Hey, thanks for the request.
This is expected behavior. All LLMs have a knowledge cutoff date, which means they don’t have information after that point. Because of that, they may give inaccurate answers about Cursor settings and features.
For Cursor requests, I recommend:
- Use Auto mode, it automatically picks the best model for the task
- Use chat web search to get up-to-date info
- Check the official docs: Cursor Docs
- Check the changelog: Changelog · Cursor
You can read more here: AI provides inaccurate information about Cursor itself - destroys user trust