Hi. I have been reluctant to update to newer versions of cursor due to the many issues my colleagues have experienced. I have been on 1.6.35 and would prefer to stay. However, Cursor is now forcing me to update to the latest version. Is there a way to avoid this for now? Given all my deadlines I really can’t afford to deal with new issues or even getting up to speed on a new console. Would appreciate any helpful advice. Thank you.
6 Likes
same issue, newest versions are pretty much unusable, really need to stay on an older version
5 Likes
Same issue. It’s really annoying. I’m on 1.7 and “You are using a very old version.” Really? It’s 1.5 months old or so…
Dear Cursor Team, have you repaired the buggy 2.x mess? I don’t have time or money to be your beta tester!! I pay you $200 every month and I require stability. Not destroying my workflow regularly!
What is the current state of 2.x? I was auto-updated to 2.0 more than a month ago - the effect was that I had to pay 4-5x more money for 2x slower software, and the results are 10x worse. The agent in 2.x is behaving as if he knows everything better than me and doesn’t want to follow my instructions + does a lot of things I don’t want him to do. It’s completely unusable!
And if you want me to update, then change your release policy and finally start actually testing new versions before you release them and break people’s work.
I propose that in new versions you provide multiple agent versions - let 2.x stay for people who don’t know what they’re doing and need the model to think 10 times before doing anything. But also let power users use the tools like in 1.6 and 1.7, since they worked very well!
I’ve been your loyal user for a year now and I really don’t want to look for alternative solutions and switch to competitors, but 2.x looks like an attempt to get rid of me. And now, constantly nagging me about the update, I feel like you really don’t want users like me.
4 Likes
Gonna be real here - I don’t think they care at all.
Like they genuinely do not care if its an issue for you, me, or other users like us.
-
We have been consistently asking that they do provide proper changelogs all the way back since the start before 0.5 etc.
→ They refuse:
Changelogs are still not properly provided, if they are provided they are done so post release and as far as I know never actually cover all actual changes. -
Perfectly reasonable requests that they stop providing obviously broken patches that has seen no actual testing, instead of using userbase to find obvious flaws and broken implementation.
→ Partial:
At least the no longer force updates without the option to opt out, like it used to be and you had to spend time to disable the damn thing yourself. However, they still don’t understand the fundamental concept development stability, and will persist in being intrusive to make you change your workflow.
I really hate this one, because they don’t seem to understand that I don’t want to randomly have to change my workflow, configure a new environment, have stuff that worked previously not work, adjust to UI changes, troubleshoot whatever random issue suddenly appear etc. Cursor is a tool, not the product.
If I’m making dinner I don’t want my oven to suddenly ask me to upgrade its software to use the new eco-induction and smart oven with a new UI that may or may not work.
I’m making food, I don’t care. And when I’m done cooking I still want to just keep using the same oven because I don’t feel like learning to use a new configuration only for that to change to yet another new one 2 weeks out - oven isn’t my main interest or hobby, I just use oven to make food.
It is quite obvious that they don’t want people to stay on previous versions, because they require the userbase as their beta testers, hence why there are no LTS either.
I genuinely do not think the developers of Cursor has any kind of development experience themselves, because if they did they would know how stupid their whole approach is.
Unstable, frequent patches without proper changelogs or clear end goal.
On that note, the lack of any actual direction is the major issue in all this.
The problem is that the patches aren’t even clearly moving in a productive direction, it’s more about persistently performing changes for the sake of generating new hype words to throw out at new users and have social media influencers review - never mind that it’s not actually maintaining the actual workflow or provides increase in user satisfaction or quality.
Just throw stuff in there, mess around with the UI, remove something, rename something and re-introduce something we removed with a new name.
Happy coding.
4 Likes
The consensus out there seems to be that 1.7.12 is the most stable release before 2.x so I installed that and I’m back in business, for now. I still have to deal with the nagging message that I need to update to the latest version. I included the github url below.
3 Likes
Well this answers my question on my other topic thread - one should not upgrade from 1.7 if one has a working workflow! But for sure, I would like them to remove the message to upgrade - I know there is another version and will get to updating to it when I don’t see a sea of bug reports every time I open this forum 
1 Like
again yet another annoying forced “feature”…people are not ■■■■■■■ stupid, we use a earlier version for a reason
1 Like
same issue here. I was forced to update and now I cannot even use my terminal normally until i reload the window, which also works 50% of the time. very annoying and I hope I can use pre 1.6 version again soon
1 Like
As frustrating as it is, I feel the same way. This is a feature firehose to prod. That’s what Cursor has become. Even new features, which are always half-baked, buggy or incomplete, get left behind immediately by the next release, and nothing is ever done to fix the growing cesspool of bugs in the product.
A GOOD product balances features and bug fixes. This is standard. There has been a very LONG STANDING process of cycling between new feature implementation, and maintenance releases. Sadly, Cursor seems to completely ignore the latter. There are severe bugs that have left critical features COMPLETELY broken for many months now. They had DEVASTATING fundamentally broken terminal issues for about four months, and then they went and spent all this time adding a fundamentally borked “sandbox” mode that IMO is very dangerous, broken and basically unusable. The legacy Terminal works, but still has its issues…some of which have probably been present for six months now.
