We introduce a computational workflow for predicting the reactivity of covalent drug warheads with nucleophilic amino acid residues. Covalent inhibitors form permanent chemical bonds with the target protein through well-defined reaction mechanisms, and accurate prediction of their reactivity is crucial for drug design. This reactivity can be quantified by accurately estimating the reaction barrier heights, which require high-level quantum mechanical (QM) methods that have a prohibitive computational cost. To overcome this challenge and to fill the gap in the availability of a relevant barrier height reference data, I will show how we develop an automated quantum chemical workflow for generating a diverse dataset of reaction barrier heights comprising representative covalent warhead-nucleophile systems. We focus on diverse warhead functional groups and key reaction classes relevant to covalent drug design, including Michael addition, nucleophilic addition, nucleophilic substitution, and others involving biologically relevant nucleophiles. I will further discuss how this dataset will serve as a foundation for in-house developments of data-driven models, including Δ-machine learning approaches that aim to predict corrections to low-level reaction barrier heights with near high-level QM accuracy at significantly reduced computational cost. Overall, I will highlight how our approach provides a computational framework for predicting the reactivity of targeted covalent inhibitors as well as means to generate high-quality datasets that can ultimately facilitate high-throughput prediction for the in silico design of covalent therapeutics.