You're right, but here's the problem: things that are robust to evolutionary change tend to kill everything (e.g. chlorine). There's a tension between finding targets that are specific to an organism, but well conserved, so that the organism can't simply evolve away from the specific threat.
This is really the alpha and the omega of antibiotics and antiviral research: finding the specific lethal target that can't easily be changed. It's damned hard, and it's why most of our antibiotics have been generated by nature, not by rational design.
I used to do antibiotic drug discovery for a living. You hit the nail on the head.
I'll go so far as to say that we've found all of the important bacterial targets (e.g., ribosome, peptidoglycan, DNA gyrase & a few others).
That's probably an overstatement....but I (and many others) spent a large amount of time/$/effort looking for antibacterial targets that might have been overlooked. In the end, my conclusion was that we've already found the softest, most vulnerable targets within bacteria and we've known about them for a long time.
What to do? I think 2 things. Find new ways to hit the existing validated targets (people are doing this, but it ain't easy). The only other "idea" that I've had (and I'm sure others have had it, too), is that maybe we can find multiple "weak" targets that, when hit in combination, are just as good at inactivating a single "thing" like the bacterial ribosome. But that's probably even more difficult than finding new ways to hit well-validated targets.
It's easy to kill bacteria, viruses, cancer, you name it. We have lots of tools that can kill them, from guns to flame throwers to acid. The trick is killing them without also killing the patient, as you say.
This is really the alpha and the omega of antibiotics and antiviral research: finding the specific lethal target that can't easily be changed. It's damned hard, and it's why most of our antibiotics have been generated by nature, not by rational design.