A Model Context Protocol server that provides browser automation capabilities.
Add this configuration to your MCP-compatible client:
{
"mcpServers":{
"browser-use": {
"command": "docker",
"args": ["run", "-i", "--rm", "groos12/browser-use-mcp"]
}
}
}This MCP server provides a tool for browser automation:
This tool allows you to navigate to a URL and perform browser actions according to provided instructions.
Parameters:
url(string): The URL to navigate toinstructions(string): The instructions to follow in the browser
Example:
{
"url": "https://example.com",
"instructions": "Search for 'climate change' and extract the first three results"
}Returns: A text summary of the browser interaction history and results.
The tool leverages the browser-use library with Google's Gemini model to perform intelligent browser automation.
Build the Docker image with:
docker build -t mcp/browser-use .Run the MCP server in a Docker container:
docker run -i --rm --init -e DOCKER_CONTAINER=true mcp/browser-useIf you need to provide API keys or other environment variables:
docker run -i --rm --init \
-e GOOGLE_API_KEY=your_api_key \
-e DOCKER_CONTAINER=true \
mcp/browser-useFor development with volume mounting:
docker run -i --rm --init \
-e DOCKER_CONTAINER=true \
-v $(pwd):/app \
mcp/browser-use- The Docker container includes Chromium for headless browser automation
- Make sure any required API keys are provided as environment variables
- Required Environment Variables:
GOOGLE_API_KEY: API key for Google Generative AI (Gemini model)
This project uses:
- browser-use - Browser automation library
- langchain-google-genai - LangChain integration for Google's Generative AI
- Model Context Protocol SDK - MCP implementation