Overview

Browser-Use MCP Server provides a powerful way for AI assistants to interact with web browsers. Using the Model Context Protocol (MCP) with Server-Sent Events (SSE) transport, this tool allows AI systems to navigate websites, extract information, and perform complex web tasks.

Installation

Prerequisites

• Python environment
• UV package manager

Setup Instructions

1. Install UV package manager:

   curl -LsSf https://astral.sh/uv/install.sh | sh
   

2. Install dependencies:

   uv sync
   uv pip install playwright
   uv run playwright install --with-deps --no-shell chromium
   

3. Configure environment variables: Create a .env file with the following:

   OPENAI_API_KEY=[your api key]
   CHROME_PATH=[optional: path to custom chrome build]
   

4. Start the server:

   uv run server --port 8000
   

Docker Deployment

You can also run the server in a Docker container:

docker build -t browser-use-mcp-server .
docker run --rm -p8000:8000 -p5900:5900 browser-use-mcp-server

For production environments, you can set a custom VNC password:

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

Connecting to the MCP Server

Configuration

Add the MCP server to your AI client by configuring it with the URL http://localhost:8000/sse.

For Cursor:

Create or edit ./.cursor/mcp.json:

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

For Claude Desktop:

Edit the configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the same MCP server configuration as shown above.

For Claude Code:

Use the same configuration format in the appropriate settings location.

Available Tools

The MCP server provides the following tools:

1. browser_use - Initiates browser tasks with a URL and action
2. browser_get_result - Retrieves results of asynchronous browser tasks
3. VNC server - Streams the browser interface for real-time observation

Using the VNC Viewer

To view the browser actions in real-time:

1. Clone the noVNC repository:

   git clone https://github.com/novnc/noVNC
   cd noVNC
   

2. Start the proxy:

   ./utils/novnc_proxy --vnc localhost:5900
   

3. Connect to the VNC server using a web browser (default password: browser-use)

Example Usage

Once connected, you can ask your AI assistant to perform web tasks like:

• "Open https://news.ycombinator.com and return the top ranked article"
• "Search for recent AI research papers on arXiv"
• "Check the weather forecast for New York City"

Support

For issues or questions, reach out at https://cobrowser.xyz/