if you want to use local agents to do rote tasks for you I believe the architecture they chose makes sense. It’s a single server running as a daemon and all integrations including the agents themselves are hooked into it via websocket.
my description isn’t the beginning and the end of it. Were that the case, you’d be correct, but the architecture and implementation are more robust than what I described I was just giving you a quick answer since the repo is public.
if you want to use local agents to do rote tasks for you I believe the architecture they chose makes sense. It’s a single server running as a daemon and all integrations including the agents themselves are hooked into it via websocket.
Unbelievably low bar for “conceptually sound”.
Idk having agents that can’t cleanly separate between data and instructions seems conceptually fucked to me.
my description isn’t the beginning and the end of it. Were that the case, you’d be correct, but the architecture and implementation are more robust than what I described I was just giving you a quick answer since the repo is public.
I doubt anyone has read the 200k lines of code