extract coding rules from merged PR review findings, you identify HIGH or CRITICAL bug-class findings and transform them into structured, reusable rule files with frontmatter. The skill automates classification, deduplication, and drafting for human-reviewed PR proposals.

Closed-loop learning for coding rules converts verified post-merge bug-class findings into formal rules. Agents reference these rules at write-time to avoid repeating anti-patterns, ensuring structural bug classes go extinct across repositories.

Deduplicate new bug-class rules against existing project rules by checking extracted findings against the current rules ecosystem. The skill automatically compares structural patterns to prevent duplicate entries before generating a draft rule proposal.

Yes, you can automate rule proposals for post-merge PR workflows across multiple repositories. The skill processes merged PR artifacts, applies automated classification, and drafts structured rule proposals with traceability frontmatter for human review.

The best way to convert structural bug patterns into formal coding rules is to extract verified findings into a structured draft with frontmatter. This ensures traceability and requires human review before merging into the rules ecosystem.

Yes, human review is required before merging automated rule proposals. The skill generates structured drafts with frontmatter from bug-class findings, but a human must review the PR proposal before it merges into the project rules.