Skip to main content
The Minitap MCP (Model Context Protocol) server enables AI assistants like Windsurf, Cursor, and Claude Desktop to control and interact with real mobile devices through natural language commands.
The MCP server integrates seamlessly with your IDE, allowing you to automate mobile testing, implement Figma designs, and analyze device screens directly from your AI assistant.

​
Installation

​
Prerequisites

Before using the MCP server, ensure you have:
Complete the Platform Quickstart to set up your environment, including Python 3.12+, device requirements, and API key configuration.
  • Xcode Command Line Tools with xcrun - Maestro - For mobile automation - iOS Simulator running

​
Running the Server

Start the MCP server with your API key: 
minitap-mcp --server --api-key your_minitap_api_key
The server will start on localhost:8000 by default.
Optional: Customize the port with --port 9000 or use environment variables for configuration.

​
Configuration Options

  • CLI Arguments
  • Environment Variables
# Basic usage
minitap-mcp --server --api-key YOUR_KEY

# With custom port
minitap-mcp --server --api-key YOUR_KEY --port 9000

# With LLM profile (default: 'default')
minitap-mcp --server --api-key YOUR_KEY --llm-profile PROFILE_NAME
Available options:
  • --api-key: Your Minitap API key (required)
  • --port: Custom port (default: 8000)
  • --llm-profile: LLM profile name (optional)