I think I understand better where you were coming from now. Your complaint (about how solving the A and B vs. A-and-B issue doesn't address the infinite regress issue) seems like it's basically answered by TOD.

Except that you don't need to substitute in something specific for "anything", since it's just the set of all things - including all those possible things he lists. This might be clearer if we said "everything" or "something"?

Talking about "the set of all things" can be quite problematic in itself! But brushing abstract set theory paradoxes aside, I think once you pin down "the set of all things" or "everything" or "something" tightly enough, you have effectively substituted in something specific: you've given me enough information to discern precisely what you're asking about, and rendered the question well-posed. At that point TOD's reply kicks in.