It is bad to lose chats, but are you genuinely going back to week old chats to continue submitting requests or is it just for your records? If you need the model to be “trained” for a week on your project with all of these chats, you should be having it create md files or some other documentation that you can place into any future chat to give it awareness, since creating new chats with fresh context is an important aspect to using these models.
I don’t think its like that at all, at least not on the week scale. Maybe the hour or day scale.
I may be wrong, but I don’t think having a week of archived chats makes new chats more aware of your project. Chats don’t build some shared context model behind the scenes, right? If it does, I should probably stop deleting all my chats lol. I think the reason tons of people aren’t complaining is because it doesn’t really affect them that much if they lose a ton of chats, just short term if they were still working in a chat that was lost, or long term if they need to keep records of all the requests. Unless of course your workflow was to hunt down week old chats and start working in them again.
Also you don’t have to update Cursor when it prompts and you can get opt out of the beta if you are in it which should reduce your updates drastically.
Also other people are reporting this too