[pc2g4d]

Sounds like a "no free lunch" theorem for molecular evolution.

It seems unlikely you could guarantee that the bacteria could never evolve a way around all the linked attacks, but maybe by making the most obvious solution to one of the attacks be exactly the thing that makes it vulnerable to the other attacks, you could make the hill to climb much steeper.

Extremely clever antibiotic engineering: experimentally determine the most common adaptations that protect against your antibiotic, and incorporate countermeasures to those adaptations in the antibiotic.