What problem does it solve? When a pull request fails CI, developers often guess at fixes without reading the actual error logs, waste effort fixing flaky tests, or get stuck trying to access external CI systems. This Skill provides a systematic workflow for identifying, diagnosing, and fixing failing PR checks using the GitHub CLI. ## Core Features & Use Cases - Log-Driven Diagnosis: Pulls failed job logs via gh run view --log-failed and extracts the 20-50 lines around each failure before proposing any fix. - Flaky-vs-Real Triage: Reruns failed jobs on the same commit and inspects run history to distinguish flaky tests from genuine regressions, then scopes the breaking commit with git log or git bisect. - Scope Boundaries: Handles GitHub Actions failures directly while correctly flagging external CI systems (Buildkite, CircleCI) as out of scope, requesting logs from the user instead. - Use Case: A PR shows three failing checks. Use this Skill to identify which are GitHub Actions jobs, pull their logs, confirm one failure is flaky via rerun, scope the real regression to a specific commit, and verify the fix with lacquer wait pr. ## Quick Start Ask the AI to diagnose why the checks on my current pull request are failing and propose a fix based on the actual CI logs.
