What problem does it solve? Working through a round of code review feedback by hand means reading every thread, deciding whether to fix or push back, writing tested fixes, and replying on each thread individually. This Skill automates that entire loop on GitLab merge requests and GitHub pull requests while keeping the human reviewer in control of disagreements. ## Core Features & Use Cases - Thread triage and classification: Reads every open review thread and classifies it as already fixed, needs fix, or disagree, then acts accordingly. - Tested, atomic fixes: Implements each agreed fix as its own commit with a failing test first, runs the full test suite, and pushes only to the source branch. - Host adapters for GitLab and GitHub: Ships adapter references covering authentication, thread listing, threaded replies, and thread resolution via glab and gh, with graceful degradation to other hosts through tool discovery. - Use Case: A reviewer leaves twelve comments on your GitHub PR. Invoke the command with the PR URL, and each thread gets a tested fix commit, an in-thread reply citing the commit SHA, and a resolution, while threads you disagree with get a reasoned reply and stay open. ## Quick Start Ask the AI to address all open review comments on your merge request or pull request by providing its URL, for example by invoking /address-pr-comments with the MR or PR link.