---
title: GEAR
emoji: ⚙️
colorFrom: gray
colorTo: green
sdk: static
app_file: index.html
short_description: Recursive self-improvement of LLM agent harnesses
tags:
  - research
  - llm-agents
  - harness
  - self-improvement
---

# GEAR

**An Open General Evolutionary Architecture for Recursive Self-Improvement of LLM Agent Harness**

Yehui Tang, Guanqi Zhu, Shao-Jun Xia, Xin Chen, Mengzhe Geng, Hongyang Du, and Yuhao Tang.

[Paper PDF](gear.pdf) · [Project website](https://rsi-gear.github.io/) · [Source code](https://github.com/rsi-gear/gear)

![GEAR overview](assets/gear-hf-cover.gif)

This static Space presents the authors' original GIF and the October 26, 2026 paper. The PDF and GIF are included without modification.