Skills repo splits into commands, autonomous workflows

Separating user commands from model-invocable skills marks a maturation in AI engineering, moving from manual prompting to structured agentic autonomy.

• –User-invocable commands like `/improve-codebase-architecture` provide predictable high-level entry points for human developers.
• –Model-invocable background skills allow agents to autonomously trigger complex reasoning workflows like `/deep-modules` based on context.
• –This architecture prevents "instruction bloat" by loading only the contextually relevant logic required for the immediate task.
• –It reinforces the "Deep Modules" philosophy by forcing agents to follow disciplined, multi-step engineering processes rather than "vibe coding."
• –With over 100,000 stars, the repository is effectively defining the standard protocol for professional AI agent interaction.