Efficient Failure Management for Multi-Agent Systems with Reasoning Trace Representation

2026-03-22

AI Summary

This work addresses the limitation of existing large language model–based multi-agent systems in fault management, which often fail to leverage historical fault patterns, thereby constraining diagnostic accuracy and response efficiency. To overcome this, we propose EAGER, a novel framework that, for the first time, integrates historical fault knowledge into multi-agent fault management. EAGER employs unsupervised reasoning scope contrastive learning to construct a unified representation of reasoning trajectories, effectively modeling both intra-agent reasoning and inter-agent collaboration. Furthermore, it introduces a knowledge-guided reflexive mitigation mechanism to enhance system resilience. Experimental evaluation on three open-source multi-agent systems demonstrates that EAGER significantly improves the timeliness and accuracy of fault detection and mitigation, establishing a new paradigm for building highly reliable multi-agent systems.

Technology Category

• Multiagent Systems: Multiagent Learning
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Knowledge Representation and Reasoning: Reasoning with Beliefs

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Large Language Models (LLM)-based Multi-Agent Systems (MASs) have emerged as a new paradigm in software system design, increasingly demonstrating strong reasoning and collaboration capabilities. As these systems become more complex and autonomous, effective failure management is essential to ensure reliability and availability. However, existing approaches often rely on per-trace reasoning, which leads to low efficiency, and neglect historical failure patterns, limiting diagnostic accuracy. In this paper, we conduct a preliminary empirical study to demonstrate the necessity, potential, and challenges of leveraging historical failure patterns to enhance failure management in MASs. Building on this insight, we propose \textbf{EAGER}, an efficient failure management framework for multi-agent systems based on reasoning trace representation. EAGER employs unsupervised reasoning-scoped contrastive learning to encode both intra-agent reasoning and inter-agent coordination, enabling real-time step-wise failure detection, diagnosis, and reflexive mitigation guided by historical failure knowledge. Preliminary evaluations on three open-source MASs demonstrate the effectiveness of EAGER and highlight promising directions for future research in reliable multi-agent system operations.

Problem

• Failure Management
• Multi-Agent Systems
• Reasoning Trace
• Historical Failure Patterns
• Large Language Models

Innovation

• reasoning trace representation
• unsupervised contrastive learning
• failure management
• multi-agent systems
• historical failure patterns