I don’t think what you’re proposing here will work. Only if the Kr99* excited state is a known and defined metastable state (like Tc99m), the G4 will create it as a particle. Otherwise, the excited daughter nucleus will automatically be passed to the de-excitation code in the same step as radioactive decay.
What you’ll see in your stepping action (or SD) is the ground state Kr85, the electron and neutrino from the beta decay, and gammas from the de-excitation process. So you could include in your SteppingAction a “quick” analysis of the list of secondaries – if you see one or more gammas, then it was the excited-state channel.
Another option, if you want to simulate only the excited-state channel, would be to edit the $G4RADIOACTIVEDECAY/z36.a85 file (as suggested by Vladimir), removing the five lines for the ground-state channel.