Central agent organization is very problematic

[AidenTag]

We should be following a coherent structure. Central agent is an agent just like the others, it shouldn't have a wildly different structure. We should give some thought to the flows, because currently it's just very confusing. I will work on this, but I think we need to discuss it. Too much output and way too much random code, and not enough prompting for the AI itself. This needs to be by FAR the most robustly instructed model.

• No need for the upload_input_text --> assign_tasks flow, makes no sense for our use case
• It is using model initializer all wrong. Passing system instructions as part of prompt in assign_tasks??? Makes no sense
• We should follow invoke --> ai_server --> process_response flow for this agent too, no reason this one should be different
• Create, edit, and delete functions in this class are not necessary, we just invoke the agents!!

[DhyeyMavani2003]

All great points @AidenTag! This has just been merged in as a first working draft, where we modified the central agent intuition and logic that we had in the source repo (Minerv-AI) from which this project is forked!

You can work with me & @liamjdavis to iron this out further!