arXiv cs.AI· Xinyuan Song, Zekun Cai·· 12 小时前AI 评分38
ReCore：面向长程智能体的规划图核心修复方法
Repair the Core, Not the Symptom: Planning-Graph Repair for Long-Horizon Agents
AI 导读
研究者提出 ReCore（Repair the Core），通过建模节点与边错误的耦合传播来定位失败规划图中的核心区域，只将该核心交给 LLM 诊断修复，而非修复最大可见错误或固定邻域。在合成 Agent Calling-Tree 图与 Who&When 归因基准上，ReCore 提升了长程稳定性与根因定位能力，每次成功修复的 token 消耗比全上下文恢复最多少 7.4 倍。
正文 · 原文
Abstract:Long-horizon language agents make interdependent decisions across multiple rounds, forming growing planning graphs in which later decisions depend on earlier states, observations, and tool outputs. Errors in these graphs can therefore propagate through subsequent decisions and amplify over long rollouts. Correcting the most visible failure may therefore fix a symptom while leaving the core that drives downstream errors unchanged. Replaying the entire planning graph provides the opposite extreme, but becomes increasingly expensive as the graph grows and forces an LLM to reason over substantial irrelevant context. We study planning-graph repair: given a failed rollout, the goal is to identify and correct the compact core responsible for continued error propagation while preserving the remainder of the graph. We first evaluate a range of practical engineering repair strategies, including single-error and TopK repair, temporal windows, neighborhood expansion, cascade-aware repair, and full-graph LLM correction. We then propose ReCore (Repair the Core), which models the coupled propagation of node and edge errors to identify the core of a failed planning graph. Rather than repairing the largest observed errors or expanding a fixed local neighborhood, ReCore selects a connected core whose correction maximally reduces residual error amplification, and exposes only this core to the LLM for diagnosis and repair. We also theoretically establish a bound on long-horizon planning regret. Extensive experiments on synthetic Agent Calling-Tree graphs and the Who\&When attribution benchmark show that ReCore improves long-horizon stabilization and root-cause localization by targeting high-amplification repair regions, matching full-context recovery with up to $7.4\times$ fewer tokens for every successful repair.
| Comments: | Under Review |
| Subjects: | Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.01767 [cs.AI] |
| (or arXiv:2607.01767v3 [cs.AI] for this version) | |
| https://doi.org/10.48550/arXiv.2607.01767 arXiv-issued DOI via DataCite |
Submission history
From: Zekun Cai [view email]
[v1]
Thu, 2 Jul 2026 06:31:45 UTC (8,011 KB)
[v2]
Sun, 5 Jul 2026 07:41:29 UTC (7,564 KB)
[v3]
Tue, 20 Oct 2026 02:42:27 UTC (4,700 KB)
来源：arXiv cs.AI · arxiv.org