Targeted covalent inhibitors (TCIs) form covalent bonds with targets following initial non-covalent binding. The advantages of TCIs have driven a resurgence in rational TCI design over the past decade, resulting in the approval of several blockbuster covalent drugs. To support TCI discovery, various computational methods have been developed. However, accurately predicting TCI reactivity remains challenging due to interference between non-covalent scaffolds and reactive warheads, leading to inefficiencies in computational screening and high experimental costs. In this study, we enhanced the SCARdock protocol, a validated computational screening tool developed by our lab, by incorporating quantum chemistry-based warhead reactivity calculations. By integrating these calculations with non-covalent docking scores, docking ranks, and bonding-atom distances, non-covalent and covalent inhibitors of S-adenosylmethionine decarboxylase (AdoMetDC) were correctly classified. Using the optimized SCARdock, we successfully identified twelve new AdoMetDC covalent inhibitors from 17 compounds, achieving a 70.6% hit rate. From these novel inhibitors, we analyzed the contributions of non-covalent interactions and covalent bonding, enabling a structure-activity relationship (SAR) analysis for AdoMetDC covalent inhibitors, which was previously unexplored with substrate-based inhibitors. Overall, this work presents an efficient computational protocol for TCI discovery and offers new insights into AdoMetDC inhibitor design. We anticipate that this approach will stimulate TCI development by improving computational screening efficiency and reducing experimental costs.