I use Cursor Pro for work. I’ve noticed many bugs that cursor is unable to fix due to context length. If I paste an entire file into the same LLM, it fixes it fine.
I’d like to point out a paradox to the devs. They think they are saving resources by truncating the context, but in fact this increases the resources because I’m making 10+ requests on a bug or feature (and often giving up) when a single call with more context would have resolved the issue.
In short, longer context saves resources in the long run.