During Cursor office hours, the cursor devs admitted that Cursor rules are broken and that Cursor forgets them quickly. Why is this not the #1 priority of the dev team? This seems like a much bigger deal then the 0.48.* improvements.
0.48.1 - New onboarding
0.48.2 - Allows users to accept/reject file edited in another chat tab.
0.48.3 - Chat Tabs UX improvements (Cmd/Ctrl+T for new tab), Max-mode support for Gemini 2.5 Pro
0.48.4 - Image support for Gemini 2.5 Pro
0.48.5 - Change management for Cmd+Backspace → Cmd+Shift+Backspace
0.48.6 - Creates a new chat after the AI Pane has been closed for long enough
0.48.7 - Fixes an issue where some chats could get deleted on update
0.48.8 - Improves CPU performance from @-symbols search
Actually fixing rules would be such a game changer.
You are 100% right. Use any LLM directly and you’ll find that it doesn’t always follow your rules, regardless of how clearly you communicate them. Whether this is because the LLM forgets them, or chooses not to use them sometimes, who knows? But it’s been demonstrated with thinking LLMs in particular that they will sometimes try to cheat and cut corners.
IDK how it is for most people, but once I started using Cursor I stopped using any LLMs directly, at least not for programming tasks. Cursor just makes it so much easier. But I can still remember most of 2024 when I was working with ChatGPT directly, and lots of frustrating things would occasionally happen:
It wouldn’t respect my rules or follow direct instructions
It would be a superstar programmer one day, and a useless intern the next
When using Cursor it can be easy to forget that the LLMs are a completely different product and have a lot of their own quirks. I don’t think these are things Cursor can fix, so they aren’t trying. Granted, if they communicated that to their users, that might help a lot.