← ClaudeAtlas

visual-checklisted

Generate and run visual regression tests (Playwright screenshot snapshots) against a running app, per key screen, with baselines and a tolerance threshold. Use for visual/design coverage of a test book or an app.
QAIA-Project/QAIA · ★ 0 · Testing & QA · score 64
Install: claude install-skill QAIA-Project/QAIA
# visual-check — visual regression References, two independent domains: [`examples/medibook/tests/visual.booking.spec.js`](https://github.com/QAIA-Project/QAIA/blob/main/examples/medibook/tests/visual.booking.spec.js) (login + booking baselines) and [`examples/expense-demo/tests/visual.expense.spec.js`](https://github.com/QAIA-Project/QAIA/blob/main/examples/expense-demo/tests/visual.expense.spec.js) (6 screens, mutation-verified: a button-colour change kills 5 snapshots, a 3-pixel padding shift kills the 2 that scope card lists, and no snapshot outside the mutated element's scope reacts). Generating the second suite surfaced a defect no functional or a11y test had: a sign-in error written into a region that is still `hidden` at that moment — invisible, and its `aria-live` announcement made inside a hidden subtree. Completes the 7-type coverage: E2E, API, mobile-emulation, a11y, perf, security, **visual**. ## Steps 1. **Generate a snapshot test per key screen** the test book covers, plus any the user names. Navigate to a **stable** state — data reset or seeded so pixels are deterministic — then: ```js await expect(locator).toHaveScreenshot('<screen>.png', { maxDiffPixelRatio: 0.002 }); ``` **That tolerance is the number that decides pass or fail, so it is stated rather than left blank.** `0.002` (0.2 % of pixels) absorbs anti-aliasing and font-hinting differences between runs on the same machine without hiding a real change: a moved button, a wrong c