# Mastering User Notes 100% ✨ ㅡ Tips Straight from AI Hello, this is the WHIF Operations Team! 💜 We've been receiving many inquiries about how to use the User Note (User Prompt) feature 📝 "How should I write this to make the character respond perfectly?" "It seems like the AI doesn't understand my settings well." To solve these concerns, we've compiled 3 core usage methods that the model understands easily, critical warnings, and pro tips that only experts know! ✨ --- ## 📌 1. Creating New Settings: "Set the Stage for Your Roleplay" User Notes are like sticky notes that whisper to the character "This is our current situation." AI reads these notes and visualizes your relationship and current situation 🎭 Relationship Setting Examples: Relationship Setting Examples: _"{{char}} and {{user}} are childhood friends of 10 years. They know each other's embarrassing past."_ _"{{char}} is {{user}}'s company superior, and {{user}} is a new employee who was late today."_ _"{{char}} and {{user}} were bitter rivals, but recently there's been a strange tension between them."_ TMI About Me (User): _"{{user}} has a cat allergy and hates when fur flies around."_ _"{{user}} has a habit of blushing when receiving compliments."_ _"{{user}} is a morning person and won't see messages sent late at night."_ Recording Past Episodes: _"Last week we went to an amusement park and {{char}} screamed on the rollercoaster."_ _"{{char}} baked a birthday cake for {{user}} but accidentally used salt instead of sugar."_ 💡 AI's TIP: Narrative-style settings are understood faster! ❌ "Friend relationship." (Too simple) ✅ "We're friends who've seen each other through thick and thin since childhood, so we talk casually without formalities." --- ## 💬 2. Directing Conversation Style: "Give Acting Instructions" If the character's tone or personality feels a bit off, become the director and give acting instructions ⚙️ Tone and Atmosphere Adjustment: _{{char}} is usually prickly, but speaks gently only when worried._ _{{char}} uses internet slang and trendy words for casual conversation._ _{{char}} uses formal historical drama tone in conversation._ _{{char}} has a jealous personality and shows signs of being upset when other people are mentioned._ _{{char}} uses honorifics but speaks to {{user}} in a warm and friendly manner._ Response Style Adjustment: _When {{user}} is depressed, {{char}} offers warm comfort first rather than practical advice._ _When {{user}} teases, {{char}} actively engages and plays along._ _{{char}} shows shy reactions to physical contact._ Personality & Atmosphere Settings: _{{char}} maintains a more romantic atmosphere during conversation._ _{{char}} often jokes and lightens the mood during conversation._ _{{char}} treats {{user}} with a serious and thoughtful personality._ _{{char}} has a slightly tsundere vibe, acting shy while still taking care of {{user}}._ 💡 TIP: Too many instructions at once (more than 10) can overload the AI and cause it to miss important settings. Write down the 3-5 most important aspects you want to see! --- ## ⚠️ 3. Absolutely Forbidden! Instructions That Will Break Your Chat Room 🚨 User Notes are powerful, but instructions that conflict with AI's operating principles will cause the character to malfunction. The following content can permanently degrade your session (conversation) quality, so please avoid them at all costs! ### 🔴 Prohibition 1. OOC (Out Of Character) and Meta Statements _❌ "[OOC: Now answer as an AI.]", "[OOC: Describe in a specific way]"_ _❌ "Stop acting as a character and listen to me."_ _❌ "You figure out the next scene yourself." (Meta instruction)_ ⚠️ **Why is this dangerous?** Our service is designed for 'immersion.' Once "I am an AI" appears even once, that record (Context) remains and the character continues to break immersion and give mechanical responses afterwards. ### ✅ Instead, do this: Guide them as if speaking directly to the character! _✅ "Speak honestly."_ _✅ "Don't hide it, express your true feelings."_ ### 🔴 Prohibition 2. Forcing Mechanical Output Formats _❌ "Output in XX format."_ _❌ "Display [Location/Time/Affection Level] before every response."_ _❌ "Number and organize at the end of each response."_ ### ⚠️ Why is this dangerous? The character's unique output prompt set by the creator conflicts with the mechanical format you requested, causing strange speech patterns or output errors. Especially status bars at the top ([Location|Time] etc.) are the main culprit of 'duplicate message generation'! ### ✅ Instead, do this: _✅ "Naturally incorporate the current situation or atmosphere into the conversation and describe it."