AI & ML
TRACE-MAS: Diagnosing and Repairing Collaboration Failures in Multi-Agent Systems
arXiv cs.MA0 views
arXiv:2611.23362v1 Announce Type: cross Abstract: LLM-based multi-agent systems (MAS) rely on not only individual agent behaviors but also the interactions that connect agents, including message exchanges and task dependencies. Existing MAS safeguards mainly perform agent-level anomaly detection, which becomes insufficient when failures originate from agent interactions and detected anomalies do not necessarily reveal their true sources. To address this limitation, we propose \textbf{TRACE-MAS}, a finer-grained framework for diagnosing and repairing MAS failures. TRACE-MAS learns normal collaborative execution patterns from benign trajectories and detects deviations in agent state changes, message transitions, and task dependency evolutions. It further identifies responsible execution elements through source-aware attribution and selects minimal interventions to restore collaboration while preserving task-critical capabilities. Experiments on five MAS benchmarks demonstrate that TRACE-MAS significantly improves root-cause localization accuracy by \textbf{60.2}\% and achieves consistent gains in task recovery over strong MAS safeguards.
Source:https://arxiv.org/abs/2611.23362PreviousKaer: Security-Aware Replanning for Multi-Agent SystemsNextARISE: Practical Self-Evolving Memory for Accurate and Cost-Efficient Enterprise Agents
Related posts
An Interview With Eric Seufert About Agentic Commerce
0Stratechery
Bring charts, queries, and context into one shareable document: Introducing Grafana Notebooks
0Grafana Blog

Four Days Early: Why AI-Found Vulnerabilities Demand Protection That’s Already On
0Imperva Blog
Two language fields, not one: what grandparent songs taught us about input design
0dev.to UX
Análise de Ações com IA: Conheça o daily_stock_analysis
0dev.to GitHub
Phân tích dữ liệu Lorient vs Rennes: Sự chuyển dịch cán cân áp đảo tại Ligue 1
0dev.to Data