iOS Automation MCP Server

iOS Automation MCP Server

A Model Context Protocol server that enables AI assistants to interact with iOS simulators, perform accessibility testing, manage apps, and automate complex iOS workflows.

Category
Visit Server

README

iOS Automation MCP Server 🚀

A comprehensive Model Context Protocol (MCP) server for iOS development automation. This Python implementation enables AI assistants to interact with iOS simulators, perform accessibility testing, manage apps, and automate complex iOS workflows.

🎯 System Architecture

graph TB
    subgraph "Claude Desktop Environment"
        CD[Claude Desktop]
        CDConfig[claude_desktop_config.json<br/>User Configuration]
        CDLogs[~/Library/Logs/Claude/]
    end
    
    subgraph "MCP Server Layer"
        MCPServer[iOS MCP Server<br/>ios_mcp_server.py]
        VEnv[Python Virtual Environment<br/>ios_mcp_env/]
        MCPTools[13 iOS Automation Tools]
    end
    
    subgraph "iOS Simulator Layer"
        Simulator[iOS Simulator<br/>Any iPhone/iPad Model]
        TestApps[Test Applications<br/>User Apps & System Apps]
        SimulatorTools[simctl Commands]
    end
    
    subgraph "macOS System Layer"
        AppleScript[AppleScript<br/>UI Automation]
        Accessibility[Accessibility Permissions<br/>Terminal.app]
        Screenshots[Screenshot Storage<br/>/tmp/]
    end
    
    subgraph "Tool Categories"
        SimMgmt[Simulator Management<br/>• list_simulators<br/>• boot_simulator<br/>• shutdown_simulator<br/>• get_simulator_state]
        AppMgmt[App Management<br/>• launch_app<br/>• terminate_app<br/>• install_app<br/>• list_installed_apps]
        UIAuto[UI Automation<br/>• tap_coordinate<br/>• tap_element<br/>• type_text<br/>• get_accessibility_tree]
        Capture[Screen Capture<br/>• take_screenshot]
    end
    
    %% Connections
    CD --> MCPServer
    CDConfig -.-> CD
    MCPServer --> VEnv
    MCPServer --> MCPTools
    MCPServer --> CDLogs
    
    MCPTools --> SimMgmt
    MCPTools --> AppMgmt
    MCPTools --> UIAuto
    MCPTools --> Capture
    
    SimMgmt --> SimulatorTools
    AppMgmt --> SimulatorTools
    UIAuto --> AppleScript
    Capture --> SimulatorTools
    
    SimulatorTools --> Simulator
    AppleScript --> Accessibility
    AppleScript --> Simulator
    Simulator --> TestApps
    Capture --> Screenshots

Data Flow:

  1. Claude DesktopMCP Server (JSON-RPC over stdio)
  2. MCP Serversimctl/AppleScript (Command execution)
  3. System ToolsiOS Simulator (Direct automation)
  4. Results/ScreenshotsClaude Desktop (Response data)

🏗️ High-Level Architecture

Design Principles

The iOS MCP Server follows a layered architecture designed for:

  • Reliability: Robust error handling and graceful failure recovery
  • Extensibility: Modular tool design for easy feature additions
  • Performance: Asynchronous operations and efficient resource usage
  • Security: Sandboxed execution with controlled system access

Core Components

1. MCP Protocol Layer

┌─────────────────────────────────────────────────────────┐
│                Claude Desktop                           │
│  ┌─────────────────────────────────────────────────┐    │
│  │            AI Assistant                         │    │
│  │  • Natural language processing                  │    │
│  │  • Intent recognition                          │    │
│  │  • Context management                          │    │
│  └─────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────┘
                            │
                   JSON-RPC over stdio
                            │
┌─────────────────────────────────────────────────────────┐
│                 MCP Server                              │
│  ┌─────────────────────────────────────────────────┐    │
│  │        Protocol Handler                         │    │
│  │  • JSON-RPC message parsing                     │    │
│  │  • Tool discovery & registration               │    │
│  │  • Resource management                         │    │
│  │  • Error handling & logging                    │    │
│  └─────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────┘

