What problem does it solve? A fully-covered test suite can still be vacuous: assertions that pass for the wrong reason, tautological checks, or tests that only exercise the happy path. This is especially risky when the same LLM wrote both the implementation and its tests. This Skill adversarially audits whether a unit's tests can actually fail before you declare the work done. ## Core Features & Use Cases - Mutation probe: Introduces deliberate, realistic bugs (negated guards, off-by-one errors, dropped steps) and confirms at least one test goes red for each, recording surviving mutants as coverage holes. - Tautology and narrowing probe: Reads each assertion for weaknesses such as existence-only checks, over-narrowed inputs, can't-fail properties, and assertions on mocks. - Structured report: Emits a markdown report listing surviving mutants, weak assertions, and clean assertions, routing findings back to the human or to test-strengthening workflows without rewriting tests itself. - Use Case: After an AI assistant generates both a Go module and its tests, run this audit to verify the tests genuinely constrain behavior before proposing the unit for merge. ## Quick Start Ask the assistant to run a vacuity audit on the unit you just finished to confirm its tests would actually catch real bugs.
