AutoHarness: improving LLM agents by automatically synthesizing a code harness

TL;DR

AutoHarness automatically creates code harnesses for LLM agents, preventing illegal actions and enabling smaller models to outperform larger ones by synthesizing policies in code, leading to cost-effective improvements.

Contribution

The paper introduces a method for automatically synthesizing code harnesses for LLM agents, enhancing safety and performance without manual intervention.

Findings

Automatically prevents illegal moves in 145 games

Smaller Gemini-2.5-Flash outperforms larger models

Synthesizing policies in code improves reward scores

Abstract

Despite significant strides in language models in the last few years, when used as agents, such models often try to perform actions that are not just suboptimal for a given state, but are strictly prohibited by the external environment. For example, in the recent Kaggle GameArena chess competition, 78% of Gemini-2.5-Flash losses were attributed to illegal moves. Often people manually write "harnesses" around LLMs to prevent such failures. In this paper, we demonstrate that Gemini-2.5-Flash can automatically synthesize such a code harness, using a small number of rounds of iterative code refinement given feedback from the (game) environment. The resulting harness prevents all illegal moves in 145 different TextArena games (both 1-player and 2-player), enabling the smaller Gemini-2.5-Flash model to outperform larger models, such as Gemini-2.5-Pro. Pushing our technique to the limit, we can get Gemini-2.5-Flash to generate the entire policy in code, thus eliminating the need to use the LLM at decision making time. The resulting code-policy receives a higher average reward than Gemini-2.5-Pro and GPT-5.2-High on 16 TextArena 1-player games. Our results show that using a smaller model to synthesize a custom code harness (or entire policy) can outperform a much larger model, while also being more cost effective.