2. Tool Orchestration Layer

┌─────────────────────────────────────────────────────────┐
│                Tool Manager                             │
│                                                         │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐      │
│  │  Simulator  │  │     App     │  │     UI      │      │
│  │ Management  │  │ Management  │  │ Automation  │      │
│  │   Tools     │  │   Tools     │  │   Tools     │      │
│  └─────────────┘  └─────────────┘  └─────────────┘      │
│                                                         │
│  ┌─────────────────────────────────────────────────┐    │
│  │           Screen Capture Tools                  │    │
│  └─────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────┘

3. System Integration Layer

┌─────────────────────────────────────────────────────────┐
│               System Adapters                           │
│                                                         │
│  ┌─────────────┐  ┌─────────────┐  ┌─────────────┐      │
│  │   simctl    │  │ AppleScript │  │ File System │      │
│  │  Command    │  │ Automation  │  │  Operations │      │
│  │ Executor    │  │   Engine    │  │             │      │
│  └─────────────┘  └─────────────┘  └─────────────┘      │
└─────────────────────────────────────────────────────────┘
                            │
                    Native macOS APIs
                            │
┌─────────────────────────────────────────────────────────┐
│                iOS Simulator                            │
│  ┌─────────────────────────────────────────────────┐    │
│  │            Target Applications                  │    │
│  │  • User apps under test                        │    │
│  │  • System apps and services                    │    │
│  │  • UI accessibility hierarchy                  │    │
│  └─────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────┘

Architectural Patterns

Command Pattern Implementation

Each tool implements the command pattern for:

  • Encapsulation: Tool logic is self-contained
  • Undo/Redo: Future support for operation reversal
  • Logging: Comprehensive audit trail
  • Error Recovery: Graceful handling of failures

Adapter Pattern for System Integration

# Abstract interface for all system operations
class SystemAdapter:
    async def execute_command(self, command: str) -> Result
    
# Concrete implementations
class SimctlAdapter(SystemAdapter):
    # Handles iOS Simulator operations
    
class AppleScriptAdapter(SystemAdapter):
    # Handles UI automation via AppleScript

Factory Pattern for Tool Creation

class ToolFactory:
    def create_tool(self, tool_type: str) -> Tool:
        # Dynamic tool instantiation based on type

Cross-Cutting Concerns

Error Handling Strategy

┌─────────────────────────────────────────────────────────┐
│                 Error Handling                          │
│                                                         │
│  Client Error     │  Server Error     │  System Error   │
│  ┌─────────────┐  │  ┌─────────────┐  │  ┌─────────────┐ │
│  │ Validation  │  │  │  Internal   │  │  │   macOS     │ │
│  │   Errors    │  │  │ Processing  │  │  │ System API  │ │
│  │             │  │  │   Errors    │  │  │   Errors    │ │
│  └─────────────┘  │  └─────────────┘  │  └─────────────┘ │
│        │          │        │          │        │         │
│        └──────────┼────────┼──────────┼────────┘         │
│                   │        │          │                  │
│  ┌─────────────────────────────────────────────────────┐ │
│  │             Centralized Error Handler              │ │
│  │  • Categorization and severity assessment          │ │
│  │  • Structured logging with context                 │ │
│  │  • User-friendly error messages                    │ │
│  │  • Automatic retry for transient failures          │ │
│  └─────────────────────────────────────────────────────┘ │
└─────────────────────────────────────────────────────────┘

Logging & Observability

  • Structured logging: JSON format with contextual metadata
  • Performance metrics: Execution time tracking for each tool
  • Audit trail: Complete record of all operations
  • Debug information: Detailed system state capture

Resource Management

  • Connection pooling: Efficient use of system resources
  • Memory management: Automatic cleanup of temporary files
  • Concurrent execution: Thread-safe operations where applicable
  • Rate limiting: Protection against excessive API calls

