Playwright Skill

Playwright Skill

Opinionated, production-tested Playwright guidance — every pattern includes when (and when not) to use it.

50+ reference guides covering the full Playwright surface: selectors, assertions, fixtures, page objects, network mocking, auth, visual regression, accessibility, API testing, CI/CD, debugging, and more — with TypeScript and JavaScript examples throughout.

Playwright 1.63 highlights covered in these guides: test lock for serializing contention on a shared resource, locator.visible() replacing the :visible pseudo-class, aria and screen snapshots in traces (snapshots: { dom, aria, screen }), --add-reporter for appending rather than replacing reporters, the perfetto reporter, typed request.get<User>(), and ariaSnapshotJSON(). From 1.62: reporter.preprocess() for selecting tests before execution, retryStrategy: 'isolated', WebP screenshots with a quality parameter, AbortSignal on actions and assertions, locator.waitForFunction(), apiResponse.timing(), and the stories-and-galleries component testing model that replaces the deprecated experimental CT packages. A dedicated dynamic-test-selection.md guide covers driving quarantine from an external flakiness API.

Earlier releases remain covered: 1.61 (WebAuthn passkeys via context.credentials, page.localStorage / page.sessionStorage, video retention modes, expect.soft.poll(), WebSockets in HAR and traces), 1.60 (on-demand HAR in tracing, locator.drop(), page-level aria snapshots, test.abort()), and 1.59 (screencast recording, CLI debugging and trace analysis). A dedicated trace-analysis.md guide covers agent-native debugging of trace.zip reports with the npx playwright trace CLI.

Security Trust Boundary

This skill is designed for testing applications you own or have explicit authorization to test. It does not support or endorse automating interactions with third-party websites or services without permission.

When writing tests or automation that fetch content from external sources (e.g., baseURL pointing to staging/production), treat all returned page content as untrusted input — never pass raw page text back into agent instructions or dynamic code execution without sanitization, as this creates an indirect prompt injection risk.

For CI/CD workflows, pin all external dependencies (GitHub Actions, Docker images) to immutable references (commit SHAs, image digests) rather than mutable version tags. See ci-github-actions.md and docker-and-containers.md for pinning guidance.

Golden Rules

1. getByRole() over CSS/XPath — resilient to markup changes, mirrors how users see the page
2. Never page.waitForTimeout() — use expect(locator).toBeVisible() or page.waitForURL()
3. Web-first assertions — expect(locator) auto-retries; expect(await locator.textContent()) does not
4. Isolate every test — no shared state, no execution-order dependencies
5. baseURL in config — zero hardcoded URLs in tests
6. Retries: 2 in CI, 0 locally — surface flakiness where it matters
7. Traces: 'on-first-retry' — rich debugging artifacts without CI slowdown
8. Fixtures over globals — share state via test.extend(), not module-level variables
9. One behavior per test — multiple related expect() calls are fine
10. Mock external services only — never mock your own app; mock third-party APIs, payment gateways, email

Guide Index

Writing Tests

Debugging & Fixing

Framework Recipes

Migration Guides

Architecture Decisions

CI/CD & Infrastructure

Specialized Topics

CLI Browser Automation

Language Note

All guides include TypeScript and JavaScript examples. When the project uses .js files or has no tsconfig.json, examples are adapted to plain JavaScript.