Browser Use MCP Server
Web browser automation for AI agents using the browser-use framework.
Data last scanned today · Reviewed today
Overview
Browser Use MCP Server provides an interface for AI agents to execute browser-based tasks by leveraging the browser-use library. It utilizes Playwright for automation, enabling models to interact with web pages directly from the MCP environment. The server supports both SSE and stdio transport modes, requiring the mcp-proxy tool for the latter. Installation involves using uv for package management and includes specific setup steps for Playwright dependencies and environment configuration via a .env file.
Our verdict
With a security score of 48/100, this server is classified as high-risk. This rating reflects that the project is community-maintained rather than vendor-backed, and key security aspects such as OAuth support and read-only mode have not yet been editorially reviewed. Users should exercise caution, as browser automation tools inherently interact with external web content, and this implementation lacks external verification.
- Zero direct dependencies minimizes the supply-chain attack surface.
- Licensed under the permissive MIT license for flexible use.
- Supports multiple transport modes including SSE and stdio.
- Includes active documentation for local setup and integration.
- Not maintained by an official vendor organization.
- Critical security features like read-only modes remain unreviewed.
- Authentication methods have not undergone editorial security assessment.
- Last repository commit was 72 days ago.
Compatibility
| Client | Local | Docker | Remote | Read-only |
|---|---|---|---|---|
| Claude Desktop | ||||
| Cursor | ||||
| VS Code | ||||
| Windsurf | ||||
| ChatGPT |
Frequently asked questions
›What are the core software prerequisites for this server?
You must install uv as the package manager, Playwright for browser automation, and mcp-proxy for stdio mode support.
›How do I configure the server to wait for tasks to complete?
You can set the PATIENT environment variable to true in your .env file to force API calls to wait for task completion.
›Does the server support multiple communication modes?
Yes, it supports both SSE (Server-Sent Events) and stdio modes for client communication.
›How can I monitor the browser automation process in real-time?
The server provides VNC streaming, allowing you to watch browser actions in a VNC viewer on port 5900.
›What is the default password for the VNC connection when using Docker?
The default password is 'browser-use', though you can override this by mounting a custom password file into the container.
Badge
Maintain this server? Add the live badge to your README.