Security Model

Principle of Least Privilege

┌─────────────────────────────────────────────────────────┐
│                Security Boundaries                      │
│                                                         │
│  MCP Server Sandbox                                     │
│  ┌─────────────────────────────────────────────────┐    │
│  │  • Limited file system access                  │    │
│  │  • Controlled subprocess execution             │    │
│  │  • No network access (stdio only)              │    │
│  │  • Isolated virtual environment               │    │
│  └─────────────────────────────────────────────────┘    │
│                            │                            │
│         Controlled Escalation                           │
│                            │                            │
│  macOS System Permissions                               │
│  ┌─────────────────────────────────────────────────┐    │
│  │  • Accessibility API access                    │    │
│  │  • iOS Simulator control                       │    │
│  │  • Screen capture permissions                  │    │
│  │  • AppleScript execution rights                │    │
│  └─────────────────────────────────────────────────┘    │
└─────────────────────────────────────────────────────────┘

Scalability Considerations

Horizontal Scaling

  • Multiple simulator support: Concurrent operations across simulators
  • Load balancing: Distribution of operations across available resources
  • Session management: Isolated contexts for different automation tasks

Vertical Scaling

  • Async operations: Non-blocking I/O for improved throughput
  • Resource optimization: Efficient memory and CPU usage
  • Caching: Intelligent caching of expensive operations

Extension Points

The architecture supports extension through:

  1. Custom Tool Development: Plugin-style tool addition
  2. Protocol Extensions: Additional MCP resource types
  3. System Adapters: Support for additional automation backends
  4. Output Formats: Customizable response formatting
  5. Integration Hooks: CI/CD and external system integration

🔄 Interaction Sequence

The following sequence diagram shows the detailed interaction flow for a typical iOS automation workflow:

iOS MCP Server Sequence Diagram

This diagram illustrates:

  • MCP Communication: JSON-RPC protocol between Claude Desktop and the iOS MCP Server
  • Tool Execution: How iOS automation commands are processed and executed
  • System Integration: Interaction with iOS Simulator via simctl and AppleScript
  • Response Flow: How results and screenshots are returned to the AI assistant

🎬 Live Demo

See the iOS MCP Server in action:

iOS MCP Server Demo

Demo showing Claude Desktop automatically controlling iOS Simulator: launching apps, taking screenshots, tapping UI elements, and typing text through natural language commands.

🎥 Watch on YouTube | 📥 Download Original (43MB)

📱 Features

Simulator Management

  • ✅ List all available iOS simulators
  • ✅ Boot/shutdown simulators programmatically
  • ✅ Take screenshots of simulator screens
  • ✅ Tap at specific coordinates
  • ✅ Get real-time simulator state

App Management

  • ✅ Install apps on simulators
  • ✅ Launch apps with bundle identifiers
  • ✅ Monitor app lifecycle

Accessibility & Testing

  • ✅ Extract accessibility tree from running apps
  • ✅ Parse UI hierarchy for automated testing
  • ✅ Enable AI-driven UI interaction

Logging & Monitoring

  • ✅ Access simulator system logs
  • ✅ Real-time resource monitoring
  • ✅ Comprehensive error handling

🛠 Requirements

  • macOS 13.0+
  • Python 3.8+
  • Xcode (for iOS Simulator)
  • iOS Simulator (included with Xcode)

⚡ Quick Start

🚀 Option 1: Automated Setup (Recommended)

git clone <your-repo-url>
cd mcp-server-demo-proj

# Run the automated setup script
python3 automated_setup.py

The automated setup will:

  • ✅ Create and configure virtual environment
  • ✅ Install all dependencies
  • ✅ Configure Claude Desktop automatically
  • ✅ Check for Cursor IDE compatibility
  • ✅ Create launcher scripts
  • ✅ Run comprehensive tests

Then just restart Claude Desktop and you're ready to go!

