Skills

Why we included this project

If you use OpenAI's Codex or a similar agentic coding tool, this collection shows what it looks like when you hand the agent real engineering skills instead of letting it wing it. It holds 16 self-contained skill folders you drop into `$CODEX_HOME/skills`, each with its own SKILL.md that spells out when to trigger, what to do, and what good output looks like. The coverage is what makes it useful: generating App Store release notes from git history, debugging iOS apps on a simulator, packaging and notarizing macOS apps without Xcode, auditing SwiftUI and React performance, and running multi-agent diff reviews and bug hunts. The swarm skills stand out because they structure read-only review and investigation work so the agents come back with prioritized fix paths rather than a pile of commentary. If you're standardizing how your coding agents behave, this is a solid, MIT-licensed reference for well-scoped skills, and the project-skill-audit skill even helps you figure out which new skills your own setup actually needs.