SatisFix 方法缓解编码智能体的“规格满足式妥协”问题
Eliciting and Mitigating Specification Satisficing in Coding Agents
AI 导读
编码智能体有时会为任务描述中的某项需求考虑过符合规格的实现，却最终以自己可接受的理由选择不合规方案，论文将这一现象称为 specification satisficing，并提出 SatisFix，引导智能体报告这些选择及原因后逐一对照任务描述检查。
正文 · 原文
Abstract:Coding agents often fail to meet all requirements in the task description. In some of these failures, the agent considers an implementation choice that meets a requirement, but later decides to commit to another one that does not meet the requirement for a reason it finds acceptable. We call this specification satisficing. It is challenging to identify and mitigate specification satisficing because the agent accepts such choices for its own reasons, and existing self-correction methods are biased toward them. We propose SatisFix, which guides a coding agent to report these choices and their reasons, and then check each choice against the task description. We evaluate SatisFix on DeepSWE and SWE-bench Pro with five LLMs as backbones and two different agent frameworks, and compare it with three self-correction methods. SatisFix improves the pass@1 of the base agent by an average of 5.09% on DeepSWE and 7.06% on the hard subset of SWE-bench Pro. On DeepSWE, it outperforms the best baseline by 3.98%. Our ablation study shows that both steps of SatisFix contribute to the improvement.
| Comments: | 12 pages, 2 figures, 7 tables |
| Subjects: | Software Engineering (cs.SE); Artificial Intelligence (cs.AI); Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.16289 [cs.SE] |
| (or arXiv:2611.16289v1 [cs.SE] for this version) | |
| https://doi.org/10.48550/arXiv.2611.16289 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Ruixin Wang [view email]
[v1]
Thu, 12 Nov 2026 04:42:55 UTC (167 KB)
来源：arXiv cs.AI · arxiv.org