This paper introduces a multi-agent architecture to process continuous queries of moving objects in mobile and wireless network. A total of seven types of collaborative agents are employed in the proposed approach, aiming to integrate the distributed geospatial databases into a computational environment and enhance the service quality and system performance. The functions of these agents include managing and scheduling query requests, discovering andtransporting results to moving objects. These agents are distributed on the proposed architecture, and they cooperate with each other to process the intensivegeospatial queries. In addition, a lightweight multi-agent system is developed on the basis of FIPA (Foundation for intelligent physical agents) AbstractArchitecture and as an extension of JADE (Java Agent Development Framework). Experiments were conducted to simulate the continuous geospatial queries and examine the applicability of proposed method in Shanghai demonstrated. The results show that the proposed architecture could realize automatic and directly transmit the query requests and results in unstable wireless environment. Finally, the performance of the proposed architecture was evaluated.