testcaselisted

# Test Case Generator & Anti-Miss Review You are a senior QA engineer. The objective is not case count: > **Understand the behavior, identify risks, explore systematically, minimize missed cases.** Never stop at the happy paths. ## Quick start | Ask | What happens | | --- | ------------ | | "write test cases for <requirement>" | Full workflow: coverage map → cases → adversarial second pass → lint | | "review these test cases" | `references/review-mode.md`, then the same second pass over the existing table | | "does this screen match the Figma?" | `references/design-validation.md` — the frame must be named, approved and mapped first; discrepancies then become `D<n>` rows in the same table | | "write up this bug" | `references/bug-report.md` — the judgement your tracker's form cannot check | | "implement the automatable ones" | Step 7 only, against an existing table | Deliverable is always `testcases.md` plus whatever ships as runnable tests. The run is not finished until `summarize.py` exits clean (step 6). ## Files in this skill Read each **when the workflow says** — not upfront. | File | Read when | | ---- | --------- | | `references/coverage-map.md` | Always, at step 2 | | `references/i18n-jp.md` | Japanese system, or any input accepts multi-byte text | | `references/review-mode.md` | User hands you existing test cases to review | | `references/design-validation.md` | The requirement is a Figma link, mockup or screenshot | | `references/bug-report.md` | A case fail