Before you commit that multi-step refactor or complex feature implementation, this skill runs a structured QA pass on your AI agent's work. It walks through six phases: gathering context from requirements and CLAUDE.md files, verifying you actually hit all the acceptance criteria, checking file paths and syntax, hunting for logic bugs and rule violations like console.log or any types, confirming everything on your todo list got done, and evaluating whether the solution makes sense. The honest take is that AI agents are surprisingly good at introducing subtle bugs or drifting from requirements when tasks get complex, and this catches that drift before it becomes technical debt. Run it when someone says "check your work" or proactively after anything over five steps.
npx skills add https://github.com/shipshitdev/library --skill qa-reviewer