Android-MCP and android-remote-control-mcp

These are complementary tools serving different deployment architectures: one runs on a host machine to control Android devices remotely, while the other runs directly on the Android device itself as a server, allowing bidirectional control and file operations over tunneled connections.

Android-MCP
68
Established
Maintenance 13/25
Adoption 17/25
Maturity 18/25
Community 20/25
Maintenance 13/25
Adoption 5/25
Maturity 9/25
Community 14/25
Stars: 447
Forks: 67
Downloads: 1,031
Commits (30d): 0
Language: Python
License: MIT
Stars: 11
Forks: 3
Downloads: —
Commits (30d): 0
Language: Kotlin
License: MIT
No risk flags
No Package No Dependents

About Android-MCP

CursorTouch/Android-MCP

MCP Server for interacting with Android Devices.

Implements an MCP server that bridges LLM agents to Android devices via ADB and the Accessibility API, enabling app navigation, UI interaction, and automated testing without computer vision models. Exposes a rich toolset including gesture controls (tap, swipe, drag), text input, view hierarchy inspection, and shell command execution—with typical action latency of 2-4 seconds. Integrates directly with Claude Desktop and any LLM/VLM, supporting Android 10+ devices and emulators.

About android-remote-control-mcp

danielealbano/android-remote-control-mcp

An MCP Server for Android running on the phone, optmized for token usage, supports also files downloads and cloudflare (free) and ngrok automated tunnelling.

Exposes 54 MCP tools via an HTTP server running directly on Android, enabling AI models to interact with any app through accessibility services and UI node introspection. Uses Ktor with optional HTTPS, streamable JSON transport, and token-optimized compact screen state representation with annotated screenshots rather than verbose XML dumps. Integrates with Cloudflare Quick Tunnels and ngrok for internet-accessible remote control, plus file operations via Storage Access Framework and camera/notification management via Android system APIs.

Scores updated daily from GitHub, PyPI, and npm data. How scores work