With the rapid advance of e-learning, the online peer help is playing increasingly important role. This paper explores the application of MAS to an online peer help system (MAPS). In the design phase, the architecture of MAPS is proposed, which consists of a set of agents including the personal agent, the course agent, the diagnosis agent, the DF agent and the communication agent. Through cooperation and negotiation, these agents are able to provide just-in-time and personalized peer help service on demand. In particular, it makes easy to perform such tasks as registering, searching and subscribing help services in a flexible fashion in MAPS. To clearly demonstrate the interactivity and message passing among agents, an UML sequence diagram is used to illustrate the peer help process. On this basis, the agent modeling approach is also discussed. In the phase of implementation, the JADE based agent behavior model was described in detail and the implementation platform was also proposed. In the end, a demonstration is used to exhibit the feasibility and reasonability of the proposed model and approach.