What problem does it solve? Manually driving a browser to test web pages, debug failing Playwright tests, or capture page state is slow and error-prone. This Skill lets an AI agent control a real browser session through the playwright-cli command line, producing snapshots, generated TypeScript code, traces, and videos without writing boilerplate automation scripts. ## Core Features & Use Cases - Interactive Browser Control: Open pages, click, fill forms, navigate tabs, manage cookies and storage, and inspect elements via snapshot refs or CSS/role locators. - Test Plan, Generate, and Heal: Explore an app through a seed test, write a spec file, generate Playwright test files from recorded actions, and debug or heal failing tests by attaching to paused debug sessions. - Network Mocking and Diagnostics: Intercept and mock requests, capture console output, record traces and annotated videos, and run arbitrary Playwright code for advanced scenarios like geolocation or downloads. - Use Case: A failing CI test needs investigation. Run it with --debug=cli, attach playwright-cli to the paused session, snapshot the page to find the changed locator, rehearse the fix, and paste the generated TypeScript back into the test. ## Quick Start Ask the agent to open a website with playwright-cli, take a snapshot, and click or fill elements using the refs shown in the snapshot.