arXiv:2611.23362v1 Announce Type: new 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.
TRACE-MAS: Diagnosing and Repairing Collaboration Failures in Multi-Agent Systems
A new arXiv paper (2611.23362v1) introduces TRACE-MAS, a framework that learns normal collaborative execution patterns from benign trajectories to detect deviations in agent state changes, message transitions and task dependency evolutions in LLM-based multi-agent systems. TRACE-MAS uses source-aware attribution to identify responsible execution elements and selects minimal interventions to restore collaboration, improving root-cause localization accuracy by 60.2% across five MAS benchmarks versus strong MAS safeguards, with consistent gains in task recovery. The work targets a gap the authors identify in existing safeguards, which perform agent-level anomaly detection that misses failures originating from agent interactions.
Monitor inter-agent message transitions and task dependency evolution, not just individual agent outputs, to localize multi-agent failures; the source reports 60.2% better root-cause accuracy across five benchmarks.
more in playbook: Designing agent systemsRun your AI side-project on zahid.host
EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.