OP here. I built this because I kept seeing people hacking together Claude swarms on local Mac Minis, but they're fragile. For example if the process dies the swarm state dies with it.
I wanted a production grade orchestration which survives restarts. It’s designed to be cloud-agnostic so you don't need dedicated local hardware to run persistent multi-agent workflows and a benefit of this is its accessible anywhere. I've also given my own version of this access to its GCP project/services and its GitHub and its now creating its own improvements.
OP here. I built this because I kept seeing people hacking together Claude swarms on local Mac Minis, but they're fragile. For example if the process dies the swarm state dies with it.
I wanted a production grade orchestration which survives restarts. It’s designed to be cloud-agnostic so you don't need dedicated local hardware to run persistent multi-agent workflows and a benefit of this is its accessible anywhere. I've also given my own version of this access to its GCP project/services and its GitHub and its now creating its own improvements.
Happy to answer questions!
How are you keeping agents alive even when the server dies?