toy project. it's used for learning AutoGPT, LangChain and other things related to LLMs.
My goal is to build a customized AI assistant that can help me understand codebase, organize informations such as article or books, and increase the efficiency of my daily work.
I want to use the architecture from AutoGPT.
Basically, we have Agent, Planning, Resources, Abilities and Memory. I've removed Plugins and Workspace since i don't know how to utilize them right now.
TODO(sheep): more architecture details
python3 -m venv .venv
source .venv/bin/activate
cd arcane_intelligence
poetry install
python main.py
- Implement LLM provider, chat with LLM first
- [] Test LLAMA Index
- [] Implement llama index based on MoonshotAI
- [] Test LangChain
- [] Implement Planning
- [] Implement Abilities
- [] Combine above components and implement simple agent
- [] Introduce memory subsystem