fix-ci

Diagnoses failing GitHub PR checks, reproduces failures locally, fixes code, and pushes verified commits.

3|Updated Feb 15, 2025
One-click install
npx skills add https://github.com/mark-torres10/ai_tools --skill fix-ci-mark-torres10
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: fix-ci
Source: https://github.com/mark-torres10/ai_tools/tree/main/skills/fix-ci
Command: npx skills add https://github.com/mark-torres10/ai_tools --skill fix-ci-mark-torres10

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Red CI checks on a GitHub pull request block merges and require tedious manual investigation across logs, local environments, and commits. This Skill automates the full loop: identifying the PR, reading failing checks, reproducing failures locally, applying scoped fixes, pushing, and re-verifying CI. ## Core Features & Use Cases - PR and Failure Triage: Resolves the PR from a URL or the current branch using gh, groups failing checks by root cause, and classifies them as in-scope, stale-base, or out-of-scope. - Local Reproduction and Fix: Reads workflow files and repo configs to run the same commands CI runs, then applies minimal root-cause fixes without weakening checks or touching unrelated code. - Commit, Push, and Re-verify: Commits only the relevant files with a clear message, pushes to the PR head branch, watches checks with gh pr checks --watch, and reports a structured summary. - Use Case: A teammate's PR shows red GitHub Actions jobs for lint and unit tests. Invoke this Skill to pull the failing logs, reproduce the failures locally, fix the offending test and lint errors, push the commit, and confirm the checks turn green. ## Quick Start Fix the failing CI checks on the pull request for my current branch and push the fix.

Frequently Asked Questions about fix-ci

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I fix failing GitHub Actions checks on a pull request?

Use gh pr checks to list failing checks, pull logs with gh run view --log-failed, then reproduce the same commands locally from the workflow file. Fix the root cause, commit only the relevant files, push to the PR head branch, and re-watch checks.

How to reproduce CI failures locally before pushing a fix?

Read the failing job's run steps in .github/workflows/*.yml and check repo docs like Makefile, package.json scripts, or tox.ini. Run the same test paths, lint targets, and build flags locally until the previously failing steps pass.

Can this fix CI failures caused by a stale base branch?

Yes. Failures already present on main are classified as stale-base, and the workflow merges or rebases the base branch first, then re-runs checks. It does not hack around failures that originate outside the PR.

What CI failures should not be fixed by changing code?

Flaky infrastructure, broken main, missing secrets, or unrelated services are out of scope and should be reported rather than patched. The workflow never weakens or skips checks in workflow YAML just to make red checks green.

Does the fix-ci workflow push commits automatically?

Yes, invoking it implies committing and pushing unless the user says investigate-only. It stages only files related to the CI fix, writes a why-focused commit message, and never force-pushes, amends, or merges without explicit request.