What problem does it solve? Handling pull request review feedback is repetitive: reading each comment, judging whether it is valid, editing code, committing, pushing, replying, and resolving threads. This Skill automates that entire loop while distinguishing AI reviewers from human reviewers so the right level of caution is applied. ## Core Features & Use Cases - Automated review triage: Fetches PR review comments with the GitHub CLI, filters out pending reviews, self-comments, and already-answered threads, then classifies each comment as a valid fix or a reply-only rejection. - Reviewer-aware workflow: Bot reviewers (Copilot, CodeRabbit) and self-authored reviews are handled autonomously and their threads resolved via GraphQL, while any human reviewer comment pauses the flow for user confirmation. - End-to-end execution: Applies suggestion blocks verbatim, commits and pushes fixes, posts replies in the review's language, and prints a summary of fixes, replies, and resolved threads. - Use Case: After opening a PR that received four Copilot comments and one human comment, run the skill to fix the two valid findings, reject the over-defensive suggestions with stated reasoning, and resolve only the bot threads. ## Quick Start Ask the agent to resolve the review comments on PR 42, for example by saying "resolve the review comments on pull request 42".