levisnkyyy-browser-use-mcp/README.md

MCP server for Browser-Use API

# Browser-Use MCP Server

An MCP (Model Context Protocol) server that provides browser automation capabilities through the Browser-Use API.

## Installation

```bash
cd mcp
npm install
npm run build
```

## Configuration

The MCP server requires the following environment variables:

- `BROWSER_USE_API_URL`: The URL of the Browser-Use API server (default: `http://localhost:8000`)
- `BROWSER_USE_API_KEY`: Optional API key for authentication with the Browser-Use API

### LLM Provider API Keys

LLM provider API keys (OpenAI, Anthropic, Google) must be configured on the Browser-Use API server side as environment variables:

- `OPENAI_API_KEY`: For OpenAI models (e.g., gpt-4o)
- `ANTHROPIC_API_KEY`: For Anthropic models (e.g., claude-3-5-sonnet)
- `GOOGLE_API_KEY`: For Google models (e.g., gemini-pro)

These keys are not passed through the MCP protocol for security reasons.

## Usage

### With Claude Desktop

Add the following to your Claude Desktop configuration (`~/Library/Application Support/Claude/claude_desktop_config.json`):

```json
{
  "mcpServers": {
    "browser-use": {
      "command": "node",
      "args": ["/path/to/browser-use/mcp/dist/index.js"],
      "env": {
        "BROWSER_USE_API_URL": "http://localhost:8000"
      }
    }
  }
}
```

### With other MCP clients

```bash
export BROWSER_USE_API_URL=http://localhost:8000
node dist/index.js
```

## Available Tools

### Task Management

- `browser_use_create_task` - Create a new browser automation task
- `browser_use_get_task` - Get task status and results
- `browser_use_list_tasks` - List all tasks
- `browser_use_cancel_task` - Cancel a running task
- `browser_use_pause_task` - Pause a running task
- `browser_use_get_task_history` - Get task execution history
- `browser_use_get_screenshot` - Get current browser screenshot
- `browser_use_stream_task` - Stream real-time task updates

### Session Management

- `browser_use_create_session` - Create a reusable browser session
- `browser_use_close_session` - Close a browser session

### System

- `browser_use_health` - Check API health status

## Examples

### Create a task

```typescript
{
  "task": "Navigate to example.com and take a screenshot",
  "browser_config": {
    "headless": false,
    "viewport": { "width": 1920, "height": 1080 }
  }
}
```

### Create a reusable session

```typescript
{
  "headless": false,
  "viewport": { "width": 1280, "height": 720 },
  "cookies": [
    {
      "name": "session",
      "value": "abc123",
      "domain": ".example.com"
    }
  ]
}
```

## Development

```bash
# Run in development mode with hot reload
npm run dev

# Build for production
npm run build

# Start production server
npm start
```