mcpField
M
Mcp Android Agent Python

Mcp Android Agent Python

Local
github.com/nim444AI & Agents59Updated 49w ago

About Mcp Android Agent Python

MCP Android Agent Python is a Python MCP server for controlling Android devices through uiautomator2. It is intended for AI agents and automation or QA workflows, with tools for device, app, screen, input, and UI inspection tasks.

Key features

  1. 01Detect and inspect connected devices
  2. 02Manage installed apps and app state
  3. 03Control screen power and unlocking
  4. 04Simulate taps, gestures, keys, and text
  5. 05Inspect UI elements and hierarchy
  6. 06Capture screenshots and wait for UI states

Use cases

  1. 01Automate Android app testing
  2. 02Let an AI agent operate a connected device
  3. 03Inspect UI state while debugging
  4. 04Manage apps and device status

MCP support

The MCP server exposes Android device-control and inspection tools to MCP clients. It supports stdio for agent integration and Streamable HTTP for web or API use.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

No install method listed yet

We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.

Repository

Tools

No tool list yet

Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.

Details

From the project README

Python 3.13 CI Pipeline Coverage: 90% Code style: ruff License: MIT

MCP Android Agent

This project provides an MCP (Model Context Protocol) server for automating Android devices using uiautomator2. It's designed to be easily plugged into AI agents like GitHub Copilot Chat, Claude, or Open Interpreter to control Android devices through natural language.

Modular Architecture

The server has been refactored into a clean, modular architecture with tools organized by functionality:

mcp-android-server-python/
├── server.py                    # Main server (61 lines - clean & focused)
├── server_original_backup.py    # Backup of original monolithic version
└── tools/                       # 🆕 Modular tools package
    ├── __init__.py             # Central registration & imports
    ├── device_tools.py         # Device connection & status tools
    ├── app_tools.py            # Application management tools
    ├── screen_tools.py         # Screen control & unlock tools
    ├── input_tools.py          # User input simulation (click, swipe, text)
    ├── inspection_tools.py     # UI inspection & screenshots
    └── advanced_tools.py       # Advanced features (toast, activity wait)
Benefits of Modular Architecture
  • Maintainability: Easy to add/modify/remove tools without touching main server
  • Organization: Tools grouped logically by functionality
  • Testing: Individual tool modules can be unit tested separately
  • Reusability: Tool modules can be reused in other projects
  • Scalability: New tool categories can be added as separate modules
  • Clean Code: Main server reduced from 1321 lines to 61 lines

Quick Demo

Demo


Demo

Requirements

  • Python 3.13 or higher
  • Android Debug Bridge (adb) installed and in PATH
  • Connected Android device with USB debugging enabled
  • uiautomator2 compatible Android device

Features

Device Management
  • Smart Device Detection: Automatically finds and connects to available devices
  • Comprehensive Device Info: Get serial, resolution, battery, WiFi IP, Android version
  • ADB Diagnostics: Check ADB availability and connection status
  • Health Monitoring: Built-in server health checks
Application Management
  • App Discovery: List all installed applications (system + user apps)
  • App Lifecycle Control: Start, stop, force-stop apps by package name
  • App State Monitoring: Track current foreground app and activity
  • Data Management: Clear app data/cache for testing
Screen & Display Control
  • Screen Power Management: Turn screen on/off programmatically
  • Smart Unlock: Automatic screen unlocking with standard methods
  • Screen State Monitoring: Wait for screen activation (async support)
User Input Simulation
  • Precision Interactions: Click by text, resource ID, or content description
  • Advanced Gestures: Long click, swipe, drag operations
  • Text Input: Smart text entry with optional field clearing
  • Hardware Keys: Simulate home, back, menu, volume keys
UI Inspection & Debugging
  • Element Analysis: Get detailed UI element properties and bounds
  • Screen Capture: Take screenshots for debugging and documentation
  • UI Hierarchy: Export complete screen structure as XML
  • Smart Waiting: Wait for elements to appear with custom timeouts
  • Scroll Detection: Auto-scroll to find elements in long lists
Advanced Capabilities
  • Toast Detection: Capture system toast messages for verification
  • Activity Monitoring: Wait for specific Android activities
  • Background Operations: Async support for time-consuming operations

Use Cases

Perfect for:

  • AI agents that need to interact with real devices
  • Remote device control setups
  • Automated QA tools
  • Android bot frameworks
  • UI testing and automation
  • Device management and monitoring

Installation

1. Clone the repo
git clone https://github.com/nim444/mcp-android.git
cd mcp-android
2. Create and activate virtual environment
# Using uv (https://github.com/astral-sh/uv)
uv venv
source .venv/bin/activate  # On Windows: .venv\\Scripts\\activate
3. Install dependencies
uv pip install

Running the Server

The server supports two different transport modes depending on your use case:

Option 1: MCP stdio (For AI Agent Integration)

This is the standard mode for integrating with AI agents like Claude Desktop, VS Code, or other MCP clients.

