What problem does it solve? AI-authored pull requests churn through repeated review rounds when fixes are pushed comment-by-comment, each push re-triggering external reviewers and CI. This Skill closes each review round in a single batched, self-reviewed push so the PR converges in as few external rounds as possible. ## Core Features & Use Cases - Batched fix pass: Collects every open finding across all threads and reviewers, fixes them in one pass using verify-then-fix discipline, and pushes exactly once. - Adversarial internal self-review: Dispatches internal reviewers against the current diff to hunt the same failure classes already found in the PR plus regressions introduced by the fixes themselves. - Round-level measurement: Re-checks branch cohesion and reports cumulative commits and LOC delta versus base every round, catching PRs that quietly grow past size gates across sessions. - Use Case: A reviewer leaves 8 comments on an open PR. Instead of 8 push-and-reply cycles, you batch all findings, fix and verify them, run an internal adversarial review until clean, push once, and only then re-trigger the external bot. ## Quick Start After addressing review feedback on my pull request, run the closing-review-loops process to batch the fixes, self-review the diff, and push once before re-triggering the external reviewer.
