Energy model is one of the most significant tool to investigate building energy saving, which has been widely used by government in energy policy evaluation. However, there are still many limitations in current research technology and methodology. Especially the interaction between energy consumption and occupant behavior in daily activities should be further discussed. Lots of the research method and conception have not been well expanded by modeling and quantizing. Agent Based Domestic Community Energy Model is proposed to develop the simulation framework of energy consumption in the stock of domestic buildings. The occupants are taken as the agent to simulate the occupant behavior factors that influence building energy-saving potential under various policy scenarios. The proposed model can associate the standard ISO-13790 for calculation the cooling and heating load, but also incorporate the interaction of agent behavior and energy use, which realize the organic combination of environment and behavior simulation. The results demonstrate that the agent behavior is one of the significant factors that influence building energy use intensity, and strengthening the guidance of occupant behavior is a scientific and effective way to achieve the goal of building energy efficiency.