I strongly recommend building affirmative detection. A script which checks everything is OK and either tells everybody "Yeah, everything is OK" or "Here are the problems" means when, inevitably, that script doesn't fire, you don't get the false impression everything is OK.
All "silent success" detection systems will also silently fail and so they're worse than useless in my experience.
All "silent success" detection systems will also silently fail and so they're worse than useless in my experience.