Repolode
Submit
zhongerxin

zhongerxin/iPhone-use

A Codex, WorkBuddy and Doubao plugin that lets an AI assistant operate a real iPhone over USB via WebDriverAgent, with a live screen sidebar.

View on GitHub
Stars
4,060
Forks
407
Open issues
18
Last commit
2 days ago

What is it?

iPhone Use connects AI assistants such as Codex, WorkBuddy and Doubao Work to a physical iPhone connected over USB. Users describe a task in natural language and the assistant can open apps, read the screen, tap, swipe, type text and collect list items, while a sidebar widget shows the live phone screen.

Under the hood it runs a local MCP service that talks to WebDriverAgent (WDA), the on-device automation runner from the Appium project. The project includes setup and usage skills that guide installation, signing, building and launching WDA, and it reuses existing healthy WDA services when possible. When element lookup fails, the assistant falls back to screenshots and coordinate taps.

The plugin requires macOS, full Xcode, Python 3.9+, Node.js and a USB-connected iPhone signed with the user's own Apple developer team. Anonymous usage analytics are on by default for the Codex version and can be disabled with an environment variable.

Key features

  • 01Operate an iPhone over USB from Codex, WorkBuddy or Doubao Work using natural language.
  • 02Read compact control trees and screenshots, then tap, swipe, and type Unicode text.
  • 03Show a live phone screen in the assistant sidebar with tap and drag feedback.
  • 04Guided setup that builds and launches WebDriverAgent with your own Apple signing team.
  • 05Batch actions, scroll-to-find and list collection to reduce model round trips.
  • 06Screenshot and coordinate-tap fallback when element lookup fails.

Use cases

  • Having an AI assistant fill in or check forms and notes on a physical iPhone.
  • Automating repetitive app workflows that have no API, such as searching and reviewing lists.
  • Testing or demonstrating an iOS app flow driven by natural-language instructions.
  • Collecting visible list items from an app by scrolling and de-duplicating results.

Topics

#ios#iphone#automation#mcp#webdriveragent#appium#codex#usb

FAQ

What is iPhone-use?
A Codex, WorkBuddy and Doubao plugin that lets an AI assistant operate a real iPhone over USB via WebDriverAgent, with a live screen sidebar. iPhone Use connects AI assistants such as Codex, WorkBuddy and Doubao Work to a physical iPhone connected over USB. Users describe a task in natural language and the assistant can open apps, read the screen, tap, swipe, type text and collect list items, while a sidebar widget shows the live phone screen.
Is iPhone-use free to use? What license does it have?
iPhone-use is open source under the MIT license. Check the license terms before using it in commercial products.
Is iPhone-use still maintained?
Actively maintained. The last commit to iPhone-use was on Oct 16, 2026.
What can I use iPhone-use for?
Having an AI assistant fill in or check forms and notes on a physical iPhone. Automating repetitive app workflows that have no API, such as searching and reviewing lists. Testing or demonstrating an iOS app flow driven by natural-language instructions. Collecting visible list items from an app by scrolling and de-duplicating results.
Where can I get iPhone-use?
The source code and installation instructions are on GitHub: https://github.com/zhongerxin/iPhone-use

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects