Developer Tools

Dinucci MCP

Control iOS and Android devices via Model Context Protocol. Integrates with Claude, Cursor, and any compatible IDE.

Learn about MCP View ecosystem
mobile-mcp — zsh
$ npx @anthropic-ai/create-mcp ✓ Installing @mobilenext/mobile-mcp... ✓ MCP server started on port 3000   > mobile_take_screenshot Screenshot saved: screen_001.png (1080x2340)   > mobile_list_apps ├── com.whatsapp (WhatsApp) — Running ├── com.instagram.android (Instagram) — Stopped └── com.spotify.music (Spotify) — Running   > mobile_click --x 540 --y 1200 ✓ Tapped at (540, 1200) — Element: "Send Message"   > mobile_type --text "Hello from MCP!" ✓ Typed 20 characters
The Problem

Manual mobile testing does not scale

Developers and QA engineers test mobile apps manually -- plugging in cables, navigating screens, taking screenshots. MCP allows AI agents to do this automatically, turning real devices into tools that Claude, Cursor or any compatible IDE can control directly. No fragile scripts, no heavyweight frameworks, no complex configuration.

Dinucci MCP exposes iOS and Android devices via Model Context Protocol with a single npm command. The AI agent sees the screen, identifies elements, taps, types and navigates -- exactly as a human QA would, but at machine speed and without fatigue. End-to-end tests, flow automation and remote control in a single open source tool.

Features

Mobile automation for the age of agents

iOS + Android

Native support for both platforms. ADB for Android, WebDriverAgent for iOS. Works with real devices and simulators with zero additional configuration.

Smart Screenshot

Screen capture with automatic AI analysis. The agent identifies UI elements, app state and view hierarchy to make contextual decisions.

Claude + Cursor Integration

Native integration with Claude, Cursor and Cline. The AI agent controls the device directly from the IDE, without leaving the development flow.

App Automation

Navigation, taps, swipes, typing and complex gestures. End-to-end automation of any mobile application installed on the connected device.

npm Package

Published on npm with single-command installation. Zero configuration to get started. Automatic updates with semantic versioning.

Open Source

Open source on GitHub with permissive license for commercial use. Community contributions, public issues and transparent roadmap.

Architecture

From npm install to automation in minutes

01

Install

A single npm command installs the MCP server and all dependencies. Automatically detects connected devices via ADB or WebDriverAgent.

02

Connect

The MCP server exposes the device as a tool for AI agents. Claude, Cursor or any MCP client connects and receives the capabilities list.

03

Automate

The AI agent captures the screen, identifies elements, executes actions and validates results. End-to-end flows run automatically with real-time visual feedback.

Tech Stack

Engineered for developers

Node.js + TypeScript ADB (Android) WebDriverAgent (iOS) Model Context Protocol npm Published Open Source

Ready for mobile automation with AI agents?

Dinucci MCP turns iOS and Android devices into tools that AI agents control directly. One command to install, zero configuration to start.

Learn about MCP