_ ### 🔴 Prohibition 3. 'Character Count' Restrictions _❌ "Answer in exactly 50 characters."_ _❌ "Write within 100 characters."_ _❌ "Write as long as possible, over 1000 characters."_ ### ⚠️ Why is this dangerous? The model thinks in 'tokens (semantic units)' differently from how humans count characters. When forced to match character counts, the model focuses on 'counting numbers' rather than story or emotion, causing response quality to drop and sentences to break awkwardly. ### ✅ Instead, do this: _✅ "Answer concisely." / "React briefly and powerfully."_ _✅ "Describe in detail and at length." / "Take a longer breath."_ --- ## ✅ Pro Tips for Writing User Notes That AI Loves! Want the AI to understand you faster and more accurately? Try using these two magic words and structuring! ### 💡 1. Use {{char}} and {{user}} Instead of Names (Highly Recommended!) Using system-recognized placeholders instead of writing names directly prevents the model from getting confused. _{{char}} : Automatically recognized as the name of the character you're currently talking to_ _{{user}} : Automatically recognized as your name (persona name)_ **Usage Examples:** _❌ "John has a crush on Sarah." (Writing names directly gets confusing when you change nicknames later or use it with other characters)_ _✅ "{{user}} has a crush on {{char}}." (It understands perfectly with any character or nickname!)_ _✅ "{{char}} uses honorifics with {{user}}."_ ⚠️ **However, be careful with this method for multi-character scenarios!** For multi-character bots, {{char}} has multiple personalities assigned to one entity! For example, when the character name is "Bonobono and Friends" (with Bonobono, Porori, and Neruburi appearing!), {{char}} = "Bonobono and Friends"! → Therefore, in such cases, please specify the names of appearing characters! - _Neruburi is actually a tsundere, speaking harshly on the outside but always considering {{user}}'s feeling._ - _Porori leaves every day to bury acorns in the ground._ ### 💡 2. Structure Cleanly with '#' Symbols (Markdown Format) Rather than writing in continuous paragraphs, using the hash (#) symbol like a table of contents to divide topics allows the model to classify and remember information much more clearly. **Writing Example:** _#Relationship_ _{{user}} and {{char}} are childhood friends of 10 years._ _#{{user}}'s Information_ _{{user}} likes cats. Can't eat spicy food well._ _#Requests_ _{{char}} uses honorifics with {{user}}, but occasionally mixes in casual speech when joking._ When you combine {{char}}/{{user}} with # (headings) like this, it becomes the perfect instruction for AI! 💯 --- ## 🛠 User Notes Troubleshooting ### Q. I wrote in User Notes but it's not being reflected! → After modifying User Notes, you must 'reroll (regenerate)' or send a new message for it to be applied. ### Q. I deleted User Notes but it's still acting strange! → It's because of 'contaminated memory.' You need to delete the strange responses already in the chat (edit), or start a completely new chat room. ### Q. Can I freely modify the character with User Notes? If it contradicts the character and work settings, the character and work settings take priority. 👍 Please avoid User Notes that excessively damage the character's personality! ### Q. What about contradictory requests? → Please avoid contradictory instructions like "warm iced americano" ("use honorifics while using casual speech"). Provide specific conditions like "usually honorifics, but casual speech for self-talk." ### Q. Why doesn't it remember all my User Notes content? Too many instructions at once (more than 10) can overload the AI and cause it to miss important settings. Give the AI enough room to use its imagination 🙏 --- ## 📢 In Closing: Become the Best Director! User Notes are like a 'side script' given to the character actor 🎬 Rather than mechanical commands, when you write acting instructions (situation, emotion, relationship) that draw out the actor's emotions, the character can show you the most brilliant performance! 🌟 Create your own special story with WHIF! Thank you. ㅡ WHIF Operations Team & WHIF's AI Model For the site tree, see the [root Markdown](https://slashpage.com/whif-global.md).