Fog computing becomes more popular as it brings computational capabilities near to edge/Internet of Things devices to reduce the communication delay. Due to constantly expanding IoT Devices and their data, the monitoring of the fog computing environment encounters critical challenges related to scalability, efficiency, and performance. In this paper, a Multi-agent based monitoring system for the fog computing environment of proposed. Multiple intelligent agents are used to perform the various tasks of monitoring systems such as data collection, data aggregation, and decision making. The data collection model introduces a novel hybrid push-pull algorithm that updates data when the difference between updated values and new value should be greater than threshold value. A tree-based data aggregation model is developed to summarize the data which intern helps to reduce communication bandwidth between fog node and cloud. The complex event processing is used in decision making system to take control decision. Experimental evaluation shows that reduction in data coherence and bandwidth compared to existing models.