qa-scenarios

Generate user-impact QA test scenarios from a code change, in the style of a human QA tester probing real failure states. Use this whenever the user wants QA help on a commit, diff, branch, or PR; asks "what should I test here", "what would break", "what would a user hit"; wants to bootstrap or grow a failure-mode corpus for a repo; or is setting up per-commit / per-PR automated QA. Trigger even if the user does not say "QA" or "test" explicitly, as long as they are asking what could go wrong from a user's perspective for a code change. This skill is product-agnostic and reads the repo it runs in to learn the domain, so it works for any codebase.