fleet-greening-ci

To automate fixing failing GitHub Actions CI runs, this skill watches the repository's workflow logs, applies local code fixes, commits, and pushes until the run succeeds. It uses budgeted polling and log-tail analysis to detect failures and trigger the fix-and-push loop automatically.

A fix-and-push loop for keeping CI green is an automated process that monitors GitHub Actions runs, surfaces the first failure log, applies a local code fix, and pushes the commit to re-dispatch the workflow. It repeats this cycle until the CI run lands green or a wall-clock budget expires.

Yes, you can use this to monitor matrix builds in GitHub Actions. The skill supports single jobs and matrix builds across modes like release for build-server matrices with fail-fast polling, and cool for confirming the rest of a matrix after a freshly-pushed fix.

Yes, you need a working git checkout of the target repo and a Node.js runtime to run automated CI fixes. You also need appropriate permissions to read CI logs, edit code, commit, and push changes back to the repository.

The best way to handle a red main branch in GitHub Actions is using the fast mode, which targets the ci.yml workflow to automatically apply fixes locally, commit, and push until the run succeeds. This mode is specifically designed for quickly responding to red main branches.

When the wall-clock budget expires during CI monitoring, the automated fix-and-push loop stops attempting further fixes. The skill uses budgeted polling to ensure it only attempts to drive the GitHub Actions run back to green within a predefined time limit.