iPhone Use — Claude Code MCP server: what it does, how to install
Connects an AI assistant to a physical iPhone from your Mac. It reads screen elements as structured text, clicks buttons, and types directly into apps lacking public APIs.
Open on GitHubWhat iPhone Use does
A local Rust background service runs on your Mac, communicating with an XCTest runner on iOS. Rather than capturing raw pixels, the daemon parses accessibility hierarchies in roughly 0.1 s. Completed interactions can be recorded into repeatable sequences that execute later without LLM tokens.
The agent handles app launches by title, verifies element settlement before completing an action, and inputs text up to 20000 characters in one step. Your Mac pointer and windows stay untouched. The connection runs across USB or an encrypted Wi-Fi link.
Initial configuration fails unless the iPhone runs iOS 15 or newer with Developer Mode active and connects via USB to a Mac on macOS 15+. Xcode must be signed in with an Apple ID beforehand.
What it helps with
- audit multi-step sign-in sequences on a physical iOS device
- read private interface text from apps lacking public developer APIs
- insert thousands of characters into mobile documents without manual typing
- automate mobile regression checks across native iOS views
Example prompts for the agent
- Open Health and check the daily step count for this week
- Launch our staging build on the attached phone and report where the login sequence fails
- Export the full transaction list from the open banking screen
How to install iPhone Use
Install
The MCP registry has no launch command for it: see the repository README
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about iPhone Use
How to install iPhone Use in Claude Code?
Plug your unlocked iPhone into the Mac with a USB cable and run curl -fsSL https://raw.githubusercontent.com/leeguooooo/iphone-use/main/install.sh | sh. The script registers the server with Claude Code and configures the helper app.
How does iPhone Use read mobile screens?
It parses the native accessibility tree into text instead of processing visual screenshots. Because of this structure, element detection takes about 0.1 s and works even in views that restrict screen capture.
What are the hardware and OS requirements?
You need a Mac running macOS 15+ with Xcode logged into any Apple ID. The iPhone must run iOS 15 or later with Developer Mode turned on in privacy settings.