Hi all, newbie here.
On first 2 days on trial I was amazed, wanted to suggest all the friends of mine use Cursor. It did well with merging codes (where CoPilot or ChatGPT failed because of context limits), it did not hallucinate, kept the rules, etc.
On the last three days (still first week of trial) it has a context warning (too big file). Ok, I get that, but have no idea how to bughunt without big number of lines.
But: I started a very basic task which it made successfully on first day (clean start) and it was not able to make it. Tested with competitors, they made it on first or max. second attempt perfectly. For now it misinterprets everything, adds unwanted features, cannot simply write a 300 line code.
I think Cursor switched back the model to a ‘Dumb and Dumber’; now I literally make copilot and chatgpt free to write parts (2-5 minutes), then spend 2 hours the properly add that code as a replacement of the cursor code (sometimes a day). It was able to add immediately functions on first 2 days.
Looks like business strategy: impressing newcomers, let them pay immediately by impressing them, then you face the of the business model.
Where am I wrong?
