CI Failure Resolution and Pre-commit Infrastructure Maintenance

Rule

When CI is red on master or pre-commit scripts are corrupted, diagnose systematically before fixing: check failure pattern, run locally, fix root cause, verify.

Context

When pre-commit CI failures block commits or indicate infrastructure degradation. Not for normal pre-commit runs that flag code issues — those are expected workflow.

Detection

• Pre-commit workflow runs failing repeatedly on master (not just a single commit)
• prek run --all-files fails on a clean checkout
• Corrupted or missing pre-commit scripts after a bad merge/rebase
• Multiple agents failing on the same hook

Pattern

# 1. Diagnose locally
prek run --all-files

# 2. Fix common issues
make format          # formatting
make typecheck       # typing

# 3. Restore corrupted scripts from last good commit
git log --oneline -- scripts/precommit/
git show <commit>:scripts/precommit/script.py > scripts/precommit/script.py

# 4. If hooks keep failing after fixes
prek clean           # clear cache
prek install --force # reinstall hooks

# 5. Verify all clean
prek run --all-files

Outcome

• Master branch stays green
• No blocking CI issues for subsequent sessions
• Infrastructure integrity maintained

Related

• Git Workflow — Commit practices