What problem does it solve? Manually debugging failing CI checks on a pull request is slow and repetitive: you read logs, reproduce failures locally, push a fix, and wait again. This Skill automates that loop by detecting failed checks, reproducing them locally, committing fixes, and polling until every check passes. ## Core Features & Use Cases - Automated failure diagnosis: Reads failed GitHub Actions logs via gh run view and maps each failure type (fmt, clippy, build, tests, fixtures, docs) to the exact local reproduction command. - Fix-push-wait cycle: Commits fixes with Conventional Commits, pushes, and polls CI status with a dedicated wait-ci.sh script that reports ALL_PASSED, FAILED, TIMEOUT, or API_ERROR. - Bounded retry loop: Repeats the fix cycle up to 5 times and reports status to the user if the limit is exceeded, never bypassing pre-commit hooks. - Use Case: A Rust project PR fails clippy and test checks. Invoke the Skill with the PR number; it reproduces the failures with cargo clippy and cargo test, applies fixes, pushes, and waits until all checks turn green. ## Quick Start Ask the assistant to fix the failing CI on pull request 42 and keep iterating until every check passes.
