# DemoEvolve: Overcoming Sparse Feedback in Agentic Harness Evolution with Demonstrations

Source: [DemoEvolve: Overcoming Sparse Feedback in Agentic Harness Evolution with Demonstrations](<https://www.sota2.com/research/paper/6a152932dbd7150506f7dff8>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-27T05:00:05.520Z

Authors: Lirong Che, Yuzhe yang, Peiwen lin, Chuang wang, Xueqian wang, Jian su

Published: 2026-05-23T12:10:52.000Z

## Abstract

Agent harness evolution improves frozen language-model agents by modifying the executable structures around them. We study this paradigm as a form of sample-efficient fast adaptation: instead of updating model weights, an agent can acquire task-specific competence by changing its external harness, while leaving the base model's general capabilities intact. Prior work shows that self-generated rollouts can support harness search, suggesting that agents may acquire new task competence through practice. Yet in long-horizon stochastic environments, self-practice becomes fragile: rewards are sparse, outcomes are high-variance, and failures are hard to attribute to concrete harness mechanisms. We introduce DemoEvolve, a demonstration-bootstrapped approach to harness evolution. When reward-only search is too broad and noisy, competent human trajectories serve as expert reference experience for the coding proposer, guiding harness-level diagnosis and editing. Experiments on Liar's Dice show that self-rollout evolution can work when episodes are short and failures are attributable. In contrast, Balatro exposes a harder long-horizon stochastic regime, where self-rollout evolution is misled by sparse feedback and candidate-selection noise, while tutorial-like textual knowledge alone does not yield stable improvement. Under the same limited budget, DemoEvolve produces more effective and auditable harness edits and achieves better performance. Overall, demonstrations make sparse-feedback harness evolution more diagnosable, localizable, and stable.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Game Playing on Balatro In-distribution seeds](<https://www.sota2.com/research/sota/game-playing-on-balatro-in-distribution-seeds.md>): Game Playing; Balatro In-distribution seeds; Capped Mean Final Round: 24.

- [Game Playing on Balatro Out-of-distribution seeds](<https://www.sota2.com/research/sota/game-playing-on-balatro-out-of-distribution-seeds.md>): Game Playing; Balatro Out-of-distribution seeds; Capped Mean Final Round: 24.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.24539>)

[PDF](<https://arxiv.org/pdf/2605.24539.pdf>)