MCP Server

Run browser automation tasks from your AI coding assistant. Connect to Claude, Cursor, Windsurf, or any MCP client.

Browser Use provides two Model Context Protocol server endpoints:

Endpoint	URL
v2	https://api.browser-use.com/mcp
v3	https://api.browser-use.com/v3/mcp

The v2 endpoint uses a task-based agent — each task runs independently and is dispatched to a queue. The v3 endpoint uses a session-based agent that runs in a persistent sandbox container, supporting keep-alive for multi-turn interactions, model selection (bu-mini, bu-max, bu-ultra), structured output, and shared workspaces for file persistence across sessions.

Get your API key from the Browser Use Dashboard.

Claude Code

v2

claude mcp add -t http -H "x-browser-use-api-key: YOUR_API_KEY" browser-use https://api.browser-use.com/mcp

v3

claude mcp add -t http -H "x-browser-use-api-key: YOUR_API_KEY" browser-use https://api.browser-use.com/v3/mcp

Replace YOUR_API_KEY with your actual API key. Claude Code requires the API key to be passed via the -H (header) flag.

Claude Desktop

Add to claude_desktop_config.json:

v2

{
  "mcpServers": {
    "browser-use": {
      "url": "https://api.browser-use.com/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

v3

{
  "mcpServers": {
    "browser-use": {
      "url": "https://api.browser-use.com/v3/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Cursor

Add to .cursor/mcp.json:

v2

{
  "mcpServers": {
    "browser-use": {
      "url": "https://api.browser-use.com/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

v3

{
  "mcpServers": {
    "browser-use": {
      "url": "https://api.browser-use.com/v3/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

v2

{
  "mcpServers": {
    "browser-use": {
      "serverUrl": "https://api.browser-use.com/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

v3

{
  "mcpServers": {
    "browser-use": {
      "serverUrl": "https://api.browser-use.com/v3/mcp",
      "headers": {
        "x-browser-use-api-key": "YOUR_API_KEY"
      }
    }
  }
}

Available Tools

v2

Tool	Description	Cost
browser_task	Run a full automation task	$0.01 init + per-step (default$0.006/step). See Pricing.
execute_skill	Run a skill	$0.02/call
list_skills	List available skills	Free
get_cookies	Extract cookies for auth	Free
list_browser_profiles	List profiles	Free
monitor_task	Check task progress	Free

v3

Tool	Description
run_session	Create a new browser session and run a task. Supports keep_alive to reuse the session, model selection (bu-mini, bu-max, bu-ultra), output_schema for structured output, and profile_id for authenticated browsing.
get_session	Poll a session for status and output. Returns status (running, idle, stopped, error), step count, cost breakdown, and a live URL to watch the browser.
send_task	Send a follow-up task to an idle keep-alive session. Enables multi-turn interactions that reuse the same browser context.
stop_session	Stop a running session. Use strategy: "task" to stop only the current task (session stays idle), or "session" to destroy the sandbox.
get_session_messages	Get the agent’s step-by-step messages for a session, including browser actions, reasoning, and results.
list_sessions	List recent sessions with status and cost info.
list_browser_profiles	List cloud browser profiles for authenticated tasks.