# Edit server.py to use stdio mode (default commented out)
# Uncomment the stdio section and comment out http section

# Then run:
uv run python server.py
Option 2: Streamable HTTP (For Web/API Integration)

This mode runs the server as an HTTP API, useful for web applications, curl testing, or direct HTTP calls.

# Current default configuration - runs as HTTP server
uv run python server.py

# Server will be available at: http://localhost:8080
Switching Between Modes

Edit server.py and modify the if __name__ == "__main__": section:

For stdio mode (AI agents):

if __name__ == "__main__":
    mcp.run(
        transport="stdio",
        show_banner=False,
    )

For HTTP mode (Web API):

if __name__ == "__main__":
    mcp.run(
        transport="streamable-http",
        host="0.0.0.0",
        port=8080,
    )

Usage

For AI Agent Integration (Claude Desktop, VS Code, etc.)

An MCP client is needed to use this server. The Claude Desktop app is an example of an MCP client.

Important: For AI agent integration, make sure to configure the server in stdio mode (see "Option 1" above).

To use this server with Claude Desktop:

Locate your Claude Desktop configuration file
  • Windows: %APPDATA%\Claude\claude_desktop_config.json
  • macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Add the Android MCP server configuration to the mcpServers section
{
  "mcpServers": {
    "mcp-android": {
      "type": "stdio",
      "command": "bash",
      "args": [
        "-c",
        "cd /path/to/mcp-adb && source .venv/bin/activate && uv run python server.py"
      ]
    }
  }
}

Replace /path/to/mcp-adb with the absolute path to where you cloned this repository. For example: /Users/username/Projects/mcp-adb

Using with VS Code

You can also use this MCP server with VS Code's agent mode (requires VS Code 1.99 or newer). To set up:

  1. Create a .vscode/mcp.json file in your workspace:
{
  "servers": {
    "mcp-android": {
      "type": "stdio",
      "command": "bash",
      "args": [
        "-c",
        "cd /path/to/mcp-adb && source .venv/bin/activate && uv run python server.py"
      ]
    }
  }
}

Replace /path/to/mcp-adb with the absolute path to where you cloned this repository.

After adding the configuration, you can manage the server using:

  • Command Palette → MCP: List Servers to view and manage configured servers
  • Command Palette → MCP: Start Server to start the server
  • The server's tools will be available in VS Code's agent mode chat

Vscode

For HTTP API Integration (Direct API Calls)

When running in HTTP mode (Option 2), you can interact with the server directly via HTTP requests:

# Check if server is running
curl http://localhost:8080/

# List available tools (you'll need to implement proper tool discovery endpoints)
# This depends on your FastMCP version and configuration

Use Cases for HTTP Mode:

  • Web applications with Android automation
  • Testing tools that can't use stdio
  • Direct API integration with other services
  • Debugging and development with curl/Postman

UI Inspector

The project includes support for uiauto.dev, a powerful UI inspection tool for viewing and analyzing your device's interface structure.

  1. Install the UI inspector:
uv pip install uiautodev
  1. Start the inspector:
uiauto.dev
  1. Open your browser and navigate to https://uiauto.dev

Ui

Available MCP Tools

Device Management Tools
Tool NameDescription
mcp_healthCheck if the MCP server is running properly
get_device_statusGet complete device status and readiness information
connect_deviceConnect to an Android device and get basic info
get_device_infoGet detailed device info: serial, resolution, battery, etc.
check_adb_and_list_devicesCheck if ADB is installed and list connected devices
Application Management Tools
Tool NameDescription
get_installed_appsList all installed apps with version and package info
get_current_appGet info about the app currently in the foreground
start_appStart an app by its package name
stop_appStop an app by its package name
stop_all_appsStop all currently running apps
clear_app_dataClear user data/cache of a specified app
Screen Control Tools
Tool NameDescription
screen_onTurn on the screen
screen_offTurn off the screen
unlock_screenUnlock the screen (turn on and swipe if necessary)
wait_for_screen_onWait asynchronously until the screen is turned on
User Input Tools
Tool NameDescription
press_keySimulate hardware key press (e.g. home, back, menu, etc.)
clickTap on an element by text, resourceId, or description
long_clickPerform a long click on an element
send_textInput text into currently focused field (optionally clearing before)
swipeSwipe from one coordinate to another
dragDrag an element to a specific screen location
Inspection Tools
Tool NameDescription
get_element_infoGet info on UI elements (text, bounds, clickable, etc.)
wait_for_elementWait for an element to appear on screen
scroll_toScroll until a given element becomes visible
screenshotTake and save a screenshot from the device
dump_hierarchyDump the UI hierarchy of the current screen as XML
Advanced Tools
Tool NameDescription
get_toastGet the last toast message shown on screen
wait_activityWait until a specific activity appears

License

This project is licensed under the MIT License - see the LICENSE file for details.

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.