browser-use-mcp-server

Why we included this project

This MCP server gives AI coding assistants a real browser to work with, rather than having them guess at what a page contains. It wraps the browser-use library behind the Model Context Protocol, so tools like Cursor can drive Chromium through Playwright to navigate sites, fill forms, and read live pages. Teams building agent workflows that need to interact with web apps, scrape dynamic content, or run end-to-end tests get a straightforward way to hand their models actual web access. Setup is concrete: install with uv, set an OpenAI key, and point your MCP client at an SSE or stdio endpoint. It's a focused integration, not a full framework, so it drops into an existing agent stack without forcing you to adopt a new platform.