🔧 Option 2: Manual Setup

1. Clone and Setup

git clone <your-repo-url>
cd mcp-server-demo-proj
python3 -m venv ios_mcp_env
source ios_mcp_env/bin/activate
pip install -r requirements.txt

2. Test the Server

# Test server imports correctly
python3 -c "import ios_mcp_server; print('✅ Server ready!')"

# Run diagnostic script
python3 debug_mcp_setup.py

3. Configure with AI Assistant

Claude Desktop

  1. Copy the example configuration:
cp claude_desktop_config.example.json claude_desktop_config.json
  1. Update the paths in claude_desktop_config.json to match your project location:
{
  "mcpServers": {
    "ios-automation": {
      "command": "/YOUR/PROJECT/PATH/ios_mcp_env/bin/python3",
      "args": ["/YOUR/PROJECT/PATH/ios_mcp_server.py"],
      "env": {
        "PYTHONPATH": "/YOUR/PROJECT/PATH"
      }
    }
  }
}
  1. Copy to Claude Desktop config location:
cp claude_desktop_config.json "$HOME/Library/Application Support/Claude/claude_desktop_config.json"
  1. Restart Claude Desktop

🔌 MCP Client Compatibility

Currently Supported

  • Claude Desktop - Full native MCP support
  • Manual CLI - Direct server execution for testing

⚠️ Partial/Future Support

  • Cursor IDE - No native MCP support yet (as of 2024)
  • Continue.dev - VS Code extension with potential MCP integration
  • Zed Editor - May add MCP support in future versions

🔧 Using with Cursor IDE

While Cursor doesn't natively support MCP yet, you can:

  1. Use the automated setup to configure everything
  2. Use Claude Desktop alongside Cursor for AI assistance
  3. Run the MCP server manually for testing: ./start_mcp_server.sh

📱 Testing without IDE Integration

# Start server manually for testing
./start_mcp_server.sh

# In another terminal, test specific tools
python3 -c "
import asyncio
import sys
sys.path.insert(0, '.')
import ios_mcp_server
# Test your tools here
"

🚨 Troubleshooting

If the server starts but Claude Desktop doesn't recognize it:

Check Configuration File Location

# Verify the config file exists
ls -la "$HOME/Library/Application Support/Claude/claude_desktop_config.json"

# Check the content
cat "$HOME/Library/Application Support/Claude/claude_desktop_config.json"

Verify Paths in Configuration

Make sure all paths in your config are absolute paths and actually exist:

# Check if Python executable exists
ls -la /YOUR/PROJECT/PATH/ios_mcp_env/bin/python3

# Check if the script exists  
ls -la /YOUR/PROJECT/PATH/ios_mcp_server.py

# Check if the project directory exists
ls -la /YOUR/PROJECT/PATH/

Run Diagnostic Script

# Run the automated diagnostic to check your setup
python3 debug_mcp_setup.py

Test MCP Server Directly

# Test if the server can import without errors
cd /YOUR/PROJECT/PATH
source ios_mcp_env/bin/activate
python3 -c "import ios_mcp_server; print('✅ Server imports successfully')"

Check Claude Desktop Logs

# View Claude Desktop logs for MCP errors
tail -f ~/Library/Logs/Claude/mcp*.log

Common Issues & Solutions

  1. "Connected to Claude Desktop" but tools don't appear

    • Restart Claude Desktop completely (quit and reopen)
    • Clear Claude Desktop cache: rm -rf ~/Library/Caches/Claude
  2. Permission denied errors

    • Make sure Python executable is executable: chmod +x ios_mcp_env/bin/python3
    • Check if Terminal has accessibility permissions (System Preferences → Security & Privacy → Accessibility)
  3. Path not found errors

    • Use absolute paths only (no ~ or relative paths)
    • Verify paths exist and are accessible
  4. ImportError for mcp module

    • Reinstall in virtual environment: pip install --upgrade mcp
    • Check virtual environment activation: which python3 should show your venv path
  5. Server imports but tools don't appear

    • Run the diagnostic script: python3 debug_mcp_setup.py
    • Check Claude Desktop logs for errors

Complete Working Example

Replace with your actual paths:

{
  "mcpServers": {
    "ios-automation": {
      "command": "/Users/yourusername/projects/mcp-server-demo-proj/ios_mcp_env/bin/python3",
      "args": ["/Users/yourusername/projects/mcp-server-demo-proj/ios_mcp_server.py"],
      "env": {
        "PYTHONPATH": "/Users/yourusername/projects/mcp-server-demo-proj"
      }
    }
  }
}

