Xcode MCP Server

What is Xcode MCP Server?

The Xcode MCP Server is an MCP (Model Context Protocol) server that provides comprehensive Xcode integration for AI assistants. It enables AI agents to interact with Xcode projects, manage iOS simulators, and perform various Xcode-related tasks with enhanced error handling and support for multiple project types (standard projects, workspaces, and Swift packages).

How to use Xcode MCP Server?

Install via the automated setup script (./setup.sh) or manually: clone the repo, run npm install and npm run build, then create a .env file with your projects base directory. Start the server with npm start. Connect your AI assistant to the server URL (default http://localhost:3000). Configuration can also be done via command-line arguments (e.g., --projects-dir, --port, --debug).

Key features of Xcode MCP Server

• Set active projects, create new projects from templates, add files
• Read/write files, search content with patterns and regex
• Build projects, run tests, analyze code, archive for distribution
• CocoaPods integration: install, update, add/remove pods
• Swift Package Manager: init, add/remove dependencies, generate DocC
• iOS simulator tools: list, boot, install apps, screenshots, recordings
• Xcode utilities: compile asset catalogs, generate app icons, switch Xcode versions

Use cases of Xcode MCP Server

• AI-assisted project creation and file management for iOS/macOS apps
• Automated build, test, and analyze workflows in CI or development
• Managing CocoaPods dependencies through natural language commands
• Controlling iOS simulators (boot, install, record) via AI agents
• Preparing and validating archives for App Store submission

FAQ from Xcode MCP Server

What are the prerequisites?

macOS with Xcode 14.0 or higher, Node.js 16+, npm or yarn, Swift 5.5+ for Swift Package Manager features. CocoaPods is optional and requires Ruby.

How do I configure the server?

You can configure via environment variables in a .env file (e.g., PROJECTS_BASE_DIR, DEBUG, PORT) or via command-line arguments (e.g., --projects-dir, --port, --debug). The PROJECTS_BASE_DIR is required.

Where does the server run and how do AI assistants connect?

The server runs locally on the configured port (default 3000). AI assistants that support the Model Context Protocol connect to the server URL (typically http://localhost:3000).

Is there automated setup available?

Yes, the included setup.sh script verifies prerequisites, installs dependencies, prompts for configuration, and optionally integrates with Claude Desktop.

What about security and path restrictions?

The server uses a path management component that validates all file access against allowed directories. The ALLOWED_PATHS configuration parameter lets you specify additional directories that the server is permitted to access.