New framework uses communication topology to diagnose multi-agent LLM failures

Researchers have developed MAScope, a novel framework designed to diagnose failures in multi-agent Large Language Model (LLM) systems. This system leverages the communication topology between agents to distinguish between different failure modes, which often present with similar symptoms. MAScope includes a Trace Structural Extractor to reconstruct communication topology from execution traces and a Topology-Conditioned Judge to classify failures based on topology, trace data, and empirical priors. Experiments demonstrated that incorporating topology context significantly improves diagnostic accuracy, raising the Macro-F1 score for gpt-mini from 0.173 to 0.350 and achieving a score of 0.346 with predicted topology, approaching the baseline of gpt-5.4. AI

IMPACT This research could lead to more robust and efficient debugging of complex multi-agent AI systems, improving their reliability.

RANK_REASON The cluster contains an academic paper detailing a new framework and experimental results for diagnosing LLM failures. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.MA (Multiagent) →

• arXiv
• gpt-5.4
• gpt-mini
• LLM
• TC-Judge

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →