What problem does it solve? Addressing GitHub Pull Request review comments one by one is tedious and error-prone: you must find every unresolved thread, decide what to fix, commit changes, reply to reviewers, and resolve threads. This Skill automates that entire loop in a structured 7-phase workflow so no review comment is missed. ## Core Features & Use Cases - Unresolved Thread Collection: Uses the GitHub GraphQL API via gh CLI to fetch only unresolved (isResolved=false) review threads, with pagination support for large PRs and correct handling of fork-based PRs. - Classification & Planning: Classifies each comment by priority (required / recommended / question) and category (architecture, performance, tests, etc.), then builds a sorted fix plan persisted to a mapping JSON file so work can resume after context loss. - Safe Reply & Resolve Automation: A guarded script replies only to entries with verified commit SHAs and resolves threads, with dry-run support and skip logic that prevents accidental or duplicate replies. - Use Case: A reviewer leaves 15 comments on your PR. Run this Skill to gather all unresolved threads, get a prioritized fix plan, implement and push the fixes, then automatically reply with commit SHAs and resolve each thread. ## Quick Start Ask the AI to address all unresolved review comments on the current branch's pull request using the pr-review-fix skill.