Give your AI agent
eyes on mobile

MobAI connects AI coding tools to iOS and Android devices. Screenshots, UI trees, taps, swipes — all through MCP or HTTP.

macOS, Windows & Linux · Free tier available · No credit card

Works with Claude Code Cursor Windsurf Codex or any MCP / HTTP client
MobAI desktop app showing AI agent controlling an Android device with Claude Code integration, MCP server, and HTTP API

iOS Devices

Physical, USB

iOS Simulators

Xcode

Android Devices

Physical, USB

Android Emulators

AVD, Genymotion

What you get

One app that handles device connection, bridge installation, and exposes everything through MCP and HTTP.

MCP Server

Native MCP integration for any AI coding tool

One command: npx mobai-mcp. Works with Claude Code, Cursor, Windsurf, and anything MCP-compatible.

HTTP API

40+ REST endpoints with OpenAPI spec

Screenshots, UI trees, taps, swipes, typing, app management. Language-agnostic, works from Python, Go, JS, curl.

DSL Engine

Batch execution with smart element matching

20+ action types, conditional logic, retry strategies, and assertions. Run complex multi-step automations in a single call.

Web + Native

Automate both native UI and web content

CSS selectors, JS execution, DOM manipulation for Safari/Chrome/WebViews alongside accessibility-based native automation.

Perf Metrics

Real-time CPU, memory, FPS, network, battery

Anomaly detection with health scoring. Stream via SSE or get LLM-optimized summaries.

Local Only

Everything runs on your machine

Screenshots, UI data, automation — nothing leaves your computer. Credentials stored in OS keychain.

Setup

  1. 1.

    Download MobAI

    Native desktop app for macOS, Windows, or Linux. Handles device detection and bridge installation automatically.

  2. 2.

    Connect a device

    Plug in Android or iOS via USB, or launch an emulator/simulator. Auto-detected instantly.

  3. 3.

    Point your AI at it

    $ npx mobai-mcp

    Or use the HTTP API at localhost:8686/api/v1

API at a glance

Full OpenAPI docs at /api/v1/docs

HTTP REST

# devices
GET  /api/v1/devices
GET  /api/v1/devices/{id}/ui-tree
POST /api/v1/devices/{id}/tap
     { "index": 12 }
POST /api/v1/devices/{id}/type
     { "text": "Hello" }

DSL Batch

POST /api/v1/devices/{id}/dsl/execute

{
  "actions": [
    { "action": "tap", "selector": "Settings" },
    { "action": "tap", "selector": "Wi-Fi" },
    { "action": "assert", "selector": "Wi-Fi", "state": "visible" }
  ]
}

Two ways to connect

Pick the one that fits your workflow. Both give full device control.

MCP Server

For AI coding tools

Claude Code, Cursor, Windsurf, Cline — anything MCP-compatible.

$ npx mobai-mcp
$ claude plugin add mobai-marketplace/mobai

HTTP API

For custom integrations

40+ REST endpoints. OpenAPI spec. Use from any language or tool.

base: localhost:8686/api/v1

Screenshots, UI trees, taps, swipes, typing, DSL execution, web automation, perf metrics

More than an API

MobAI is a desktop app with screen mirroring, element picker, and a built-in AI agent.

Screen Mirroring & Element Picker

See your device mirrored in real-time. Click any element to inspect properties, bounds, and accessibility info.

Built-in AI Agent

Run tasks directly from the app. Bring your own API key from OpenAI, Anthropic, or Google. Natural language commands.

Context Pipeline

Element Picker captures context → AiBridge injects it into Claude Code, Codex, or Gemini CLI. No copy-paste. AiBridge →

Auto Bridge Setup

Handles bridge installation and device pairing automatically. Just plug in and go.

Pricing

Start free. Open source tools are always free regardless of plan.

Free

$0

Devices1
API calls100/day
Machines1
Offline mode

Plus

$4.99/mo

Devices1
API callsUnlimited
Machines1
Offline mode

Pro

$9.99/mo or $99/yr

DevicesUnlimited
API callsUnlimited
Machines3
Offline mode7 days

Need enterprise features? Contact us

Chrome Extension

ContextBox

Select any element on a webpage — capture its HTML, CSS, screenshot, and surrounding context. Send it to your AI coding assistant via AiBridge or clipboard.

