Hyperagent

Purpose

HyperAgents is an experimental agent framework for improving task agents through evaluation and code evolution. The idea is to let one agent inspect past results, modify another agent, and measure whether the new version performs better. This project release opensource at GitHub.

Features

1. Built a LangChain and LangGraph powered agent loop.
2. Added task domains examples such as calculator, scoring, bash, paper review, and git evolution.
3. Stored evaluation results and code diffs for each generation.
4. Created examples and documentation for repeatable agent experiments.

I designed and built this project independently as the sole developer.