Hi I’m getting a number of queries simply stalling. I’m a premium user and this has neutered my use of cursor.sh – anyone else experiencing this?
The same is happening to me. Cursor is really getting bogged down with the AI outputs, especially the 200k long chats. It seems there would be a way to separate the processes of the IDE away from the AI functions so the IDE runs fast while the AI can take it’s time with an output. I have had a few crashes with large codebases as well and getting bogged down. I am using cursorignore files in every project so the indexing does not get bogged down. Some prebuild curosignore files would be a nice touch to start with automatically to keep the chat from searching all the system files. It always seems to go straight for the bin and pycache first instead of the code we need to work on, but i have also been adding the files needed, but that has been getting a pain to repeat three to five folders and 10 files at once.
There’s no question Cursor is likely suffering from its own success – it’s truly a next-gen AI companion I’m sure the world will quickly get used to. But it’s right there in the inner loop of software developers so it’s only going to work if the response times are consistently good. Also, Cursor uses out of date data (using just the generic LLMs) so it’s bordering on useless for any API that’s changed in recent months (but luckily that never happens right? 
)
I’m really noticing a slow down for sure.
Thanks for reporting this.
Are you noticing this in one specific feature (i.e. chat) or everywhere?
Are you using Claude, by any chance?
This is very helpful, thanks!
Do you remember if Cursor crashed during a 200k long chat, or during normal code-editing?
Yeah just chat with the animated dots not responding, 30 seconds, sometimes up to a minute.
Do you notice if the autocomplete works?
It was happening when I was using the long chat with Gemini 500k. And like my post before, I was using cursorignore files and only giving it two or three folders and maybe 6 scripts. I like that there is a message of how many tokens are in each files as it is added but it disappears too fast. It would be nice to have a token counter on the chat boxes so we could delete, add, based on the tokes, and a max token warning as well. I also have not seen the reprioritizing function working, but maybe that is because I am leaving out all the system files and keeping it to the code.
1 Like
Im using composer and I’m stuck at the perpetual Generating… loader. I’m not sure if its a slow down or the process is stuck. Additional logs would definitely help. What do you guys regard as slow? Im using Claude Opus to generate a full next js app. How are you guys addressing these issues?