What it does
Playwright drives Chromium, Firefox and WebKit through one API, for tests, scripts and AI agent workflows. It is available for TypeScript, Python, .NET and Java, runs on Linux, macOS and Windows, and works headless or headed. For testing, Playwright Test is a full test runner with auto-waiting, assertions that retry until they pass, tracing and parallel runs that can be sharded across machines. Each test gets a fresh browser context, so state stays isolated, and saved authentication state can be reused across tests. Locators follow how users see the page, such as by role, label or placeholder. For AI agents, a Playwright MCP server gives agents browser control through structured accessibility snapshots rather than screenshots. A separate Playwright CLI is aimed at coding agents such as Claude Code and GitHub Copilot, with installable skills, and a session dashboard shows live previews of running browser sessions. Tooling includes a test generator that records browser actions and writes the test code, a Trace Viewer with DOM snapshots, network requests and console logs, and a VS Code extension for running and debugging tests.
Capabilities
- Test runner with auto-waiting, retrying assertions and parallel runs
- Records browser actions and writes the test code for you
- Trace Viewer with DOM snapshots, network requests and console logs
- MCP server that gives AI agents browser control through accessibility snapshots
- Command-line tool for coding agents with installable skills
- Chromium, Firefox and WebKit on Linux, macOS and Windows
When to use it
- You need end-to-end tests that run across Chromium, Firefox and WebKit
- You want an AI coding agent to operate a browser through MCP or a CLI
- You want to script browser tasks from TypeScript, Python, .NET or Java
