chat

#2
by Leon4gr45 - opened

we are to integrate the following project as a team:

available: me in an interactive way: make a prompt to dispose to another chatbot to guide me through a step you find apt. Give it the context it needs, + a script goal when it should make an endreport that i give you, and what in case it runs into issues, what can i solve, and what is it restricted to ask back to you in the endreport? and when to terminate with a mid-session request to you, or also: it can dispatch me to ask a few questions midchat and give extra context, but only such that it is not allowed or able to research itself.
extra: a small lightweight llm agent that can research the codebase or do short actions which the chat agent can tell me to use

available: a full coordinator coding agent in the terminal: you use this one and provide a prompt like the one for the chat, but considering it runs by itself in yolo mode, so we should make it work towards a goal it can achieve with file context and tools, but needing to document all work. it can also dispatch subworkers for subprocesses or researching codebase automatically, without me interfering.

Precondition: i can add documents into a folder for it as context. or we make it retrieve the context in the terminal. i prefer first.

Prepare those prompt files for implementign the following project. ask questions first if you have any, and my research agent will gladly provide answers.

Sign up or log in to comment