Computer-aided docking-based inverse high-throughput virtual screening (inv HTVS) was applied for a quick evaluation of compounds as new affine ligands for SARS-CoV-2 proteins. We performed Autodock Vina-based inv HTVS using more than 450 structures of the virus proteins from Protein Data Bank (PDB) and totally ~ 2000 structures of compound with electrophilic motifs (mainly, α,β-unsaturated carbonyls) from Pubchem (CID codes) or Biogem databases. The inv HTVS calculations were semi-automatically organized, run and analyzed using our helper program tool composed of Python scripts and Excel files with macroses. Criteria for binding energy (Ebind, kcal/mol) values (no more than 6.9), distance between thiol groups of cysteine residues and electrophilic groups of the compounds within 0.4 nm (distance criterium) and total number of structures meeting the criteria were used. New in silico interactions were found for Mpro in the cases of such phytochemicals as peperomine E, 15-oxosteviol, zambesiacolactone B, isodehydroeupatundin, 2deoxycucurbitacin and jatrophone. Analogously, potential covalent ligands for NSP12 were found to include ixerin D, xindongnin C, rabdosin B, epinodosin as well as such ligands for NSP16 were found to include anhydroverlotolin-like compound CID325147, mexicanin E, insulicolide B and geoditin A. These and others hits are discussed with respect to their availability and biological properties. Thus, a set of natural compounds were pointed out as potential new covalent inhibitors of SARS-CoV-2 proteins Mpro, Nsp12 and Nsp16. This information could be used for further evaluations as prophylaxis tools or drugs against COVID-19.