The only advice worth following here is: send shit PRs back.

If you don’t understand the change, it’s too large, it contains multiple independent changes that should’ve been separate PRs, anything that doesn’t smell right - send it back.

My expectation when you review my PR is that your ass is on the line just as much as mine if something goes wrong.

PRs aren’t a checkmark exercise that validates you’re not trying to backdoor an exploit into the system. A reviewer that accepts a change is committing themselves to maintain said change going forward.

If you let your kids get a dog and the kids don’t take care of it, you will. There’s no two ways about it.