This paper explores the challenges and open problems in multi-agent systems, emphasizing the leveraging of diverse capabilities and roles of individual agents to tackle complex tasks through collaboration. Key challenges include optimizing task allocation, fostering robust reasoning through iterative debates, managing complex context information, and enhancing memory management. The paper also discusses the potential application of multi-agent systems in blockchain systems, highlighting their ability to serve as tools and integrate with blockchain nodes to enhance security and efficiency. The structure of multi-agent systems, including equi-level, hierarchical, nested, and dynamic structures, is analyzed, along with the unique challenges each structure poses. The paper further delves into planning and memory management, discussing global and local planning, task decomposition, and the classification and management of various types of memory in multi-agent systems. Finally, the potential applications of multi-agent systems in blockchain, such as smart contract analysis, consensus mechanism enhancement, and fraud detection, are explored, showcasing the promising future of these systems in distributed computing environments.This paper explores the challenges and open problems in multi-agent systems, emphasizing the leveraging of diverse capabilities and roles of individual agents to tackle complex tasks through collaboration. Key challenges include optimizing task allocation, fostering robust reasoning through iterative debates, managing complex context information, and enhancing memory management. The paper also discusses the potential application of multi-agent systems in blockchain systems, highlighting their ability to serve as tools and integrate with blockchain nodes to enhance security and efficiency. The structure of multi-agent systems, including equi-level, hierarchical, nested, and dynamic structures, is analyzed, along with the unique challenges each structure poses. The paper further delves into planning and memory management, discussing global and local planning, task decomposition, and the classification and management of various types of memory in multi-agent systems. Finally, the potential applications of multi-agent systems in blockchain, such as smart contract analysis, consensus mechanism enhancement, and fraud detection, are explored, showcasing the promising future of these systems in distributed computing environments.