We are in the midst of writing a proofreading (and much more) plugin for Discourse:
One of the modules allows us to quickly proofread a post on Discourse.
I was experimenting a bit today with prompt engineering at:
A concept I had was to have a dial from 1 → 10, that indicates how pedantic and English teacher like the proof reading would be.
Early results are quite interesting. I love that we can use prompt engineering to ensure ChatGPT does not make any egregious mistakes (like editing quotes or messing up markdown)