Skill: Fix CI

What This Skill Does

Provides a structured debugging workflow for CI failures. Instead of the common pattern of "read error → guess fix → push → wait → repeat", this skill enforces: analyze logs → identify root cause → reproduce locally → fix → verify locally → push once.

When to Use

• When CI checks fail on a PR or push
• When the user says "CI is red" or "fix the pipeline"
• After pr-ready detects CI failures

Do NOT use this for writing new CI pipelines — use ci-setup for that.

Execution Model

• Always: the primary agent runs this skill directly.
• Rationale: CI fixes require terminal access (running tests locally), file edits, and iterative debugging — all primary agent capabilities.
• Token budget: ~3-8k tokens depending on failure complexity.