An MCP server that gives AI assistants rich tooling to create, modify and build Xcode, CocoaPods and Swift-PM projects, drive iOS simulators and run other Xcode CLI utilities.
https://github.com/r-huijts/xcode-mcp-serverIf you're building iOS apps, you know the drill: ask Claude to generate some Swift code, copy it to Xcode, realize it needs tweaks for your specific project structure, switch back to Claude to explain your setup, get updated code, rinse and repeat. Now imagine your AI assistant could just directly create the project, add the files, install dependencies, and even run the build.
That's exactly what this MCP server delivers.
Your AI assistant knows Swift and SwiftUI inside and out, but it's completely blind to your actual Xcode environment. It can't see your project structure, doesn't know what pods you have installed, can't check if your code actually compiles, and definitely can't spin up a simulator to test the result.
This server bridges that gap by giving AI assistants direct access to your entire Xcode toolchain through the Model Context Protocol.
Instead of the usual back-and-forth, you can now:
"Create a new iOS project called WeatherApp, add Alamofire as a dependency, and build a network service that fetches weather data"
Your AI assistant will:
"Add a new SwiftUI view for user settings, include it in the main target, and run tests"
Your assistant handles:
This isn't just basic file operations. The server provides deep integration across your entire iOS development workflow:
Project Management: Create projects from templates, manage workspaces, handle project configurations, and organize files within Xcode's group structure.
Build & Test Automation: Run builds with specific schemes and configurations, execute test suites with detailed failure reporting, and analyze code for potential issues.
Dependency Management: Install and update CocoaPods, manage Swift Package Manager dependencies with version requirements, and handle dependency resolution conflicts.
Simulator Control: List available simulators, boot specific device configurations, install and launch apps, capture screenshots, and even record videos for documentation.
Real Xcode Workflows: Use xcrun commands, compile asset catalogs, generate app icons, trace performance, and prepare archives for App Store submission.
The included setup script handles everything:
./setup.sh
It verifies your macOS environment, installs dependencies, builds the TypeScript code, creates your configuration, and even sets up Claude Desktop integration. You're up and running in minutes, not hours.
This server understands the complexity of modern iOS projects. It handles standard .xcodeproj files, complex .xcworkspace setups with multiple projects, and Swift Package Manager projects. Whether you're working on a simple app or a multi-target enterprise project with frameworks and extensions, it adapts to your structure.
The path validation system keeps operations secure while giving your AI assistant the access it needs. Debug logging helps troubleshoot when things go wrong, and comprehensive error handling provides clear feedback instead of cryptic failures.
Since it implements MCP, it works with any AI assistant that supports the protocol. Claude Desktop integration is particularly smooth – add a few lines to your config and your AI assistant immediately gains access to your entire Xcode environment.
Your development workflow transforms from constant context-switching to natural conversation about your actual project. No more copying code snippets back and forth or explaining your project structure repeatedly.
./setup.sh for guided setupThe server runs locally, so your code never leaves your machine. It's MIT licensed and actively maintained, with comprehensive documentation and examples to get you productive immediately.
This is iOS development with AI assistance that actually works with your toolchain, not around it.