Huge spatio-temporal data (e.g., traffic flow, human mobility, and geographical data) are generated in modern cities. Accurately forecasting over spatio-temporal data enables many essential applications in intelligent cities, such as traffic management, public safety, and economy. To improve the ease of use, we propose an Enhanced Automated machine learning library for Spatio-Temporal forecasting, entitled EAST. In EAST, we mainly reconstruct three types of automated machine learning methods, namely, AutoSTPoint, AutoSTGrid, and AutoSTGraph for spatio-temporal point (STPoint) forecasting, spatio-temporal grid (STGrid) forecasting, and spatiotemporal graph(STGraph) forecasting, respectively. We adapt structureaware algorithms using neural architecture search methods. The search space is elaborately designed according to the structures and characteristics of spatio-temporal data. We implement EAST with the popular deep learning frameworks. Finally, we conduct experiments on real-world datasets to demonstrate the effectiveness and superiority of our EAST.