Hi, CursorAI team (I hope you monitoring your forums) and everyone,
I wanted to share an update regarding the ongoing issues with CursorAI (starting from version 0.46). After testing, I can confirm that reverting to version 0.45 resolves all major problems. Claude 3.5 Sonnet provides the same level of control over the codebase without introducing unnecessary changes. Additionally, version 0.45 does not have the frustrating bug where file edits are blocked, Cursor AI is creating a diff and a script to apply it instead of editing the file—something that has made the experience quite cumbersome.
A few thoughts for the CursorAI team, in case you’re following this discussion:
First and foremost, you’ve built an excellent product and were among the pioneers in this space, which is a remarkable achievement. However, it’s disappointing to see how turbulent things have become since the release of version 0.46. I’m not sure if this is due to internal constraints or resource limitations, but for a commercial product with a relatively high price point, reliability is crucial.
New features are always welcome, but not at the cost of breaking essential workflows. As developers, we rely on CursorAI to be a stable tool that enhances our productivity, not one that forces us to troubleshoot issues that previously didn’t exist. The time spent searching forums and workarounds is frustrating and counterproductive.
To help address these concerns, I strongly recommend:
-
A Dedicated Support Portal – A structured way to submit support tickets would improve user experience and ensure customers receive timely assistance. Forums are good for community support and free products.
-
Stronger Quality Assurance (QA) Processes – The current approach to QA appears inconsistent. Whether testing is limited to a small team or an individual, the impact of insufficient testing is evident. Automated testing is an industry standard for a reason, and without it, ongoing issues will likely persist.
-
Compensation for Disruptions – Given the strict request limits in place, it seems unfair that paying users end up exhausting their quotas just to test and troubleshoot product issues. A refund or credit for affected users—particularly those impacted since the release of version 0.46—would go a long way in maintaining goodwill.
The broader community has expressed similar frustrations, and some users have already started exploring alternative products. While competitors may not yet have the same issues, this should be a wake-up call. Maintaining a leadership position in this space requires both innovation and reliability.
I truly hope the team takes these concerns seriously and implements changes to prevent similar incidents in the future. CursorAI has great potential, but its success ultimately depends on the trust and satisfaction of its users.
Wishing you all the best and looking forward to improvements.