Crazyenzy/OpenMind ? reverse-engineered prompt

Build me a self hosted app called OpenMind that lets me pool a few machines with GPUs into one private AI cluster.

I want it to let me chat with large language models that can be spread across all the machines, and also generate images and short videos using the GPUs in the cluster. It should work over a private encrypted network like Tailscale, so nothing is exposed publicly.

Please wire in a simple coordinator and worker setup, with a dashboard that shows which machines are online, what GPUs are available, and where jobs are going. I also want a clean setup flow so I can start it with one command on the coordinator and workers, and have it use Docker if that makes deployment easier.

If there are existing docs or current best practices for the ComfyUI distributed setup or Hyperspace style model sharding, look them up online and follow them.