# Browser Use MCP Server
A FastMCP server that enables browser automation through natural language commands. This server allows Language Models to browse the web, fill out forms, click buttons, and perform other web-based tasks via a simple API.
## Quick Start
### 1. Install the package
Install with a specific provider (e.g., OpenAI)
\`\`\`bash
pip install -e "git+https://github.com/yourusername/browser-use-mcp.git#egg=browser-use-mcp[openai]"
\`\`\`
Or install all providers
\`\`\`bash
pip install -e "git+https://github.com/yourusername/browser-use-mcp.git#egg=browser-use-mcp[all-providers]"
\`\`\`
Install Playwright browsers
\`\`\`bash
playwright install chromium
\`\`\`
### 2. Configure your MCP client
Add the browser-use-mcp server to your MCP client configuration:
\`\`\`javascript
\{
"mcpServers": \{
"browser-use-mcp": \{
"command": "browser-use-mcp",
"args": ["--model", "gpt-4o"],
"env": \{
"OPENAI_API_KEY": "your-openai-api-key", // Or any other provider's API key
"DISPLAY": ":0" // For GUI environments
\}
\}
\}
\}
\`\`\`
Replace \`"your-openai-api-key"\` with your actual API key or use an environment variable reference like \`process.env.OPENAI_API_KEY\`.
### 3. Use it with your favorite MCP client
#### Example using mcp-use with Python
\`\`\`python
import asyncio
import os
from dotenv import load_dotenv
from langchain_openai import ChatOpenAI
from mcp_use import MCPAgent, MCPClient
async def main():
# Load environment variables
load_dotenv()
# Create MCPClient from config file
client = MCPClient(
config=\{
"mcpServers": \{
"browser-use-mcp": \{
"command": "browser-use-mcp",
"args": ["--model", "gpt-4o"],
"env": \{
"OPENAI_API_KEY": os.getenv("OPENAI_API_KEY"),
"DISPLAY": ":0",
\},
\}
\}
\}
)
# Create LLM
llm = ChatOpenAI(model="gpt-4o")
# Create agent with the client
agent = MCPAgent(llm=llm, client=client, max_steps=30)
# Run the query
result = await agent.run(
"""
Navigate to https://github.com, search for "browser-use-mcp", and summarize the project.
""",
max_steps=30,
)
print(f"\nResult: \{result\}")
if __name__ == "__main__":
asyncio.run(main())
\`\`\`
#### Using Claude for Desktop
1. Open Claude for Desktop
2. Go to Settings → Experimental features
3. Enable Claude API Beta and OpenAPI schema for API
4. Add the following configuration to your Claude Desktop config file:
- Mac: \`~/Library/Application Support/Claude/claude_desktop_config.json\`
- Windows: \`%AppData%\Claude\claude_desktop_config.json\`
\`\`\`json
\{
"mcpServers": \{
"browser-use": \{
"command": "browser-use-mcp",
"args": ["--model", "claude-3-opus-20240229"]
\}
\}
\}
\`\`\`
5. Start a new conversation with Claude and ask it to perform web tasks
## Supported LLM Providers
The following LLM providers are supported for browser automation:
| Provider | API Key Environment Variable |
|----------|----------------------------|
| OpenAI | \`OPENAI_API_KEY\` |
| Anthropic | \`ANTHROPIC_API_KEY\` |
| Google | \`GOOGLE_API_KEY\` |
| Cohere | \`COHERE_API_KEY\` |
| Mistral AI | \`MISTRAL_API_KEY\` |
| Groq | \`GROQ_API_KEY\` |
| Together AI | \`TOGETHER_API_KEY\` |
| AWS Bedrock | \`AWS_ACCESS_KEY_ID\` and \`AWS_SECRET_ACCESS_KEY\` |
| Fireworks | \`FIREWORKS_API_KEY\` |
| Azure OpenAI | \`AZURE_OPENAI_API_KEY\` and \`AZURE_OPENAI_ENDPOINT\` |
| Vertex AI | \`GOOGLE_APPLICATION_CREDENTIALS\` |
| NVIDIA | \`NVIDIA_API_KEY\` |
| AI21 | \`AI21_API_KEY\` |
| Databricks | \`DATABRICKS_HOST\` and \`DATABRICKS_TOKEN\` |
| IBM watsonx.ai | \`WATSONX_API_KEY\` |
| xAI | \`XAI_API_KEY\` |
| Upstage | \`UPSTAGE_API_KEY\` |
| Hugging Face | \`HUGGINGFACE_API_KEY\` |
| Ollama | \`OLLAMA_BASE_URL\` |
| Llama.cpp | \`LLAMA_CPP_SERVER_URL\` |
For more information check out: https://python.langchain.com/docs/integrations/chat/
You can create a \`.env\` file in the project directory with your API keys:
\`\`\`
OPENAI_API_KEY=your_openai_key_here
# Or any other provider key
\`\`\`
## Troubleshooting
- **API Key Issues**: Ensure your API key is correctly set in your environment variables or \`.env\` file.
- **Provider Not Found**: Make sure you've installed the required provider package.
- **Browser Automation Errors**: Check that Playwright is correctly installed with \`playwright install chromium\`.
- **Model Selection**: If you get errors about an invalid model, try using the \`--model\` flag to specify a valid model for your provider.
- **Debug Mode**: Use \`--debug\` to enable more detailed logging that can help identify issues.
- **MCP Client Configuration**: Make sure your MCP client is correctly configured with the right command and environment variables.
## License
MIT # browser-use-mcp
