3 Agents debug - terrytaylorbonn/auxdrone GitHub Wiki
see: #499_BBB_AI_API_deployments_devplan_
A1-3 Agent local (any code that calls LLM api) running locally
The browser is good for chatting. but usually you need workflows, loops, etc. So python script called the agent (runing locally at first, maybe deployed remotely later).
UI connects to agent.
Need a way to record the messages being sent between the agent and the model, in order to fix any errors in the agent code.
A1-5 Agent deploy (calls remote model; RISKS: tokens deployed)
Render: Use IP to control access. Deploy agent. RISKS: tokens deployed. Calls remote model.
