Fix CI

Use this skill when automation disagrees with local assumptions or blocks integration.

Trigger Conditions

Use this skill when:

• CI is red
• lint, test, build, or type-check fails in the pipeline
• the branch passes locally but fails remotely
• logs are noisy and the earliest failure needs to be extracted
• the user wants the minimal change that unblocks the pipeline

Do not use this skill for planning-only tasks or for broad product design questions.

Workflow

1. Read the earliest meaningful failure first.
2. Classify the failing stage: lint, type, unit test, integration test, build, packaging, environment.
3. Separate root-cause failure from downstream noise.
4. Check whether the failure is deterministic, flaky, or environment-specific.
5. Recommend the smallest repair with the highest confidence.
6. Re-run the narrowest useful verification.

Output

For non-trivial work, provide:

• failing stage
• first actionable error
• likely cause
• minimal repair path
• verification to re-run

Coordination

After CI diagnosis:

• hand off to the repo's coding workflow for code fixes
• use systematic-debugging if root cause is still unclear
• use ship-readiness once pipeline health is restored

Invocation Examples

• "Use fix-ci to inspect this pipeline failure and find the first actionable error."
• "Use fix-ci because local tests pass but CI is still red."
• "Use fix-ci and tell me the smallest change that should unblock the branch."