What problem does it solve? Addressing PR review feedback is tedious: you must read every comment, decide which are valid, write fixes, reply to each thread, and resolve them one by one. This Skill automates that entire loop, evaluating every unresolved review thread on a GitHub pull request, fixing valid feedback in parallel, and replying to and resolving threads via the GitHub GraphQL API. ## Core Features & Use Cases - Full and Targeted Modes: Process all unresolved threads on a PR, or address a single comment URL without touching anything else. - Parallel Resolver Agents: Spawn one agent per thread (batched with file-conflict avoidance) to evaluate validity, implement fixes, and return structured verdicts such as fixed, declined, or needs-human. - Cross-Invocation Cluster Analysis: Detect recurring feedback patterns across review rounds and dispatch a single agent to investigate the broader area holistically. - GraphQL Automation: Bundled scripts fetch unresolved threads, reply within threads, and resolve them, with pagination and bot-comment filtering built in. - Use Case: A reviewer leaves 12 comments on your PR overnight. Run the Skill with no arguments; it triages the threads, fixes the valid ones in parallel, runs the project test suite once, commits and pushes, then replies to and resolves each thread with quoted context. ## Quick Start Ask the AI to resolve all review feedback on the current branch's pull request, or pass a PR number or a specific comment URL to target one thread.
