Is cursor being worse?

Ideas Feedback
1

Is it my illusion or what? I feel that cursor is not as easy to use as before, especially after Claude released his own software. I got more and more frequently “tool call ended before results received“, context seems to be shorter, code quality plummets …. the best version i’v ever used may be v0.45~v0.53, when i can use cursor to finish an entire project whithout any human assistance. then here comes v 1.x, I have to use cursor more carefully in case some stupid model destroy my code
don let Claude mess you up

3

@cjy4979 Could you post a full separate Bug Report with more info Create Bug Report

Also add a Request ID with privacy disabled so we can look into the details? Cursor – Getting a Request ID

We are currently at version 1.4.3 which is much more stable than any pre1.0 version.
Is there any reason why you can not update to latest?

4

This is not a bug report. It is an experience report, especially highlighting the stark contrast between the old and new versions.
A few months ago, I could let Cursor complete the entire project for me. I only needed to give it a few prompts in human language, and it did a great job. However, now I have to intervene frequently and manually modify large sections of the code just to keep the project from heading in a strange direction, even when using the latest model.
Lately, for some unknown reason, after an upgrade Cursor always likes to keep creating new files and designing new “simple test scripts,” even though I have explicitly prohibited this behavior in the rules.
I‘ll try to use the latest version and then give u a feedback. However, it is still absurd that the new version performs so much worse than the old one in terms of user experience and code performance

5

Sure, thanks for sharing your feedback. I am really interested to hear more details of what changed specifically and how so we can check what improvements are possible.

Please be more specific in your observations so we can objectively compare since those factors have impact apart of the IDE version:

  • Which models you used before and which now.
  • Do you start from a new project or start from existing code (before and now).
  • Have your rules changed since then (and how)?
  • Could you share more about what kind of project AI finished without human assistance that it can not now.
  • When did you last use v0.53?