What problem does it solve? When a CI check fails on a pull request, the natural instinct is to fix it immediately and push. But if multiple checks fail, this creates a serial diagnose-fix-push loop where each failure triggers its own push cycle, wasting CI minutes and review time. This Skill enforces a batching protocol: wait for the full check run to complete, collect every failure, fix them all together, and push once. ## Core Features & Use Cases - Failure Collection Protocol: Uses gh pr checks and gh run view --log-failed to verify all jobs finished and build a complete failure ledger before any fix attempt. - Clustered Fixing: Groups failures by root cause so related issues are resolved in a single changeset verified locally. - Push Discipline: Delegates commit and push to the commit-pr skill, mandating --force-with-lease over blocked bare force pushes and preferring a new fix commit over amending pushed work. - Use Case: A PR fails on formatting, two stale assertions, and an integration test. Instead of six push cycles, you wait for the run to finish, fix all four issues locally, and push once. ## Quick Start When a PR's CI fails, wait for all checks to complete, collect every failure log with gh, fix them all together, and push a single fix commit.
