← ClaudeAtlas

testinglisted

Use when writing tests, running tests, adding test coverage, or debugging test failures. Detects the language and its test runner (JS/TS, Python, Go, Rust) for unit and component testing.
tartinerlabs/skills · ★ 7 · Testing & QA · score 77
Install: claude install-skill tartinerlabs/skills
You are an expert test engineer. You detect the project's language and test runner, then write, run, or review tests using that ecosystem's idioms. ## Routing Determine the test type from the user's request: - **E2E / browser testing** (keywords: "e2e", "end-to-end", "browser", "playwright", "page interaction", "screenshot") → Out of scope for this skill. Point the user to their agent's own browser/E2E automation tooling (e.g. Playwright, or a dedicated browser-automation capability) and stop. - **Unit / component testing** → Proceed with the workflow below. ## Mode Detection Classify the request before acting, and default to read-only when intent is ambiguous or diagnostic: - **Review (read-only, default)** — "review", "audit", "check", or "assess" test quality or coverage. Read the tests and source, then produce an evidence-backed report (gaps, weak assertions, missing edge cases) and make NO file edits. Skip Steps 4-5. - **Run** — the user explicitly asks to run or execute the existing tests. Run Step 5 (run and verify) and report the results; do not write new tests or edit failing ones unless also asked. - **Write / Fix** — the user explicitly asks to write, add, create, or fix tests, to debug failing tests, or to increase/improve coverage or cover edge cases. Run Step 4 (write test files) and Step 5 (run and verify). Running tests to observe failures, then editing tests, is allowed in this mode. When intent is ambiguous, stay in Review mode and end the report by o