xcf - AI-Powered Xcode Automation

What is xcf?

xcf is an MCP server that enables AI assistants to control Xcode. It integrates with Cursor, VSCode, and Claude Desktop, allowing users to build, run, and manage Xcode projects through natural language or function calls.

How to use xcf?

Configure by adding the server to your MCP config file (~/.cursor/mcp.json or ~/Library/Application Support/Claude/claude_desktop_config.json) with a stdio transport pointing to the xcf binary, then restart your AI tool. Activate with the use xcf command, then use commands like show, open #, build, and run.

Key features of xcf

• Execute Xcode build and run commands via AI
• File access through snippets with smart path resolution
• Security with automatic workspace boundary enforcement
• Optional environment variables for project pre-selection
• Dedicated MCP tools for AI function calls and help

Use cases of xcf

• Build and run Xcode projects hands-free using natural language
• Explore and retrieve Swift file contents during AI conversations
• Manage multiple open Xcode projects from a single session

FAQ from xcf

Which AI tools does xcf work with?

It works with Cursor, VSCode, and Claude Desktop.

How do I configure xcf?

Add the server to your MCP config JSON with "type": "stdio" and the path to the xcf binary, then restart your AI tool.

What optional environment variables can I set?

XCODE_PROJECT_FOLDER_optional and XCODE_PROJECT_optional let you pre-select a specific project and define a secure workspace boundary.

What transport does xcf use?

It uses the STDIO transport protocol.

How does xcf find files when using snippets?

It uses smart path resolution, searching the current directory, workspace folder, subdirectories (one level deep), and the project directory when the exact path isn't found.