What problem does it solve? Writing a QA test plan from scratch often produces vague coverage claims, combinatorial case explosions, or untestable exit criteria. This Skill guides the authoring of a test plan where every case traces to a real upstream behavior, the catalog is sized by risk rather than the input cross-product, and done-criteria are mechanically decidable. ## Core Features & Use Cases - Coverage-map-first method: Derives every test case from a feature-spec behavior/acceptance criterion or an api-spec operation/error, so no behavior is left untested and no case is invented. - Risk-weighted case design: Applies BVA, equivalence partitioning, decision tables, state-transition testing, and pairwise selection so high-risk areas get depth without combinatorial blow-up. - Non-functional and ML testing guidance: Carries numeric targets for performance, security, and WCAG 2.2 AA accessibility, plus metric-threshold-on-a-named-dataset cases for ML/probabilistic behavior. - Versioned amends: Handles change requests as scoped deltas with impact+risk regression selection instead of full rewrites. - Use Case: Given an approved feature-spec and api-spec for a login service, produce a test plan with a coverage matrix, executable cases (including the exact 401/429 error expectations), testable exit criteria tied to an open-defect threshold, and a performance target from the NFRs. ## Quick Start Use authoring-test-plan to turn docs/specs/auth-v2-feature-spec.md and the api-spec into a complete test plan with a traceable, risk-weighted case catalog.
