Debug failing CI jobs from a prompt

[structbinary]

Give CI-Copilot the ability to diagnose and help fix failing CI jobs. A user describes the problem (or pastes a job URL/log), and CI-Copilot analyzes the failure and suggests or applies a fix.

Examples

• "Our CI build is failing on the lint step — here's the log"
• "This GitHub Actions workflow keeps timing out on the test job"
• "Why is our Docker build failing with exit code 1?"

What needs to happen

• Accept CI job logs or URLs as input
• Parse and classify failure types (dependency issues, test failures, config errors, timeout, permissions)
• Correlate failures with the workflow definition
• Suggest targeted fixes with explanations
• Optionally apply the fix and open a PR (with approval)

Dependencies

• Requires read access to CI job logs (GitHub Actions API, Jenkins API, etc.)
• May benefit from the CI knowledge base (Issue: CI platform knowledge base)