Apparently, per a previous response, the devs assessed that tool lock down on custom agent modes “was not statistically significant”, so they assumed the feature wasnt a winner.. Versus perhaps going a step closer and recognizing that the users that were/are using that feature are generally power users building on top of their product. Or at least, that is my hope, versus it being something more nefarious like they don’t want power users in support of a agentic first approach (not that the two have to be mutually exclusive) .
1 Like
We are working on better way to handle this but it is not yet ready to be released.
5 Likes
A better way to handle this would be to just give us back the feature we loved!!! As someone who’s been a head of product before, this is insane that users have to even ask not to have the rug pulled on their favorite features
4 Likes
ETA? Can you temporarily bring back modes while you finish whatever you’re doing right now? Your message should be sent to all users BEFORE removing custom modes. You can’t treat your clients like that - degrade the quality of your product without proper transparency and react with mysterious message only when there’s lashback. “Big things coming” or “trust me bro” type shi… do better, because there’s a looooot money coming in.
I agree that communication is key here. Ideally, we should see prior announcements before major changes, or—if a feature needs to be removed temporarily or critically adjusted—a clear message explaining that an improved version is in the works. A dedicated email update or a newsletter summary would definitely help prevent these surprises and keep everyone in the loop.
I will pass this specific feedback on to the team to help improve how these transitions are communicated in the future.
I’ve never said communication is key. Actually, the team could work on “improved version” and introduce it without removing the old one. Maybe even do some A/B testing to actually check what clients/users want. Improved version according to who? This is not Apple or Google to trust your judgment guys. Be humble and listen to various groups of your users
3 Likes
Agree with Koleboxx - it’s not just about communication. You are now responsible for the workflows of a large number of teams. You guys need to think before just pulling the rug on features that are critical to people’s workflows. If this many people are upset enough to take the time out of their day to post in this thread, you are doing something wrong from a product management standpoint. This needs to be communicated to your product team
3 Likes
Absolutely! I will admit I had never been on this forum before discovering the removal of custom modes. I could deal with the little inconsistencies until then. But after that, I had to get on, this is such a huge miss-step… I gotta hope they will take note, and correct.
Please give the old custom modes feature back, you have now jeopardized a wider team rollout of cursor
1 Like
Custom Modes previously provided a persistent, selectable mode with guaranteed rule enforcement across the entire conversation. In 2.1.x, that capability was removed, and neither @file references nor slash commands replace it—they’re single-turn injections that don’t persist, don’t isolate behavior, and don’t ensure consistency. As a result, workflows built around persistent modes can’t be reproduced under the current system.
If you love the “old custom mode” feel free to open a thread in Feature Requests
Love is a strong word to describe it, but what was there worked vs now it is unusable.
I will add a thread there, but i agree with the original poster here that this is a regression, not a feature request.
I added a new topic here, additionally I sent an email to support.
2 Likes
I’m with you SanK! Custom Modes wasn’t perfect but it gave us something usable to professional developers, rather than the stinking mess the latest few builds have been.
1 Like
Custom modes is currently the #1 feature request. Two minor releases and dozens of patch releases since they pulled it, and still no adequate replacement.
Sadly, I think Cursor peaked at 2.0.77 and I’m just too busy ATM to switch my team to a different IDE. Another upside of sticking with the old version is I’m no longer greeted with a wave of regressions and fresh bugs each update.
1 Like