Saved in:
Bibliographic Details
Main Authors: Qiao, Hezhe, Tong, Hanghang, Lim, Ee-Peng, Liu, Bing, Pang, Guansong
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.17467
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909052548677632
author Qiao, Hezhe
Tong, Hanghang
Lim, Ee-Peng
Liu, Bing
Pang, Guansong
author_facet Qiao, Hezhe
Tong, Hanghang
Lim, Ee-Peng
Liu, Bing
Pang, Guansong
contents Large language model-driven multi-agent systems (LLM-MAS) excel at complex tasks, yet unreliable agents remain a key bottleneck to system-level reliability. Automatic failure attribution is therefore critical, but existing approaches, such as direct prediction of agent-error pairs and agent-first failure attribution, rely on local logs of agents and miss global failures that only manifest over full interaction trajectories, such as cross-step inconsistencies and inter-agent coordination errors. Moreover, directly predicting failures induces a large combinatorial search space, hindering fine-grained attribution. To address these challenges, we propose VerifyMAS, a hypothesis verification framework for agent failure attribution. Instead of directly predicting faulty agents and error types, VerifyMAS formulates and verifies failure hypotheses against full trajectories. This verification-based approach decomposes attribution into trajectory-level error validation and fine-grained agent localization, providing an error-first attribution approach that captures global failure patterns while substantially reducing the search space. We further introduce a hypothesis-based data construction strategy grounded in a structured error taxonomy and fine-tune a specialized LLM verifier model for trajectory-level failure verification and agent attribution. Experiments on Aegis-Bench and Who&When show that VerifyMAS consistently improves diverse backbone models, including open-source Qwen and API-based GPT models, outperforming prior methods without sacrificing inference efficiency for long multi-agent trajectories.
format Preprint
id arxiv_https___arxiv_org_abs_2605_17467
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle VerifyMAS: Hypothesis Verification for Failure Attribution in LLM Multi-Agent Systems
Qiao, Hezhe
Tong, Hanghang
Lim, Ee-Peng
Liu, Bing
Pang, Guansong
Computation and Language
Large language model-driven multi-agent systems (LLM-MAS) excel at complex tasks, yet unreliable agents remain a key bottleneck to system-level reliability. Automatic failure attribution is therefore critical, but existing approaches, such as direct prediction of agent-error pairs and agent-first failure attribution, rely on local logs of agents and miss global failures that only manifest over full interaction trajectories, such as cross-step inconsistencies and inter-agent coordination errors. Moreover, directly predicting failures induces a large combinatorial search space, hindering fine-grained attribution. To address these challenges, we propose VerifyMAS, a hypothesis verification framework for agent failure attribution. Instead of directly predicting faulty agents and error types, VerifyMAS formulates and verifies failure hypotheses against full trajectories. This verification-based approach decomposes attribution into trajectory-level error validation and fine-grained agent localization, providing an error-first attribution approach that captures global failure patterns while substantially reducing the search space. We further introduce a hypothesis-based data construction strategy grounded in a structured error taxonomy and fine-tune a specialized LLM verifier model for trajectory-level failure verification and agent attribution. Experiments on Aegis-Bench and Who&When show that VerifyMAS consistently improves diverse backbone models, including open-source Qwen and API-based GPT models, outperforming prior methods without sacrificing inference efficiency for long multi-agent trajectories.
title VerifyMAS: Hypothesis Verification for Failure Attribution in LLM Multi-Agent Systems
topic Computation and Language
url https://arxiv.org/abs/2605.17467