AWorld: Dynamic Multi-Agent System with Stable Maneuvering for Robust GAIA Problem Solving

The rapid advancement of large language models (LLMs) has empoweredintelligent agents to leverage diverse external tools for solving complexreal-world problems. However, as agents increasingly depend on multiple tools,they encounter new challenges: extended contexts from disparate sources andnoisy or irrelevant tool outputs can undermine system reliability and accuracy.These challenges underscore the necessity for enhanced stability in agent-basedsystems. To address this, we introduce dynamic supervision and maneuveringmechanisms, constructing a robust and dynamic Multi-Agent System (MAS)architecture within the AWorld framework. In our approach, the Execution Agentinvokes the Guard Agent at critical steps to verify and correct the reasoningprocess, effectively reducing errors arising from noise and bolsteringproblem-solving robustness. Extensive experiments on the GAIA test datasetreveal that our dynamic maneuvering mechanism significantly improves both theeffectiveness and stability of solutions, outperforming single-agent system(SAS) and standard tool-augmented systems. As a result, our dynamic MAS systemachieved first place among open-source projects on the prestigious GAIAleaderboard. These findings highlight the practical value of collaborativeagent roles in developing more reliable and trustworthy intelligent systems.