Over the past 20 days, I’ve had a rather strange and, frankly, frustrating experience using Cursor.ai in combination with Claude models (across all available versions).
I understand that a comment like this might cause some reaction — after all, we’re talking about constantly evolving technologies — but I don’t feel comfortable with what I’ve been observing.
Regardless of which version I select, the responses have been shallow and unproductive, often filled with dialogue that seems more like a conversation between two AIs than a real attempt to assist. Even basic issues — ones that a junior developer could solve with ease — are either ignored or answered with technically nonsensical suggestions.
What’s even more puzzling is that I do have an official Claude account, and when I use the models directly there, the answers are almost perfect. That sharp contrast is what makes this situation even more concerning. It raises the question of whether anyone is truly reviewing what’s happening with this integration between Cursor and Claude, because from what I’m seeing, something is clearly not working as it should — and that seriously affects trust in using these tools professionally. FAM
Check the forum to ensure the issue hasn’t been reported already
Provide a clear description of the bug
Explain how to reproduce the bug (if known)
Attach screenshots or recordings (e.g., .jpg, .png, .mp4).
Tell us your operating system and your Cursor version (e.g., Windows, 0.x.x).
Tell us if the issue stops you from using Cursor.