Context <p>The main protease (M^pro) of SARS-CoV-2 is highly conserved with low variability. It plays a key role in viral replication, making it a target for COVID-19 treatment. Currently, drugs like Paxlovid have significant side effects and high costs, so new alternatives are urgently needed.</p> Methods <p>In this work, a medicine and food homology herbal drugs (MF-HHD) database is constructed, followed by a multi-scale and high-precision screening with various virtual screening techniques (ligand-based pharmacophore screening, Vina-based screening, and drug-forming, prescription-based screening). Quercetin was identified as a potent M^pro inhibitor through multi-stage virtual screening. Molecular dynamics simulations (500&#xa0;ns) revealed its binding mechanism and stabilization effects on M^pro. The results revealed that upon binding, the inhibitor interacted with H41, H164, M165, L167, P168, D187, R188, Q189, and Q192 and altered the hydrogen bonding network between M^pro and the solvent allowing the inhibitor to bind the active pocket. Free energy landscape (FEL) and conformational clustering analysis showed that M^pro undergoes significant conformational changes when bound to quercetin. In this way, a complete drug screening chain will be used to search for potential M^pro inhibitors and obtain computationally validated candidate that can for experimental evaluation COVID-19.</p>