What problem does it solve? Maintaining a personal Guix channel with a weekly auto-update GitHub Actions workflow produces recurring CI failures — sandboxed git-fetch errors, GitHub API rate limits, stale test references after package removal, and docs drift — that are hard to triage without knowing which failures need code fixes versus simple reruns. ## Core Features & Use Cases - CI failure triage: Classify Auto Update Packages workflow runs into infrastructure failures (rerun only), real build failures (fix packages), and test/script desync failures (fix stale references), using gh CLI run timing and failed-step inspection. - Known failure-mode playbooks: Documented root causes and proven fixes for git-fetch sandbox EACCES (convert to url-fetch), masked downstream bugs, url-fetch tarball extraction in build phases, GitHub API rate limits, and zig version chain breakage in emacs-ghostel. - Package maintenance workflows: Recipes for adding packages, wrapping existing packages with langpack resources, lint-fix recipes for wrapper-inputs and description checkers, and cron auto-fix boundaries (what may be committed vs. left to the user). - Use Case: When the jeans channel CI fails overnight, classify the run in under a minute, rerun infrastructure failures directly, or follow the issue-fixer flow to patch the failing package and verify locally before handing the commit to the user. ## Quick Start Ask the assistant to check the latest Auto Update Packages run on the jeans repository and determine whether the failure needs a rerun or a package fix.
