What problem does it solve?
Manually triaging reviewer feedback on a pull request is tedious: inline review threads, general PR comments, and bot suggestions are scattered across multiple GitHub API surfaces, and it is easy to miss actionable items or waste effort on false positives. This Skill automates the full loop of fetching, analyzing, fixing, and replying to PR review feedback in the streamlit/streamlit repository.
Core Features & Use Cases
- Comprehensive comment retrieval: Pulls inline review comments, general PR (issue) comments, and unresolved review threads via the gh CLI REST and GraphQL APIs, with pagination support.
- Intelligent triage: Categorizes feedback using conventional comment types (issue, todo, chore, suggestion, nitpick, question), skips status-only bot messages, and validates bot suggestions before acting on them.
- Guided fix workflow: Presents a numbered list of actionable items, applies minimal fixes after user selection, flags high-complexity changes, and generates reply text plus thread-resolution commands.
- Use Case: A reviewer leaves 8 comments on your PR, including two AI bot suggestions. The Skill fetches all feedback, identifies one bot comment as a false positive, applies fixes to the 5 actionable items, and drafts replies and resolve-thread commands for each.
Quick Start
Ask the assistant to address all review comments on the PR for the current branch using the gh CLI.