AvdLee/SwiftUI-Agent-Skill
Add expert SwiftUI Best Practices guidance to your AI coding tool (Agent Skills open format).
What it solves
This project provides a specialized knowledge base and toolset for AI coding assistants to help developers build high-performance, modern SwiftUI applications. It prevents AI agents from providing outdated or inefficient SwiftUI code by providing them with concise, actionable references on state management, view composition, and the latest iOS APIs (including iOS 26+ Liquid Glass).
How it works
The project is packaged as an "Agent Skill" using an open format compatible with tools like Claude Code, Cursor, and ChatGPT. It consists of a library of Markdown reference files that the AI agent loads on demand to avoid bloating the context window. Additionally, it includes a Python-based toolchain that wraps xctrace to allow AI agents to record and analyze Instruments traces, helping them diagnose performance issues like hangs and hitches directly from trace data.
Who it’s for
- Developers and teams using SwiftUI who want to ensure their AI coding tools follow modern best practices.
- iOS/macOS developers who need to profile and analyze app performance using Instruments via an AI agent.
- Developers refactoring legacy SwiftUI code to modern APIs.
Highlights
- On-demand Reference Library: Covers everything from
@Observableand NavigationStack to Swift Charts and accessibility. - Automated Performance Analysis: Includes scripts to record and parse
xctracefiles, allowing agents to identify if a hang is CPU-bound or blocked by I/O. - Broad Tool Compatibility: Supports multiple AI clients including Claude Code, Cursor, Codex, and ChatGPT.
- Modern API Coverage: Includes specific guidance for iOS 26+ Liquid Glass and migration guides from iOS 15+.
Related
- Project
- Project
- Project
- Project
- Project