MacOS Trace

macOS-Trace

[English](README.md) | [中文](README_zh.md)

[](https://agentskills.io) [](https://skills.sh/kmgcc/macOS-Trace) [](https://developer.apple.com/macos/) [](https://developer.apple.com/xcode/) [-3776AB.svg)](https://www.python.org/) [](https://opensource.org/licenses/MIT)

Looking for iOS and iPadOS profiling on physical devices or simulators? See [iOS-Trace](https://github.com/kmgcc/iOS-Trace).

Autonomous, closed-loop application performance optimization engine for macOS applications using `xctrace` and Xcode Instruments.

Designed for AI coding agents (Claude Code, OpenAI Codex, Cursor, Google Antigravity, GitHub Copilot) and macOS systems engineers. It completely eliminates manual Instruments GUI interaction. From a single prompt, an agent can align on optimization goals with the user, capture headless diagnostic traces, implement targeted code fixes, and re-test with differential benchmarking across multiple iterations until the user's requirements are met.

---

The Autonomous Optimization Loop

+-------------------------------------------------------------------------+
|                  The Autonomous Optimization Loop                        |
|                                                                         |
|  1. Goal Alignment ──> 2. Diagnostic Trace ──> 3. Targeted Code Fix     |
|         ^                                                 │             |
|         │                                                 ▼             |
|         └────── Iterate if Target Not Met