Browser Use MCP Server

Browser AutomationPython

Control web browsers directly from AI agents

Available Tools

navigate

Navigate to a specified URL in the browser

url

click

Click on an element identified by a selector

selector

type

Type text into an input field identified by a selector

selectortext

screenshot

Take a screenshot of the current page

get_text

Get text content from an element identified by a selector

selector

get_html

Get the HTML content of the current page

Browser Use is an MCP server that enables AI agents to control web browsers through the browser-use library. It provides a seamless interface for AI systems to interact with web content, navigate websites, fill forms, and extract information from web pages. The server acts as a bridge between AI agents and browser automation capabilities, making web browsing accessible to AI applications.

Overview

Browser Use MCP Server enables AI agents to control web browsers programmatically. This server implements the Model Context Protocol (MCP) to provide a standardized interface for AI systems to interact with web browsers through the browser-use library.

Installation

Prerequisites

Before installing the Browser Use MCP Server, you need to set up the following prerequisites:

Install uv - A fast Python package manager:

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

Install mcp-proxy - Required for stdio mode:

uv tool install mcp-proxy
uv tool update-shell

Install Playwright - The browser automation framework used by Browser Use.

Installing Browser Use MCP Server

You can install the server using uv:

uv pip install browser-use-mcp-server

Or using pip:

pip install browser-use-mcp-server

Configuration

The server can be configured using environment variables. You can create a .env file based on the .env.example in the repository:

BROWSER_USE_MCP_SERVER_HOST: Host to bind the server to (default: 127.0.0.1)
BROWSER_USE_MCP_SERVER_PORT: Port to bind the server to (default: 8000)
BROWSER_USE_MCP_SERVER_LOG_LEVEL: Logging level (default: info)

Usage

Running the Server

You can run the server in different modes:

HTTP Mode (default):
```
browser-use-mcp-server
```
STDIO Mode (for direct integration with AI agents):
```
mcp-proxy browser-use-mcp-server --stdio
```

Docker Deployment

You can also run the server using Docker:

docker run -p 8000:8000 ghcr.io/co-browser/browser-use-mcp-server:latest

Integration with AI Agents

To integrate with AI agents like Claude or other MCP-compatible systems, add the following configuration to your client:

{
  "mcpServers": {
    "browser-use": {
      "command": "mcp-proxy",
      "args": [
        "browser-use-mcp-server",
        "--stdio"
      ]
    }
  }
}

For HTTP mode integration, use:

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

Advanced Configuration

For advanced users, you can customize the browser behavior by setting additional environment variables:

BROWSER_USE_MCP_SERVER_HEADLESS: Set to "false" to show the browser UI (default: true)
BROWSER_USE_MCP_SERVER_BROWSER_TYPE: Choose browser type (chromium, firefox, webkit)
BROWSER_USE_MCP_SERVER_TIMEOUT: Set default timeout for browser operations

Troubleshooting

If you encounter issues:

Check that Playwright is properly installed
Verify that mcp-proxy is installed and accessible
Check your environment variables configuration
Look at the server logs for detailed error messages

For more help, join the Discord community or open an issue on GitHub.