What problem does it solve?
It helps you turn noisy, automated CodeRabbit review output into a clear set of actionable triage decisions for a GitHub pull request, including optional fixes and replies.
Core Features & Use Cases
- PR-aware triage: Identifies the target PR from a PR URL, owner/repo#number, a bare PR number resolved via
gh, or the current branch.
- Findings parsing across sources: Extracts actionable items from inline review comments, review-body nitpick sections, and summary/walkthrough actionable findings while skipping walkthrough noise and duplicates.
- Skeptical vetting and safe actioning: Categorizes each finding as AUTO-APPLY, REVIEW, or DROPPED, and supports batch JSON output when
--batch is provided.
- Context verification: Fetches PR diff and reads full affected files to validate that the proposed change is genuinely correct and safe.
- User confirmation and structured execution: In interactive mode, you confirm what to apply; the skill then applies edits and posts appropriate replies.
Quick Start
Run the coderabbit skill with your PR identifier (for example, a PR URL or owner/repo#number) to produce a categorized triage table and suggested diffs for the valid findings.