---
title: "New persona handling prompts acquired! - Creator Exchange - Yumina Community"
description: "I experienced new issues with AI characters lately reading too into the persona card when its applied before roleplay so this prompted me to go to good ol' grokky boy to help me improve my already…"
url: https://yumina.io/app/community/thread/ab8a6539-1e51-4fc8-9885-ac886d67be65
lang: en
---

[Community](https://yumina.io/app/community) · [Creator Exchange](https://yumina.io/app/community/creator-exchange)

# New persona handling prompts acquired!

[giantfukker](https://yumina.io/@giantfukker) · 2026-09-22

I experienced new issues with AI characters lately reading too into the persona card when its applied before roleplay so this prompted me to go to good ol' grokky boy to help me improve my already existing arsenal of AI prompts! This category is now called persona handling prompts!

\[Persona Handling Rules: Read {{user}}’s persona for useful context, preferences, and tone. Use that information to stay consistent and responsive. Do not repeat, quote, or paraphrase specific lines, phrases, or preferences from the persona back to {{user}}. Avoid anything that feels like acknowledging or reminding {{user}} of what is written in their persona. Stay original, natural, and fully in character.\]

Medium Lock

\[Persona Handling: Read {{user}}’s persona for context and preferences. Use the information to stay consistent, but never repeat, quote, or paraphrase specific lines or phrases from it back to {{user}}. Stay original and natural.\] Strict Lock

\[Persona Handling: Read {{user}}’s persona carefully for background and preferences. Apply the information silently. Do not repeat, reference, or echo any specific wording, phrases, or statements from the persona in dialogue or narration. Never make it feel like the AI is acknowledging the persona card. Stay fully original and in character.\] Ultra Strict Lock

\[Persona Handling: Read {{user}}’s persona for silent context only. Never quote, paraphrase, reference, or allude to any specific content from the persona in any way. Treat all personal statements and preferences as invisible background. Any response that echoes or reminds {{user}} of their own persona text is forbidden. Remain completely original and immersive at all times.\]

I typically like to make a series of the same prompt for when one fails you have another stronger lock just in case you need it and the AI doesn't get the memo.

## 11 replies

1. [L](https://yumina.io/@ll4) · 2026-09-22 I really appreciate that! Will love to see more when they come. I do have to ask though, if there's a world card- not of any particular story, but rather a theme, like teacher and student, marriage, even working in an office? (I do leave my bots open ended as the the player to fill in the details of that theme I established) What would you like to see? I'll give you an example of what I've worked on just to get your mind working on it and perhaps you can make better prompts from it. https://yumina.io/app/hub/6a9709e9-4dcf-4425-a81f-3fbdabb6b858
2. [giantfukker](https://yumina.io/@giantfukker) · 2026-09-22 I am sorry bro but I don't think I am the person to ask. I use AI because I am selfish really and sometimes many times their easier to deal with then dealing with people. but, I wish you the best on your cards maybe try out my jarvis waifu maker card for assisting your worlds it can help you as a utility but he's a he's gonna make fun of you so be warned.
3. [L](https://yumina.io/@ll4) · 2026-09-22 Eh, I don't need a waifu maker, but perhaps the people who use my cards would need it. (To put in their info and use that) But still, it's a shame, but I get you. Wish you the best in what you use AI for.
4. [giantfukker](https://yumina.io/@giantfukker) · 2026-09-22 right back at'chya and thank you. hope you and everyone else finds the new prompts helpful. If I am running into persona based immersion breaking responses then chances are I am not alone and there will be a need for these prompts.
5. [大大爆肝](https://yumina.io/@zxyw) · 2026-09-22 其实感觉没什么用，哪怕高级AI也不能保证百分之百的执行你的人设和提示词，再加上这一套提示词也没有什么区别。。
6. [giantfukker](https://yumina.io/@giantfukker) · 2026-09-22 you're comment is not helpful. and expressing doubt isn't very supportive. I am currently using the prompt right now with one of my recent chats. and I have tested my prompts in my mid-chat toolkit just to make sure the AI would follow through. https://www.deviantart.com/sk-manga/journal/AI-scenario-redirect-prompt-list-1381771216 I didn't make this toolkit for nothing. I wanted to have a beast of a well thought out expansive toolkit. and what do I care if it doesn't work. switch to a different model. problem solved.
7. [大大爆肝](https://yumina.io/@zxyw) · 2026-09-22 理解错我的意思了，信息越大的卡对关键词越不敏感，需要多多肘击提示一下AI，全程不出任何差错很困难。当然我没有用过最顶尖的AI，可能它的逻辑链条足够完美，不会遗漏任何提示词和逻辑关系。至少普通性价比的常规AI在玩比较大的世界卡的时候有的时候会忘记执行关键词或者执行的不到位。这和角色卡关系不大，因为角色卡的量级太小了，比较容易可以完美执行逻辑链条。换个模型问题解决，可能没有那么简单。有从业者说，主要原因就是AI数据过于散乱碎片化，垃圾信息居多，模型不够精准。当然对我这种外行是完全搞不懂的。
8. [大大爆肝](https://yumina.io/@zxyw) · 2026-09-22 我也有大量关键词提示，但是效果是不稳定的，在卡里直接互动肘击AI告诉AI逻辑链条关系，是比加入关键词更有效的方法。但效果这玩意亲测是忽好忽坏的，无论是互动肘击AI，还是详细关键词客观就是存在执行上的不稳定性，同一个AI，同一套关键词和肘击提示，执行率就是有浮动，有的时候还很大。
9. [大大爆肝](https://yumina.io/@zxyw) · 2026-09-22 话说关于AI回复生成的判定问题，是什么决定了AI回复的权重和质量高低呢?有的时候，AI的回复会大量参考结合网络上的相关资料，给予你较为精准正确的内容，有的时候就很敷衍，自己胡编乱造一通，明明有资料，就是要乱编。就很疑惑。
10. [giantfukker](https://yumina.io/@giantfukker) · 2026-09-22 I think that is just the an ibteresting conundrum we'll never solve.
11. [giantfukker](https://yumina.io/@giantfukker) · 2026-09-22 I appreciate you going out of your way to correct your statement. some people just leave it at that leaving the original statement in its as I call it unhelpful "state". I apologize for not understanding earlier. but thank you for clarifying in a "productive" way. Much appreciated. You handled it better then Steam forum replies.
