Most text linters rely on simple POS tagging or regex, I've been working on this tool that allows one to use the dependency graph of a sentence and then allow you to run rules over it.

It's not as fast but it allows for pretty sophisticated rule writing since you can refer to the relation between words.

I've made a demo where I did an "AI text lint" to highlight common AI kind of text.