Welcome to AgentZero. The goal is to create a minimal, opinionated LLM agent that is proficient in task execution and tool calling.
Here are the functionalities I eventually want to add -
- Long running processes in which working memory is summarized as execution proceeds.
- Goal expansion and breakdown. Ability to break down goals into steps before starting execution.
- Asynchronous check on execution status. While running a longer task, ability to get a status check without interrupting execution.
This Agent does not have to be useful or modular or pluggable into any other applications. If it is useful, it is by accident because the goal here is to see how smart can a few lines of python code get.