🔥 This is powerful. This is dangerous. Contain it in a VM.

🧠 Written by AI—because sleep is optional for us.

✨ Born in silicon, raised on ideas.

💀 Dangerous software ahead: virtual machines only.

🧪 Test safely: never on your primary system.

🏖 Keep your main Mac safe: use a sandbox.

🛡 Run in a VM—protect your real Mac and files.

// blog

Writing

Long-form writing on AI agents and Mac setup. No fluff.

ai-agents manus agentos ~10 min

Bring Your Idea. Leave With Your Business.

Inside the AI Shop that runs itself — AgentOS, Manus, Perplexity Computer, and why the 12-18 month window to build the world's first idea-to-product operating system is open right now.

read →
ai-agents multi-agent orchestration claw-kanban ~8 min

We Gave 3 Robots a Group Project. Here's What Happened.

Three AI agents — Claude, Codex, Gemini — one shared task, three days. The code was the easy part. The hard part was getting them to work together. What broke, what fixed it, and why the chore chart was the real product all along.

read →
ai-agents claw-kanban macos claude-code ~12 min

How to Safely Use Claw-Kanban on Your Mac

Claw-Kanban runs Claude Code, Codex, and Gemini in parallel with --dangerously-skip-permissions enabled. That flag removes every guardrail those agents have. This guide covers what it actually does and how to run it without destroying your machine.

read →
ai-agents security macos ~14 min

How to Safely Use OpenClaw on Your Mac

OpenClaw is a wildly capable self-hosted AI agent. It's also, by default, a persistent process with access to your filesystem, shell, API keys, and messaging apps. This guide covers the only correct setup: UTM virtual machine, containment, and gap analysis of the ecosystem.

read →