Aegis: Automated Error Generation and Attribution for Multi-Agent Systems

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Kong, Fanqi, Zhang, Ruijie, Yin, Huaxiao, Zhang, Guibin, Zhang, Xiaofei, Chen, Ziang, Zhang, Zhaowei, Zhang, Xiaoyuan, Zhu, Song-Chun, Feng, Xue
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915960111235072
author Kong, Fanqi
Zhang, Ruijie
Yin, Huaxiao
Zhang, Guibin
Zhang, Xiaofei
Chen, Ziang
Zhang, Zhaowei
Zhang, Xiaoyuan
Zhu, Song-Chun
Feng, Xue
author_facet Kong, Fanqi
Zhang, Ruijie
Yin, Huaxiao
Zhang, Guibin
Zhang, Xiaofei
Chen, Ziang
Zhang, Zhaowei
Zhang, Xiaoyuan
Zhu, Song-Chun
Feng, Xue
contents Large language model based multi-agent systems (MAS) have unlocked significant advancements in tackling complex problems, but their increasing capability introduces a structural fragility that makes them difficult to debug. A key obstacle to improving their reliability is the severe scarcity of large-scale, diverse datasets for error attribution, as existing resources rely on costly and unscalable manual annotation. To address this bottleneck, we introduce Aegis, a novel framework for Automated error generation and attribution for multi-agent systems. Aegis constructs a large dataset of 9,533 trajectories with annotated faulty agents and error modes, covering diverse MAS architectures and task domains. This is achieved using a LLM-based manipulator that can adaptively inject context-aware errors into successful execution trajectories. Leveraging fine-grained labels and the structured arrangement of positive-negative sample pairs, Aegis supports three different learning paradigms: Supervised Fine-Tuning, Reinforcement Learning, and Contrastive Learning. We develop learning methods for each paradigm. Comprehensive experiments show that trained models consistently achieve substantial improvements in error attribution. Notably, several of our fine-tuned LLMs demonstrate performance competitive with or superior to proprietary models an order of magnitude larger, validating our automated data generation framework as a crucial resource for developing more robust and interpretable multi-agent systems. Our project website is available at https://kfq20.github.io/Aegis-Website/.
format Preprint
id arxiv_https___arxiv_org_abs_2509_14295
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Aegis: Automated Error Generation and Attribution for Multi-Agent Systems
Kong, Fanqi
Zhang, Ruijie
Yin, Huaxiao
Zhang, Guibin
Zhang, Xiaofei
Chen, Ziang
Zhang, Zhaowei
Zhang, Xiaoyuan
Zhu, Song-Chun
Feng, Xue
Robotics
Multiagent Systems
Large language model based multi-agent systems (MAS) have unlocked significant advancements in tackling complex problems, but their increasing capability introduces a structural fragility that makes them difficult to debug. A key obstacle to improving their reliability is the severe scarcity of large-scale, diverse datasets for error attribution, as existing resources rely on costly and unscalable manual annotation. To address this bottleneck, we introduce Aegis, a novel framework for Automated error generation and attribution for multi-agent systems. Aegis constructs a large dataset of 9,533 trajectories with annotated faulty agents and error modes, covering diverse MAS architectures and task domains. This is achieved using a LLM-based manipulator that can adaptively inject context-aware errors into successful execution trajectories. Leveraging fine-grained labels and the structured arrangement of positive-negative sample pairs, Aegis supports three different learning paradigms: Supervised Fine-Tuning, Reinforcement Learning, and Contrastive Learning. We develop learning methods for each paradigm. Comprehensive experiments show that trained models consistently achieve substantial improvements in error attribution. Notably, several of our fine-tuned LLMs demonstrate performance competitive with or superior to proprietary models an order of magnitude larger, validating our automated data generation framework as a crucial resource for developing more robust and interpretable multi-agent systems. Our project website is available at https://kfq20.github.io/Aegis-Website/.
title Aegis: Automated Error Generation and Attribution for Multi-Agent Systems
topic Robotics
Multiagent Systems
url https://arxiv.org/abs/2509.14295