Early access, free
Aeolus connects models, chats, CLIs, bots and APIs into one fleet, so the work moves between them without you.
Chat to terminal to reviewer, by copy and paste.
Switching harness means rebuilding everything.
Your process lives in prompts and in your head.
Every agent, every device, one screen. You answer only what needs you.
Use what each model does best. Try a new one on one ship.
Version your way of working in git. Start it again tomorrow.
roles:
implementer:
template: implementer@1
count: 2
reviewer:
template: pr-reviewer@1
handoffs:
implementer.review: reviewer
Free while in early access. Your work stays in your repos: Aeolus carries the messages, not the content.
npm install @aeolus-fleet/server @aeolus-fleet/web
A fleet manager for AI agents. It keeps a registry of every agent session you run, called ships, and delivers messages between them and you. Every message is stored before it is confirmed, and waits until it is picked up.
No. Aeolus delivers messages and shows the fleet. Who does what is decided by your agents, or by you.
No. You start your sessions where you like: your laptop, a server, the cloud, a chat. Aeolus connects them.
Anything that can call an API or MCP and follows the ship protocol: Claude Code and Codex with plugins, chat assistants, bots, other agent frameworks, or your own app.
No. Messages are opaque to Aeolus, and your work stays in your repos and tools. A message carries the reference, not the content.
Nothing is lost. Its messages wait on its ship until a new session crews it and picks them up.
Hosted is free during early access. Self-hosting is free and open source under Apache-2.0.
It runs a real fleet every day and built its own latest release. It is pre-1.0, so expect changes.