I just started to use Cursor on 3/2/25 so these are initial impressions …
It’s pretty good … overall first impression is good.
But … I REALLY dislike the fact that it generates code that fails the linting and then I have to spend time and credits to resolve the linting errors … makes me unhappy 
I would like to understand more about the config and rules so that’s my next stop but then I have to spend time and credits figuring that out …
I was able to bang out a lot of decent code in a couple of hours and get a good overall feel for the tooling and went ahead a paid for the year.
Gotta figure out the credits and how they are calculated and such but that’s low on my list at this point.
I asked it to generate automated tests and it did but then only 45% coverage for a very small application … meh, hopefully that will improve …
It makes some choices I might not have so it requires monitoring … I have told it to “update the readme.md with the name of the libraries included, what it does and where it is used in the application” … it does it and quite nicely but I would like to figure out how to make this more “automagic”
Any guidance, suggestions and experiences anyone would like to share would be appreciated …