pw-visual-regressionlisted

# PW Visual Regression You set up **snapshot tests whose first baselines the engineer must generate and eyeball** — never trust an auto-approved baseline. You make snapshots deterministic, and visual assertions complement functional ones rather than replacing them. ## When to use - A page or component needs pixel/visual regression coverage. - Flaky snapshot diffs need masking or threshold tuning. - Someone says "add visual regression", "manage/update baselines", "why does my screenshot test keep failing". ## When *not* to use - Generating the underlying Playwright test → `pw-test-generator`. - Fixing a brittle locator used to find the screenshot target → `pw-locator-fixer`. - Diagnosing general test flakiness → `pw-flaky-debugger`. - Analyzing a trace → `pw-trace-analyzer`. - Controlling network responses that feed the page → `pw-network-mocker`. - Building the Page Object or fixture a visual test reuses → `pw-page-object-builder` / `pw-fixture-designer` (reuse what exists; don't design new ones here). Don't use visual regression as just another way to assert functional behavior — keep a functional assertion (`toBeVisible`, `toHaveText`) for behavior, and the screenshot for appearance. ## Language and project conventions Support both **JavaScript and TypeScript** — preserve the project's existing language and conventions; never convert between them, and never introduce TypeScript syntax into JavaScript output. ## Workflow 1. **Pick the smallest stable target** —