How do I design tests that actually catch bugs?
Running a test is following a recipe. Designing one that catches bugs is thinking like the thing might be broken. Those are different skills, and the second is the real craft. It is normal to be able to do the first long before the second clicks. The shift is to stop testing whether the code works and start hunting for where it fails. Ask what a user might do that the maker never expected. Push the extremes, the empty input, the huge input, the wrong order, the thing done twice. A good test is not polite, it is suspicious. It goes looking for trouble on purpose. Take one feature and write down three ways it could break before you write a single test. Then test those. Designing from how things fail, instead of how they are supposed to succeed, is the whole turn. Do that a few times and it starts to feel natural.