What problem does it solve? Adding or changing a public SDK API requires consistent end-to-end test coverage, but writing those tests by hand often leads to weak assertions, missing error paths, wrong executor placement, and unverified smoke tags. This Skill enforces a structured workflow so every public API change ships with happy, sad, and error coverage that is validated locally before landing. ## Core Features & Use Cases - Test planning and scaffolding: Investigates the changed SDK surface, finds comparable existing tests, and drafts concrete test definitions plus executors in packages/sdk/tests-qvac. - Deterministic model-output assertions: Selects the strongest achievable validation strategy (contains-all, contains-any, regex, numeric-range, throws-error) instead of defaulting to weak shape-only checks. - Placement and mobile constraints: Decides shared, desktop, or mobile executor placement and addresses memory, filesystem, and platform-specific limitations with SkipExecutor entries when needed. - Use Case: After adding a new translation function to the SDK, invoke the skill to generate test definitions, register executors, pick one smoke candidate, and receive the exact run:local:desktop --filter command to verify the tests pass. ## Quick Start Ask the assistant to add e2e tests for the new SDK feature you just implemented in packages/sdk.