This endless march towards new features with not one iota of care given to producing a QUALITY, STABLE product, for a tool that professionals use every moment of their working lives, is very frustrating. The new diff/review mode stuff is a joke, and it causes severe problems over time as it seems to hang onto diffs LONG after they are utterly irrelevant (not to mention, the way it shows diffs is generally incorrect & useless for half the files that need diffing.) But…the march goes ever on… Features, features, features, to the detriment of everything else. Even recently-new features!
They are not listening to their community. Thing is, by not listening, not caring, the MOMENT a better product comes out that competes feature for feature with Cursor (i.e. has a WORKING Docs indexing and search feature!!) half their user base is instantly going to vanish, because they are so fed up of the broken features and frustrating user experience.
2 Likes
Sadly, as is with many companies in the tech and entertainment industry such as game development for example, there comes a point when strong enough market position and/or funds that you basically doesn’t have to care anymore - your position is simply self sustaining.
Cursor doesn’t have to care about the very few voices that bothers to have actual opinions or voice their legitimate concerns, like most here on the forum. We are a completely negligible part of the actual consumer base.
It’s the same way that companies like Adobe, YouTube, Blizzard, The Pokémon Company or Meta can just do whatever and it doesn’t really matter. For every disappointed of fed up user leaving, there are 10,000 new users signing up.
Only difference here is that I think that Cursor is sorely mistaken in how strong their position is and how stable the whole sector is.
However, even that doesn’t really matter to be honest - they are generating mad cash and investors are throwing boatloads of money at them, so if their goal is making money they are obviously doing the right thing.
And that is simply the main problem:
The goal of Cursor obviously isn’t to make a great product.
The goal is to make a lot of money.
The product itself is irrelevant.
I am also disappointed with the newer version.
the 2.x versions will NOT detect my browser login. It is a very bad decision to ONLY be able to login through a browser. We should have a login form inside the cursor software, and we should not depend on a browser to login.
When they force me out of 1.7, thats when I will quit cursor for good unless the new version starts working on my pc
1 Like
Well, if this is the case, it is very sad. It is amazing how greedy investors can utterly destroy a product to line their pockets. I don’t know that Cursor has quite that much investment, though, as it seems the whole AI landscape is in a pretty precarious position right now. Not that that gives users any power to get a poor quality product improved…
So it turns out they just killed 1.7 -,-
Great, now I have to scramble to find new tools and hope I can set up a workspace of similar quality to what I had before. Thanks Cursor team for pushing me towards the competition -,-
1 Like
It’s quite an interesting decision, forcing people to upgrade. The amount of bugs in the new versions is staggering, the forum is riddle with nothing but issues. And then they feel the need to spend time and energy, to punish satisfied customers on the 1.7 branch with that bug filled upgrades. And even if there wasn’t any bugs in the newer versions, why actively force new workflows on people who is contempt?
And it’s not like we’re dealing with a 10 year old LTE release, that would make sense to make deprecate at one point, we’re talking about a version that is a couple of months old.
Yeah, as of today 1.7 does not work anymore, forced into update 
Forcing an update. It’s barely been out, I can’t simply update mid workflow across projects
I already lose my chat history and even using external backup tools has proven that I need to build my own in-chat context backup system because I can’t recontext each of my chat-timelines - Dec 14 I stored my chats. Jan 2 its all wiped. I restore the physical files from my backup system and I still have 7 days of context wiped. Then I have to manually stitch back across several chats where I was, what the in context piece was, what the emphasis was on, where the chats was heading, how we were doing particular development in each case. I never know when my context is going to get nuked but if agentic development expects us to take on more and delegate mental load how are we suppose to trust that we scale our mental load when the platform lobotimizes itself frequently and sporadically and in such a way that we can’t reconstitute where we were. How do we scale intelligent use of LLM driven development if the core platform works against us retaining state !
Cursor is nothing but frustration hell and I simply do not know why I indulge this pain driven development experience.
And then you have changed the install location ( which isn’t editable anymore ) how many times now ? 3 ? Who does this !
Cursor must be run by the most inexperienced devs, severe memory leaks on the IDE itself, rife with inefficiencies and absolutely zero insight into their users.
I have zero loyalty for this brand infact I will actively steer people elsewhere
the day has arrived, Cursor 1.7 is no longer works at all,and cursor 2.X is still broken ,totally garbage.
Thanks to the warnings posted here, I was able to export my active chats. You probably saved me hours. Guess I’ll have to use some of that on finding the next most stable version. Guessing it’s a bit much to ask Cursor the courtesy of adding a few characters to their “This version is very old…” warning indicating when the version you’re using will be deprecated. Thank you all for the heads-up and I’m sorry for those of you who are set back by this. For anyone who doesn’t have it yet, here is the github with all the versions.