arXiv cs.AI· Hao Chen, Yukun Yan, Ziyu Han, Ye He, Chuanzhe Guo, Shangzhan Li, Qi Shi, Qingfu Zhu, Maosong Sun, Wanxiang Che·· 13 小时前AI 评分34
RuHarness：让评分标准与智能体框架协同进化，实现持续自我改进
When the Critic Falls Behind: Co-Evolving Rubrics and Harnesses for Continual Agent Self-Improvement
AI 导读
RuHarness 是一个评分标准与智能体框架协同进化的框架，用于智能体的持续自我改进。它从执行轨迹中持续发现并重新评估行为标准，优先处理最相关的未解决弱点，并据此指导框架的针对性更新，同时保持任务目标不变。在多种智能体任务和基座模型上的实验显示，RuHarness 优于强框架进化基线，在未见过的留出任务上尤其明显。
正文 · 原文
Abstract:Autonomous agents increasingly depend on harnesses that govern interaction, memory, tools, and execution around large language models (LLMs). While recent work has begun to evolve these harnesses from execution experience, the supervision used to guide evolution often remains relatively static, making it difficult to track changing behavioral weaknesses across rounds and potentially limiting generalization to unseen tasks. To address this issue, we propose RuHarness, a rubric--harness co-evolution framework for continual agent self-improvement. RuHarness continually discovers and reassesses behavioral criteria from execution trajectories, prioritizes the most relevant unresolved weaknesses, and uses them to guide targeted harness updates while keeping the task objective fixed. Extensive experiments across diverse agent tasks and backbone models show that RuHarness outperforms strong harness-evolution baselines, particularly on unseen held-out tasks. Further analyses show that adaptive rubric prioritization tracks shifting weaknesses and supports more sustained improvement and stable generalization across evolution rounds.
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.36826 [cs.AI] |
| (or arXiv:2610.36826v1 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2610.36826 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Hao Chen [view email]
[v1]
Tue, 27 Oct 2026 08:24:44 UTC (7,265 KB)
来源：arXiv cs.AI · arxiv.org