Autonomous Repair for Multi-Agent Systems via Monte-Carlo Tree Search

Researchers propose MARS, a Monte Carlo Tree Search-based framework for automatically repairing multi-agent systems, and introduce StateMAS, a benchmark with 1,310 replayable failure trajectories. On StateMAS, MARS outperforms state-of-the-art methods by 3.0% to 12.1% absolute improvement across settings while maintaining comparable token consumption.

arXiv:2607.29055v1 Announce Type: new Abstract: Multi-agent systems MAS are increasingly deployed to solve complex tasks. In case of incorrect or unsatisfactory outputs, users have to manually locate agent mistakes by inspecting agent trajectories i.e., {\em failure attribution} and provide feedback to refine the outputs i.e., {\em repair} . Despite some recent work in MAS failure attribution, automated mechanisms to recover from such mistakes remain largely unexplored. To bridge this gap, we propose MARS, a search-based framework that formulates MAS repair as a Monte Carlo Tree Search MCTS process and navigates the vast space of potential repairs via diagnosis-guided expansion with taxonomy-augmented evaluation. Unlike standard MCTS, which evaluates a complete simulation via full rollout, MARS evaluates the agent trajectory using partial rollout to reduce token consumption. Furthermore, we introduce StateMAS, a large-scale MAS repair benchmark with 1,310 replayable multi-agent failure trajectories spanning four types of agent architectures and four LLM backbones. Experiments on StateMAS demonstrate that MARS consistently outperforms state-of-the-art methods, achieving an absolute improvement from 3.0\% to 12.1\% across all settings, while maintaining a comparable token consumption cost. The ablation study further confirms that taxonomy-augmented evaluation and diagnosis-guided expansion are critical to achieving these performance gains.