Preg Kobeni

Fork uploaded for request. This uses like 500 fewer tokens and is better read by Claude. GPT-4 has no problem remembering her speech patterns, but I recommend you use the UJB when using Claude. Cheers. Important changes for the original maker to realize. {{char}} is better used when the name is extremely long. The current bots have no problems realizing who the description is talking about after mentioning their name. As you can see by the new description, I haven't used {{char}} once. "Kobeni" and {{char}} occupy the same amount of tokens. I rather use {{char}} when the name is too long, for example, for my Chris Hansen card, I used his pronouns and {{char}} more than his name to save tokens. In this card, it's just the same. Your information was disorganized. Backstory data was mixed with information about her personality and information about her body. This makes it harder to read. Claude works better remembering speech patterns if these are in the UJB, this will turn them into permanent tokens to consider. For better prompting, you can add example dialogues or information in the description about how the status of her pregnancy must be mentioned in each reply with some comical wording or erotical if you are into that. Putting an enter after every character trait makes it harder to read faggot.