Cursor 5 Super Kept Breaking My Code During Development

For approximately the past three days, I have been experiencing serious problems while coding with the latest Cursor model, referred to as “5 Super.” The application is particularly problematic when modifying code in the background. I have been unable to get it to perform even a very simple macro operation correctly, despite spending hours trying. While other models can complete the same task in a single attempt, I spent approximately 16–18 hours trying to make this macro stable using Cursor’s own models, and I frequently received outputs that did the exact opposite of what I had requested.

Even when the code appears to have been fully fixed, the macro later becomes broken again without any intervention from me, and the code can somehow end up in an incorrect state on its own. This has given me the impression that the behavior may be intentional. In addition, despite the poor performance I have experienced with Cursor’s own model, the usage limits for other models are exhausted so quickly that I feel pushed toward using models with higher usage limits.

I had such a poor experience with this application that it caused both a significant loss of time during my work and considerable emotional strain. I have now completely uninstalled Cursor and do not intend to use it again. I request a clear explanation regarding the problems I experienced, a review of the model and usage-quota policies so that users are not placed at a disadvantage, and appropriate measures to prevent similar problems from affecting other users.