playwright-best-practices

Flaky Playwright tests usually stem from race conditions, missing waits, or state leaks between tests. Use web-first assertions with auto-waiting instead of explicit timeouts, ensure test isolation with fixtures, and investigate failures with the trace viewer using npx playwright show-trace.

Page Object Model in Playwright encapsulates page locators and interactions into reusable classes, keeping tests readable and maintainable. The skill's references cover when to choose POM versus fixtures and how to structure test suites around them.

Playwright fixtures handle setup, teardown, and dependency injection, while POM organizes page interactions. They solve different problems and are often combined: fixtures can instantiate page objects so tests receive ready-to-use page instances.

Playwright's route interception lets you mock API responses by intercepting network requests and fulfilling them with custom data. This covers REST and GraphQL endpoints, HAR recording and playback, and simulating network failures for error-state testing.

Playwright supports GitHub Actions, GitLab CI, Docker containers, and other providers with parallel sharding for faster runs. Configure retries, artifacts like traces and screenshots, and use test tags to run subsets such as @smoke tests on pull requests.

Parallel-only failures typically indicate shared state between workers, such as reused test accounts or database records. Isolate test data per worker using fixtures, and review the performance and flaky test references for worker-scoped isolation patterns.