I want to request to have an option in which it allows users to disable emojis from being processed when agents are writing text to any type of file. It’s extremely aggravating trying to build when there are little images of clowns and horns all over my code. Please implement this ASAP. Emojis are not fun, funny or appealing to look at when building software at any level. I have tried saving multiple instructions in my rules to prevent emojis from being produced and nothing seems to work art all.
I did, and I have not had anymore issues after I understood the two dimensions of each rule.
I always ask the resonant Chinese Room to change the rules in their style so they are more abiding due to familiarity, with increasingly imperative grammar and formatting. ULTRATHINK, VERY IMPORTANT, ATTENTION, < CRITICAL >RULE< /CRITICAL>.
If that still doesn’t work, do RLHF with examples like:
instead of using unicode emoji like , use ascii text like ╰‿╯
DO NOT use the , simply output a blank space, we all know what it means.
For my liking, I actually enforce extra emojification, especially and , so I did manage to control this feature to my liking. The issue is that once a single emoji appears in your chat, it will infest everything so it’s best to guide the process.
Ok, then there’s no answer possible.
Or there is an answer and you won’t like it.
Because Cursor doesn’t have a piece of code hidden somewhere in their build where it says “// insert emojis into output even if the user doesn’t want to ”. The cloud models have this behavior built in, and you can only shape it with rules, which Cursor uses.
But Cursor offers their base rules file for EVERYONE, including me, and I want emojis. Do you want them to disable my emojis? I need them, because they represent symbolic information in my outputs, they represent functional code, some people even use them for variable names, if you “delete” them from the output, you just wiped functional code.
Ok, so you don’t want to disable this for everyone. Then a toggle is what you think will switch between the emoji and non-emoji prompt ruleset? Why do you think this would work, if users have attempted to fix this with prompts and it doesn’t work all the time?
I’m confused really about what you think is possible and maybe I don’t understand what you are asking for, but if you can fix it, you can fix it.
I do question why the model learned that its uses were normative - I think most of the models do this, some better than others. Is there no way to add a configuration item that would then adjust the system level prompt to add/subtract a prompt stating that user wishes not to use emojis?
I do get that there is a work around, so this would require some buy in from the community. But maybe to rephrase this with some kind of implementation idea (albeit, might be incomplete).
I have struggled to see consistency with cursor rules as well, so might be more to this where the rules just get ignored after losing some sort of context.
, use ascii text like ╰‿╯
and 
, so I did manage to control this feature to my liking. The issue is that once a single emoji appears in your chat, it will infest everything so it’s best to guide the process.