simple-browser-use-mcp 0.2.1

An unofficial Model Context Protocol server for automating browser tasks using Browser Use API. This package is not provided or endorsed by Browser Use.

Browser Use MCP server

Overview

A Model Context Protocol server for automating browser tasks using Browser Use API. This server provides tools to run browser automation tasks, monitor task status, and manage running tasks.

Prerequisites

• A Browser Use API key

to get a Browser Use API key, go to Cloud Browser Use and sign up.

Installation

Installing via Smithery

To install browser-use-mcp for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @hadiazarabad/browser-use-mcp --client claude

The package is not published to PyPI. You'll need to clone this repository and run it directly from source.

git clone https://github.com/hadiazarabad/browser-use-mcp.git
cd browser-use-mcp

Running the Server

Using Python directly

python /path/to/browser-use-mcp/src/server.py --api-key YOUR_BROWSER_USE_API_KEY

Using uvx (recommended)

First, install uv if you haven't already:

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

Then you can run the server using uvx:

uvx run /path/to/browser-use-mcp/src/server.py --api-key YOUR_BROWSER_USE_API_KEY

Or, for development purposes, you can use:

uv run /path/to/browser-use-mcp/src/server.py --api-key YOUR_BROWSER_USE_API_KEY

Tools

1. run task

   • Run a Browser Use automation task with instructions and wait for completion
   • Input:
     • instructions (string): Instructions for the browser automation task
     • structured_output (string, optional): JSON schema for structured output
     • parameters (object, optional): Additional parameters for the task
   • Returns: Information about the created task including final output if wait_for_completion is True

2. get task

   • Get details of a Browser Use task by ID
   • Input:
     • task_id (string): ID of the task to retrieve
   • Returns: Complete task information including steps and output

3. get task status

   • Get the status of a Browser Use task
   • Input:
     • task_id (string): ID of the task to check
   • Returns: Current status of the task

4. stop task

   • Stop a running Browser Use task
   • Input:
     • task_id (string): ID of the task to stop
   • Returns: Confirmation of task being stopped

5. pause task

   • Pause a running Browser Use task
   • Input:
     • task_id (string): ID of the task to pause
   • Returns: Confirmation of task being paused

6. resume task

   • Resume a paused Browser Use task
   • Input:
     • task_id (string): ID of the task to resume
   • Returns: Confirmation of task being resumed

7. list tasks

   • List all Browser Use tasks
   • Returns: List of all tasks with their IDs and statuses

8. check balance

   • Check your Browser Use account balance
   • Returns: Account balance information

Prompts

1. browser-use-task
   • Run a Browser Use automation task
   • Input:
     • instructions (string): Instructions for the browser automation task
     • structured_output (string, optional): JSON schema for structured output
   • Returns: Formatted task details as conversation context

Claude Desktop

Add this to your claude_desktop_config.json:

"mcpServers": {
  "browser-use": {
    "command": "uv",
    "args": [
        "run",
        "/path/to/browser-use-mcp/src/server.py",
        "--api-key",
        "YOUR_BROWSER_USE_API_KEY"
    ]
  }
}

License

This MCP server is licensed under the MIT License. This means you are free to use, modify, and distribute the software, subject to the terms and conditions of the MIT License. For more details, please see the LICENSE file in the project repository.