Search papers, labs, and topics across Lattice.
This paper introduces SymTrace, a novel evaluation framework designed to assess the effectiveness of debugging methods in large language model (LLM)-based multi-agent systems (MASs). By recording execution trajectories and establishing intervention anchors, it allows for the reliable reproduction of MAS failures and evaluates existing repair methods against a dataset of 536 human-annotated failure trajectories. The study finds that traditional unguided rerun methods are unreliable, achieving only 6.90% repair rates, while a new symptom-driven intervention method significantly improves repair success to 20.15%, highlighting the need for more effective debugging strategies in MASs.
Existing debugging methods fail to reliably repair failures in multi-agent systems, with traditional approaches achieving less than 7% repair success.
As large language model (LLM)-based multi-agent systems (MASs) are increasingly applied to long-horizon complex tasks, their reliability has emerged as the core bottleneck hindering their real-world deployment. Existing MAS debugging and repair methods typically rely on rerunning and resampling the entire execution trajectory. However, a fundamental question remains to be answered: do these methods causally repair MAS failures or merely stochastically repair by leveraging the randomness of LLM sampling? To evaluate the effectiveness of MAS repair methods, we introduce SymTrace, a controlled evaluation framework that records the MAS execution trajectory and establishes intervention anchors. During replay, it effectively reconstructs the execution before the anchor using recorded logs and only regenerates the downstream trajectory, thereby enabling the reliable reproduction of MAS failures. We further construct the dataset SymFail, comprising 536 human-annotated failure trajectories with graph-linked locations, categories, and trace evidence. Based on these foundations, we conduct a large-scale empirical study across three mainstream MAS frameworks. Our findings reveal that existing unguided rerun methods are highly unreliable, exhibiting low failure reproduction and repair rates (only 67.97% and 6.90%, respectively). Building upon these findings, we further explore the effectiveness of a symptom-driven intervention method, which successfully repairs 20.15% of the failed cases (a 191.89% improvement to state-of-the-art repair methods). This study aims to provide actionable insights for MAS debugging and repair research, paving the way for the robust deployment of multi-agent systems.