Element picker

Hover and click any element on any page

Structured capture

HTML, computed styles, and screenshot in one click

AI-ready output

Send to AiBridge or copy as a structured prompt

View on GitHub →

FAQ

MobAI is a desktop application that gives AI coding agents like Claude Code, Codex, and Cursor eyes on mobile devices. It provides an MCP server and HTTP API that lets AI agents take screenshots, read UI elements, tap, swipe, and type on Android and iOS devices.

MobAI works with any AI tool that supports MCP (Model Context Protocol) including Claude Code, or any tool that can make HTTP requests. This includes Cursor, Windsurf, Codex, and custom AI agent frameworks. The HTTP API makes it universally compatible.

Yes, MobAI supports both Android and iOS devices. You can connect physical devices via USB or use emulators and simulators. The same API works identically across both platforms.

MobAI offers a free tier with 1 device and 100 API calls per day (requires internet connection). The Pro plan at $9.99/month unlocks unlimited devices, API calls, and 7-day offline mode for professional use.

No. MobAI runs entirely on your local machine. Screenshots, UI data, and all automation happens locally. The only network communication is for license validation and updates. Your app data never leaves your computer.

MobAI is available for macOS, Windows, and Linux. It's a native desktop application that you download and run locally.

MobAI enables AI-powered mobile app testing by connecting AI coding agents to your mobile devices. Download MobAI, connect your Android or iOS device, then use natural language commands through Claude Code, Cursor, or any MCP-compatible tool. The AI agent can see your app's UI, tap elements, type text, and verify behavior - all automatically. This "vibe coding" approach lets you test mobile apps 2-3x faster than traditional methods.

An MCP (Model Context Protocol) server is a bridge that connects AI coding tools to external capabilities. MobAI's MCP server gives AI agents like Claude Code, Cursor, and Windsurf the ability to control mobile devices. Through the MCP protocol, AI agents can take screenshots, read UI accessibility trees, perform taps and swipes, and run automated tests on both iOS and Android devices.

Yes! MobAI provides a dedicated plugin for Claude Code that enables mobile app testing directly from your terminal. Install the MobAI plugin, connect your device, and Claude Code can see your mobile screen, tap UI elements, type text, and verify app behavior. This enables a true "vibe coding" workflow where you can develop and test mobile apps without leaving Claude Code.

Unlike traditional testing frameworks like Appium or XCTest that require writing test scripts in code, MobAI enables AI-powered testing through natural language. Simply describe what you want to test and the AI agent handles the automation. MobAI also provides a unified API for both iOS and Android (unlike XCTest which is iOS-only), integrates directly with AI coding tools via MCP, and runs entirely locally for privacy. It's designed for the "vibe coding" era where AI agents handle the implementation details.

Vibe coding is a development approach where you describe what you want in natural language and AI agents handle the implementation. For mobile apps, MobAI enables vibe coding by giving AI agents visual feedback - they can see your app's screen, understand the UI, and verify changes work correctly. Instead of manually testing every code change, describe the expected behavior and let the AI agent verify it automatically on real devices or emulators.

ios-builder is our open source CLI tool that lets you build iOS apps from Windows, Linux, or any platform without a Mac. It uses GitHub Actions to run Xcode builds remotely and MobAI to deploy to real iOS devices. It supports native Swift, Flutter, React Native, and Expo projects with hot reload capabilities.

Yes. MobAI provides full web automation for Safari, Chrome, and WebViews on both iOS and Android. You can use CSS selectors, execute JavaScript, manipulate the DOM, navigate pages, and interact with web content. Native UI (address bar, tabs) uses accessibility-based automation while page content uses web debugging protocols.

DSL (Domain Specific Language) execution lets you run precise, multi-step automation scripts in a single API call. It supports 20+ action types including taps, swipes, typing, assertions, conditional logic, and retry strategies. Smart element matching uses text, labels, types, and spatial relationships to find elements reliably.

AiBridge is our open source tool that wraps terminal-based AI assistants (Claude Code, Codex, Gemini CLI) and exposes an HTTP API for external text injection. Combined with our Chrome extension ContextBox, you can visually select UI elements in a browser and send that context directly to your AI assistant.