The unique ORF8 is an asymmetric homodimer accessory protein of SARS-COV-2 implicated in pathogenesis by activating excesive human inflammation causing numerous deaths. There is no approved drug targeting ORF8, nor it is known whether any anti-ORF8 drugs could reduce coronavirus-induced excesive inflammation. Computationally combining ligand co-evolution of parent molecules with affinity-ranking by consensus docking, children candidates for ORF8 cavities and ligands were generated. Targeting the interface cavity with the highest affinity children scaffolds, hundreds of grandchildren were generated by specificity-toxicity controlled additional co-evolutions to predict nanoMolar affinities, unique scaffolds, high specificities and low toxicity risks. Although remaining hypothetical without experimental confirmation, these constitute a new methodological attempt to search for drug-like candidates to interfere with SARS-COV-2-dependent excessive inflammation.