Agentlas-OS

Why we included this project

Agentlas OS takes a different approach to multi-agent setups: instead of wiring every agent into one long-lived workflow, you keep a pool of specialist agents in a hub and let a temporary orchestrator assemble them per task. Requests in plain language become runnable agents or teams, you can pull in specialists from the public Agentlas Hub, and anything you build lands in an owner-scoped cloud so it survives on any machine you install the OS on. It is local-first and works with whatever model host you already use, such as Claude Code, Codex, Gemini CLI, Cursor, or Ollama. If your team has a pile of small automations that should outlive the session that created them, this turns those scattered prompts into reusable pieces you can call on later.