Open source — works with any MCP client

Give AI Eyes and Hands on Your Devices

ScreenMCP connects AI assistants to real phone and PC screens via the Model Context Protocol. Screenshots, taps, swipes, typing, camera — your AI sees and controls the screen like a human.

Free tier available · No credit card required · Open source

How it works

Three steps to connect your AI to any device screen.

1

Install the app

Download ScreenMCP on Android, Windows, macOS, or Linux. Enable the accessibility service once and you're ready.

2

Connect via MCP

Add ScreenMCP as an MCP server in Claude Desktop, Cursor, or any compatible client. Paste your API key and connect.

3

AI controls the device

Your AI can now see the screen, tap, type, scroll, capture camera, and navigate any app in real time.

Everything your AI can do

All commands available on all plans. Limits are by usage, not features.

Screenshot

Capture the current screen as a WebP image. Adjustable quality and max resolution.

Tap & Long Press

Tap at any screen coordinate. Long press for context menus and hold-to-activate elements.

Drag & Swipe

Drag from one point to another. Use for swipe gestures, sliders, and drag-and-drop.

Scroll

Scroll any scrollable view up, down, left, or right with configurable distance.

Type Text

Insert text into any focused input field. Works with any app and keyboard.

Get Text / Clipboard

Read selected text, get clipboard content, select all, copy, and paste.

Navigation Buttons

Trigger Back, Home, and Recents (app switcher) — just like pressing physical buttons.

UI Tree (Accessibility)

Get the full accessibility tree of the current screen — element labels, bounds, clickable state, and hierarchy.

Camera Capture

Capture a photo from the front or rear camera. Useful for scanning documents, reading physical displays, or monitoring environments.

Play Audio

Play a sound or audio file on the device. Useful for alerts, TTS output, or notification sounds triggered by AI.

Multi-device

Connect multiple phones and desktops simultaneously. AI picks the right device by ID or name.

API Keys

Generate and manage API keys for programmatic access. Works with any MCP client or custom integration.

Supported platforms

AndroidStable

Android 8.0+, AccessibilityService, all commands

WindowsStable

Win32 UI tree, system tray, all commands

LinuxBeta

X11, wmctrl/xdotool

macOSBeta

CoreGraphics, menu bar

Use cases

AI Home Automation

Pair with OpenClaw or any home assistant. Monitor device screens alongside cameras and control smart-home UIs automatically.

Automated Mobile Testing

QA teams run test suites with AI. The AI sees the screen and adapts to layout changes, no brittle selectors needed.

Personal Assistant

Let AI manage your phone — book rides, order food, reply to messages, navigate complex app workflows hands-free.

Accessibility

Help users who have difficulty operating phones. Voice-driven AI can see and interact with any app on their behalf.

Development & Debugging

See live device state while coding. AI captures screenshots, inspects UI trees, and tests interactions during development.

Enterprise Automation

Manage fleets of devices controlled by AI. Automate repetitive workflows across dozens of phones with centralized API key management.

OpenClaw Voice Nudges

Generate speech with AI TTS and play it on your phone/desktop via ScreenMCP play_audio. Ideal for OpenClaw reminders, proactive prompts, and spoken guidance when users are not looking at the screen.

Works with your AI tools

Any MCP-compatible client connects out of the box.

Claude DesktopCursorClaude CodeOpenClawAny MCP Client

FAQ

Connect your AI to a real device screen today

Free tier. No credit card. Open source.