Meta-Prompt
This is a minimal re-implementation of Meta-Prompt, by Noah Goodman, for building self-improving agents. It is written in LangChain.
Meta-Prompt is a simple self-improving language agent that reflects on interactions with a user and modifies its own instructions based on its reflections. The only thing that is fixed about the agent is the meta-prompt, which is an instruction for how to improve its own instructions.
For a description of Meta-Prompt, see Noah's blog-post.
Setup
This repo is written with LangChain. See installation instructions here.
Run
Launch the Jupyter notebook main.ipynb to interact with the agent.