Communication topology improves diagnosis of multi-agent system failures
Know the Shape, Find the Fault: Topology-Conditioned Diagnosis of Multi-Agent LLM Failures
Multiagent SystemsArtificial Intelligence
Summary
When multiple AI agents work together, they share information to complete tasks, but sometimes their coordination fails. Different problems can look similar, so the authors studied how the pattern of communication between agents helps identify what exactly went wrong. They built a system called MAScope that figures out how agents interact from logs and uses this to better diagnose failures. Their tests show that knowing the communication structure makes diagnosis more accurate and efficient.
What this means in practice
- •For ai system operators: Diagnose coordination failures in multi-agent AI setups by recovering communication patterns from logs to pinpoint error causes more accurately and cheaply.
- •For software reliability teams: Improve fault detection workflows in distributed applications by integrating topology-based context to discriminate failure modes with less computational cost.
Authors
Xinwen Liu, Zhuocheng Pan, Isabella Zhu, Jawei Zhang, Xudong Liu, Tianyu Wo
Abstract
Multi-agent LLM systems coordinate task execution through exchanges of information among agents. When coordination breaks down, similar symptoms in execution traces can reflect different problems in how information is passed, used, or verified. Communication topology captures how agents exchange information and provides structural cues for distinguishing coordination failure modes. Using these cues for diagnosis requires establishing how topology relates to failure patterns and recovering the relevant structure from execution traces that lack explicit topology labels. We analyze the relationship between communication topology and failure patterns and introduce MAScope, a two-stage framework for topology-conditioned diagnosis. Its Trace Structural Extractor TSE recovers communication topology from heterogeneous execution traces by grounding an interaction graph in message evidence. The Topology-Conditioned Judge TC-Judge then classifies failures using the trace, predicted topology, an empirical failure prior estimated from separate labeled traces, and a short description of topology-specific failure patterns. Under a fixed orchestration structure, the recovered topology can be reused across executions. Experimental results show a statistically significant association between communication topology and failure type, with $χ^2 = 409.9$ and $p = 1.2 \times 10^{-70}$. On the \num{851} MAST-clean traces, ground-truth topology context raises gpt-mini's Macro-F1 from $0.173$ to $0.350$. With predicted topology, the pipeline achieves $0.346$, approaching the trace-only gpt-5.4 baseline of $0.372$. For \num{1000} traces under a fixed orchestration structure, the projected pipeline cost, including one topology extraction, is approximately $6\%$ of repeated gpt-5.4 diagnosis cost. These results show that topology-conditioned context improves failure diagnosis and supports lower-cost deployment.