This takes a methodical approach to fixing broken test suites by grouping failures intelligently before diving in. Instead of fixing tests one by one, it categorizes them by error type, affected modules, and root cause, then tackles the highest impact groups first. Infrastructure issues like imports get fixed before API changes, which get fixed before logic bugs. After each group, it runs a focused subset of tests to verify the fix before moving on. The systematic grouping is the smart part here. When you've got 15 failures after a refactor and they're all tangled up, having a process that says "these 8 are all the same import issue" prevents you from thrashing around fixing symptoms instead of causes.
npx skills add https://github.com/mhattingpete/claude-skills-marketplace --skill test-fixing