Molecule synthesis is one of the fundamental problems in drug discovery. Its objective is to generate a synthetic route to synthesize the desired target molecule beginning from starting materials via a sequence of reactions. Current data-driven strategies employ one-step retrosynthesis models and search algorithms to predict the synthetic route in an up-to-bottom manner. Despite their effective performance, these strategies face limitations in the molecule synthetic route generation due to a greedy selection of the next molecule without any lookahead. Furthermore, current strategies lack the capability to control the generation of synthetic routes according to factors such as material costs, yields, and reaction conditions. In this work, we propose a general framework via residual energy-based models (EBMs), that focus on the entire synthetic routes to improve their quality. By incorporating an energy-based function into our probabilistic model, we significantly improve the quality of the routes generated by various strategies in a plug-and-play fashion. Extensive experiments demonstrate that our framework can consistently boost performance across various strategies and outperforms previous state-of-the-art top-1 accuracy by a margin of 2.5\%.