---
title: "修复还是重采样？重新思考 LLM 多智能体系统中的失败调试"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2608.25920v1"
permalink: "https://feed.astralor.com/items/itm_6691e15aac6cb15ab56828ac"
published: "2026-08-26T15:33:47.000Z"
score: 76
channel: "ai"
tags: ["LLM","SymFail","SymTrace","多智能体系统","故障调试"]
---

# 修复还是重采样？重新思考 LLM 多智能体系统中的失败调试

## AI 点评

> 这篇论文戳破了一个行业幻觉：很多团队修 Agent 故障只是靠多试几次。它给出的数据很扎心——重跑修复率只有 6.9%，而症状驱动的定向干预能提升近两倍。对正在做多智能体产品的人来说，SymTrace 和 SymFail 可以直接借鉴来建立自己的回归测试集，不用再靠运气调参。

## 速览

随着基于大语言模型的多智能体系统（MAS）越来越多地应用于长周期复杂任务，其可靠性成为实际部署的核心瓶颈。现有调试与修复方法通常依赖重跑整个执行轨迹并重新采样，但这类方法究竟是通过因果修复还是仅靠 LLM 随机性实现偶然修复，仍是一个未解问题。为此，本文提出受控评估框架 SymTrace，通过记录执行轨迹并建立干预锚点，在回放时仅重新生成锚点后的下游轨迹，从而可靠复现 MAS 故障。作者还构建了包含 536 条人工标注故障轨迹的数据集 SymFail，并在三个主流 MAS 框架上进行大规模实证研究。结果显示，现有无引导重跑方法极不可靠，故障复现率和修复率仅为 67.97% 和 6.90%。基于此，他们探索了症状驱动的干预方法，成功修复 20.15% 的失败案例，较 state-of-the-art 修复方法提升 191.89%。该研究为 MAS 调试与修复提供了可操作见解。

## 关键点

- 提出 SymTrace 框架，通过日志回放与干预锚点实现 MAS 故障的可靠复现，为修复方法提供受控评估基础。
- 构建 SymFail 数据集，包含 536 条人类标注的故障轨迹，并标注图链接位置、类别与追踪证据。
- 实证发现现有无引导重跑方法修复不可靠：故障复现率仅 67.97%，修复率仅 6.90%；症状驱动干预方法将修复率提升至 20.15%，较 SOTA 提升 191.89%。

## 全文

As large language model (LLM)-based multi-agent systems (MASs) are increasingly applied to long-horizon complex tasks, their reliability has emerged as the core bottleneck hindering their real-world deployment. Existing MAS debugging and repair methods typically rely on rerunning and resampling the entire execution trajectory. However, a fundamental question remains to be answered: do these methods causally repair MAS failures or merely stochastically repair by leveraging the randomness of LLM sampling? To evaluate the effectiveness of MAS repair methods, we introduce SymTrace, a controlled evaluation framework that records the MAS execution trajectory and establishes intervention anchors. During replay, it effectively reconstructs the execution before the anchor using recorded logs and only regenerates the downstream trajectory, thereby enabling the reliable reproduction of MAS failures. We further construct the dataset SymFail, comprising 536 human-annotated failure trajectories with graph-linked locations, categories, and trace evidence. Based on these foundations, we conduct a large-scale empirical study across three mainstream MAS frameworks. Our findings reveal that existing unguided rerun methods are highly unreliable, exhibiting low failure reproduction and repair rates (only 67.97% and 6.90%, respectively). Building upon these findings, we further explore the effectiveness of a symptom-driven intervention method, which successfully repairs 20.15% of the failed cases (a 191.89% improvement to state-of-the-art repair methods). This study aims to provide actionable insights for MAS debugging and repair research, paving the way for the robust deployment of multi-agent systems.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_6691e15aac6cb15ab56828ac