发表机构
The Hong Kong University of Science and Technology (Guangzhou); The Hong Kong University of Science and Technology（香港科技大学（广州）; 香港科技大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
RESKILL提出结构化修复框架，通过显式失败归因、覆盖选择与重测更新，在ALFWorld和TextCraft上超越直接修复，平均成功率提升3.7个百分点。
AI 中文摘要
语言智能体越来越依赖可复用的技能，但失败后的修复通常由不透明的单次反思处理：模型生成技能补丁时，既不显式维护失败解释与候选修复之间的关系，也不考虑不成功的重测应如何影响后续编辑。我们提出RESKILL，一种在修复轮次间维护显式修复状态的结构化修复框架。给定一次失败的轨迹，该框架将失败假设与候选技能补丁关联，通过基于覆盖的归因选择局部修复，在环境中重测编辑后的技能集，并利用重测结果指导后续修复更新。语言模型提供结构化修复因素，而修复过程记录这些因素，根据候选局部技能补丁对活跃失败解释的解决程度进行比较，并将不成功的重测结果带入后续修复轮次。我们在ALFWorld和TextCraft上，以三种模型规模在固定修复预算下评估RESKILL。RESKILL在所有六个基准-模型组合中取得最强最终成功率，平均最终成功率比直接修复提高3.7个百分点，比假设条件修复提高3.3个百分点。这些结果表明，仅靠显式归因是不够的；当归因与修复选择及持续的重测条件更新相结合时，才能产生持久的改进。
英文摘要
Language agents increasingly rely on reusable skills, but post-failure repair is often handled by opaque one-shot reflection: a model generates a skill patch without explicitly maintaining how failure explanations relate to candidate repairs or how unsuccessful retests should influence later edits. We introduce RESKILL, a structured repair framework that maintains an explicit repair state across repair rounds. Given a failed rollout, the framework links failure hypotheses to candidate skill patches, selects local repairs through coverage-based attribution, retests the edited skill set in the environment, and uses retest outcomes to guide subsequent repair updates. The language model supplies structured repair factors, while the repair procedure records them, compares local skill patches by how well they address active failure explanations, and carries unsuccessful retest outcomes into later repair rounds. We evaluate RESKILL on ALFWorld and TextCraft across three model sizes under fixed repair budgets. RESKILL obtains the strongest final success in all six benchmark-model settings, improving average final success by 3.7 percentage points over direct repair and 3.3 points over hypothesis-conditioned repair. These results suggest that explicit attribution alone is insufficient; durable improvement emerges when attribution is integrated with repair selection and persistent retest-conditioned update.
CommentsAccepted to EMNLP 2026 (Main Conference)
