You best bet might be to treat the whole glutathione as a small molecule, rather than a peptide. Look into RosettaLigand small molecule docking. Here you’d be treating the whole of glutathione as a single ligand “residue” rather than as a three-mer peptide.
One caveat is the large number of rotatable bonds in glutathione. You would want to make sure you use a ligand rotamer library that had sufficient sampling of the possible conformational states.