playwright-expert

Create a class per page that exposes locators as readonly properties and actions as methods, then instantiate it in tests with the page fixture. Combine with custom fixtures via test.extend to share page objects and authenticated state across tests.

Role-based selectors like getByRole are the most resilient, followed by getByLabel, getByPlaceholder, and getByTestId. Avoid CSS class selectors and XPath because they break when styling or markup changes.

Enable traces with trace: 'on-first-retry', rerun with retries, and inspect the timeline in the trace viewer. Replace waitForTimeout with proper waits like waitForResponse or locator state assertions, and ensure tests are fully isolated.

Use page.route to intercept requests and call route.fulfill with a custom status and JSON body, or route.fetch to modify real responses. You can also record and replay network traffic with routeFromHAR.

Yes. Install browsers with npx playwright install --with-deps, run npx playwright test, and upload the HTML report as an artifact. Set retries and a single worker in CI via environment-variable checks in playwright.config.ts.

waitForTimeout introduces arbitrary delays that either slow tests or still fail under load. Playwright's auto-waiting locators and explicit waits like waitForURL, waitForResponse, and toBeVisible synchronize on actual application state instead.