conorluddy/ios-simulator-skill

What it solves

This project provides a suite of automation scripts designed to bridge the gap between AI agents and iOS development environments. It solves the problem of fragile, high-token-cost interactions with the iOS Simulator by providing semantic, accessibility-based controls that are optimized for AI agent consumption.

How it works

The project uses a collection of 29 Python scripts that interface with standard Apple development tools like xcodebuild, xcrun simctl, and Meta's idb (iOS Device Bridge). Instead of relying on brittle pixel-based coordinates, the scripts use iOS accessibility APIs to identify and interact with UI elements by their semantic labels. It features "progressive disclosure," where build logs and test results are summarized into concise, token-efficient outputs, with full details available only upon request.

Who it’s for

This tool is intended for developers and AI agents working on iOS app automation, testing, and CI/CD pipelines who need a reliable, programmatic way to build, test, and navigate iOS apps within a simulator environment.

Highlights

• Semantic Navigation: Uses accessibility trees to find and interact with UI elements, making automation robust against UI layout changes.
• Token Optimization: Automatically compresses screenshots and truncates logs to minimize token usage for LLM-based agents.
• Progressive Disclosure: Provides concise summaries of build and test results, allowing agents to drill down into specific error logs only when necessary.
• Comprehensive Tooling: Includes 29 scripts covering the entire lifecycle, from simulator management and app launching to accessibility auditing and visual regression testing.