Guys, the cursor performance of the cursor is really bad, for some reason, how big the chat conversation become, more slower the App is, and my computer is a good one M1.
I’m considering leaving in to Zed, the unique reason I didnot moved yet is because I like GPT-5.
Hey, how many chats do you have in your project? We don’t recommend keeping long-term chats, since Cursor isn’t really suited for that. You can export some important chats into a Markdown file and refer to them when needed. Try deleting part of your history, maybe the oldest chats, that should help.
You can try renaming your project or moving it to a different location on your disk, this might help clear your history.
where I can find these histories in disk, so I can delete manually and other cache related file ?
All your chat history is stored at these paths:
Windows: %APPDATA%\Cursor\User\workspaceStorage
macOS: ~/Library/Application Support/Cursor/User/workspaceStorage
Linux: ~/.config/Cursor/User/workspaceStorage
I deleted eveything inside, gonna use it for a while and see if it get’s better. Thank you
It would be helpful for Cursor to have a built-in tool for deleting old chats. I have noticed a slowdown lately as well. I suspect I have a huge chat history, and some of my chats are very large as I’ve had context-heavy tasks that ended up taking longer to solve than initially expected, and they ended up quite large (starting a new chat would have resulted in a lot of lost knowledge, reeducation of the agent, etc.)
An agent chat explorer and management tool, would be quite welcome. I do like to keep certain chats around, so I can’t just bulk delete them all. I often have research or planning chats that I end up needing to go back to once I actually get to a point where I can actually implement those ideas. Sometimes I record them in Linear our project management software, but nto always.
Anyway, some kind of agent chat management tool would be very welcome at some point.
Cursor team: Thought I would note this example here; 1.5 should be considered an exemplar of a very bad release. From various sources of feedback, I think the general experience with 1.5, has been poor to terrible. It should be marked as an example of what never to do, with a product like Cursor. There were just way too many bugs in 1.5, for it to ever have been considered ready for mainstream usage.
@jrista this product will inevitably die, people will continue leaving one after another.
The CORE team are taking pretty bad decisions, the releases are being a total disaster and the alternatives are just increasing. Unfortunately, because I really like Cursor, but we don’t have option, It’s impossible to work in a tool so slow like this. I already canceled my two subscriptions.
Other thing is that the Team is not taking our complaints too seriously, and when they do, will be too late. They need to focus URGENTLY on performance, in every corner of the IDE.
Perhaps. Time will tell. I like Cursor, and I want it to survive, so I’m doing what I can to bring to light the severity of the issues that have been mounting with each release, and particularly with 1.5. It was a very bad release, and something the Cursor team should not allow to happen again.
I don’t know that it is so much that they are not taking our complaints seriously. I think they are. I think it is more that, they are not vocal in the community, and they don’t seem to have a coordinated community team, so they aren’t interacting with us as much as they could or in the ways they could, to keep us apprised of what is going on internally.
I’ve made a lot of bug reports, and received comments from the Cursor team on them. They are not immediate, sometimes they take a little while to respond. I have also made a rather pointed and detailed feedback post about how bad the 1.5 release was, and that as professionals, we cannot be Cursor’s ongoing beta testers day to day. They have not responded to the post directly, however, they did react to it, and I think they are taking it to heart. It was a pretty hefty post, and I can understand them not responding to it immediately. However, I don’t think they are just ignoring it.
I think one of the Cursor team’s current failings is that they don’t really have a coordinated community interaction effort. Which, given the point they are at, is probably something they need. I think the team, the product, the company, would benefit from more coordinated, consistent, and proactive community interaction than they have thus-far had. I think a bit more transparency, and them being more careful about testing their releases before they shove buggy code out to make us their beta testers, would go a LONG way to stabilizing their community, bring more users into the fold, and help their product overall, primarily from a perception standpoint.
Sadly, I think a lot of the tech world is enamored with concepts like “continuous delivery” and “feature flags” and such. SOME aspects of technology, can really leverage those things. Cursor, IMO, is NOT a product where periodic instability, bugs, crashing, etc. is acceptable. We are not the billions-strong throngs of social media users, we are professionals doing a job. I mentioned this in my pointed feedback post a few days back. I’ll mention it more over time here. I think this is something the Cursor team needs to recognize, and figure out how to incorporate into their processes. Again, I don’t think they have ignored me, I think its a critical factor that they may not really have been accounting for before, and it might take some time for the impact of their current approach to not only us as professionals, but also to the companies we work for, to sink in, and for them to adapt.
I do, however, think that keeping the lines of communication open between us and the Cursor team, is the best way to help them UNDERSTAND OUR NEEDS, and accommodate them.
@jrista I appreciate the care that you have taken in making this comment, and that you elaborated very well on the discussion. I agree with you on certain aspects. I’m not here just to complain and criticize but to give feedback as a daily user of Cursor, and I don’t have much faith in a product that ignores such critical problems like performance for so long.
IDEs are becoming, in a certain way, a type of commodity, where there won’t be much difference from one to another in terms of functionality and we will choose based on soft requirements, like performance, overall look, etc. In a few more months, Zed will take users by storm because they are matching functionality with other IDEs and their performance is unbeatable.
I’m experiencing the same performance issue, but not related to chats. I mostly develop Flutter aps with a Mac M3 Max, and I need to reload the page very often, as the auto-completions and even navigating through the code (command + click) gets stuck. When it reaches its limits, the Dart process crashes and restarts. It was really fine in the version 0.51 but we were forced to update to 1.5, which has a severe performance degradation.
I really love Cursor and hope you guys can look at this issue as soon as possible! I’m having to “downgrade” back to VSCode with Copilot and Claude Code to be able to work, but that’s a bummer as Cursor’s exprience is much superior with its magic tabs and the AI code acceptance UX!