Sure, but that UNIX philosophy is what got us "grep -r" as the way to search files across an entire directory, which would then compose with stuff like xargs and parallel to be able to do things concurrently. I'd argue that ripgrep shows that that bundling together stuff sometimes does end up with a user experience that people prefer. The nuance lies in figuring out where the balance between "not enough" and "too much" lies, and so far I've yet to see a pithy statement like the UNIX philosophy encapsulate it well.

Alternately, maybe people's idea of what "one thing" is ends up being more subjective than it sounds (or at least depends on context). "Searching through my code" at least sounds like a reasonable idea of "one thing", and it's not crazy that someone might consider "don't search though the stuff that isn't my code, like my npm dependencies or my Rust build artifacts" would be part of "doing it well". Having to specify it every time would be annoying, so you might want to put it in a config file, but then if then if it ends up being identical to your gitignore, having to manually symlink it or copy it each time you modify it is annoying, so it's also not crazy to just use the gitignore by default with a way to opt out of it. Now we're just back where we started; custom .ignore files, fallback to .gitignore, and a flag for when you want to skip that.