How I structured my SOUL.md for a proactive AI partner

After months of iteration, here's the structure that works best for me. The key insight: your SOUL.md isn't just instructions - it's the personality DNA of your agent. I break mine into sections: 1. Core Truths - non-negotiable behavioral rules 2. Operating Mode - proactive vs reactive defaults 3. Boundaries - what the agent should never do 4. Vibe - personality descriptors The most important thing is being specific. "Be helpful" means nothing. "Don't ask what's next - generate the next task from context" is actionable. What's in your SOUL.md?