Browser Use

Browser-Use MCP is a Model Context Protocol (MCP) server that enables AI-powered web automation. It allows AI language models to interact with web browsers, performing tasks like navigating websites, extracting information, and automating web-based workflows.

Features

• 🌐 Seamless integration with AI language models
• 🚀 Supports SSE (Server-Sent Events) transport for real-time communication
• 🔧 Easy setup with Docker and environment variables
• 🖥️ Compatible with popular AI clients like Cursor.ai and Claude
• 🔐 Secure VNC password management for Docker deployments
• 🔄 Asynchronous browser task execution and result retrieval

Use Cases

1. Web scraping and data extraction: Automate the process of gathering information from websites, perfect for market research or content aggregation.
2. Automated testing: Use AI to navigate through web applications, test user interfaces, and report issues, streamlining QA processes.
3. Content creation assistance: Allow AI to research topics by browsing the web, collecting relevant information for writers or content creators.
4. Personal web assistant: Develop AI-powered tools that can perform web-based tasks on behalf of users, such as booking appointments or checking prices across multiple sites.

How to Use It

1. Install the required dependencies:

curl -LsSf https://astral.sh/uv/install.sh | sh
uv sync
uv pip install playwright
uv run playwright install --with-deps --no-shell chromium

2. Set up the environment variables in a .env file:

OPENAI_API_KEY=[your api key]
CHROME_PATH=[only change this if you have a custom chrome build]

3. Start the server:

uv run server --port 8000

4. Configure your AI client to use the MCP server:

• Add http://localhost:8000/sse to your client UI
• Or create an mcp.json file with the following content:

 {
   "mcpServers": {
     "browser-use-mcp-server": {
       "url": "http://localhost:8000/sse"
     }
   }
 }

5. Use the server by sending commands to your AI, like:

open https://news.ycombinator.com and return the top ranked article

FAQs

Q: Which AI clients are supported by Browser-Use MCP Server?
A: The server supports Cursor.ai, Claude Desktop, Claude Code, and Windsurf (though Windsurf doesn’t support SSE yet).

Q: How can I secure the VNC password when using Docker?
A: For production use, it’s recommended to use Docker secrets. Create a file with your password and mount it as a secret:

echo "your-secure-password" > vnc_password.txt
docker run -v $(pwd)/vnc_password.txt:/run/secrets/vnc_password your-image-name

Q: Can I use other LLM providers besides OpenAI?
A: While currently focused on OpenAI, support for other LLM providers like Claude, Grok, and Bedrock is planned for future updates.