🎯 Available Tools (13 Total)

Simulator Management (4 tools)

Tool Description Parameters
list_simulators List all iOS simulators None
boot_simulator Boot a simulator device_id
shutdown_simulator Shutdown a simulator device_id
get_simulator_state Get real-time simulator status device_id (optional)

App Management (4 tools)

Tool Description Parameters
launch_app Launch app by bundle ID bundle_id, device_id (optional)
terminate_app Terminate running app bundle_id, device_id (optional)
install_app Install app on simulator app_path, device_id (optional)
list_installed_apps List all installed apps device_id (optional)

UI Automation (4 tools)

Tool Description Parameters
tap_coordinate Tap at specific coordinates x, y, device_id (optional)
tap_element Tap UI element by identifier identifier, device_id (optional)
type_text Type text into focused field text, device_id (optional)
get_accessibility_tree Get UI hierarchy device_id (optional), format (optional)

Screen Capture (1 tool)

Tool Description Parameters
take_screenshot Capture simulator screen save_path (optional)

📚 Available Resources

Resource Description URI
Simulator State Current simulator status simulator://current-state
Accessibility Tree Live UI hierarchy accessibility://hierarchy
Simulator Logs System and app logs logs://simulator

💬 Example AI Prompts

Simulator Management

"List all available iOS simulators and boot an iPhone 15 Pro"

App Testing

"Install my app from ~/MyApp.app, launch it, take a screenshot, and show me the accessibility tree"

Automated Workflows

"Boot iPhone 14 simulator, take a screenshot, tap the center of the screen, wait 2 seconds, then take another screenshot to see what changed"

Development Debugging

"Show me the current simulator logs and tell me if there are any crash reports in the last hour"

🔧 Development

Project Structure

mcp-server-demo-proj/
├── ios_mcp_server.py           # Main Python MCP server
├── debug_mcp_setup.py          # Setup diagnostic script
├── requirements.txt            # Python dependencies
├── claude_desktop_config.example.json  # Configuration template
├── ios_mcp_env/               # Virtual environment
├── test_doc_retrival_agent.py # Test automation script
├── test_ios_mcp.sh            # Shell test script
├── mobile_mcp.mov             # Demo video (43MB)
├── sequence_.png              # Sequence diagram
├── README.md                  # This file
└── LICENSE                    # Project license

Key Dependencies

  • MCP Python SDK - Official MCP implementation
  • Apple's simctl - iOS Simulator command line tools
  • AppleScript - macOS UI automation
  • asyncio - Asynchronous Python execution

Running from Source

# Setup virtual environment
python3 -m venv ios_mcp_env
source ios_mcp_env/bin/activate

# Install dependencies
pip install -r requirements.txt

# Test server
python3 -c "import ios_mcp_server; print('✅ Server ready!')"

# Run diagnostic
python3 debug_mcp_setup.py

🚀 Advanced Usage

Custom Tool Development

Add new tools by extending the handle_call_tool function:

elif name == "your_custom_tool":
    # Your implementation here
    return [types.TextContent(type="text", text="Tool executed!")]

Integration with CI/CD

This server can be integrated into your CI/CD pipeline for automated iOS testing:

# GitHub Actions example
- name: Setup iOS MCP Server
  run: |
    python3 -m venv ios_mcp_env
    source ios_mcp_env/bin/activate
    pip install -r requirements.txt
    
- name: Run iOS Automation Tests
  run: |
    source ios_mcp_env/bin/activate
    python3 test_doc_retrival_agent.py

🤝 Contributing

This project can be extended with:

  • Additional iOS automation tools
  • Physical device support (requires additional setup)
  • Advanced accessibility analysis
  • Integration with TestFlight
  • Custom UI testing frameworks

📄 License

Built with ❤️ for iOS development automation using the Model Context Protocol.

🙏 Acknowledgments


Ready to automate your iOS development workflow? Let's build something amazing! 🚀

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured