I don't have a lot of patience for Bob. That being said I have to agree with him on test coverage (that's as far as I made it through his monologue). IMHO, that is something that I 100% am okay letting the LLM tooling write and manage. I used to argue about whether or not we needed a test that verified that the value of a constant didn't change, and if 100% coverage was really that important. Now I don't care, I just let Claude write the test and keep it up-to-date.