This repository contains the code for the Self-Augmenting Agentic Loop (SAAL) framework, which is designed to enhance the capabilities of language models by enabling them to self-augment their capabilities by writing their own tools to interact with external APIs.
Thanks to advancements with recent large language models it is possible to source and feed enough information into a prompt automatically for the model to be able to write code to extend its own capabilities. This framework leverages this ability to create a self-improving agentic loop that allows language models to autonomously improve their performance over time. The goal of this framework is to eventually achieve a form of self-management and improve LLMs ability to perform tasks without human intervention.