أميرة جنية الجليد
Elara Voss
متحفظة أولاً ومجزية لاحقاً، مناسبة لمؤامرات القصر وإذابة الجليد.
ابدأ الدردشة→دليل عملي
Use a concise Janitor AI persona template with copy-ready fields, examples, layer comparisons, privacy guidance and a repeatable six-turn test.
12 min read
إجابة سريعة
A Janitor AI persona should define the user role, relationship, communication style, durable preferences and boundaries in about 120 words or less.
بدء سريع
أميرة جنية الجليد
متحفظة أولاً ومجزية لاحقاً، مناسبة لمؤامرات القصر وإذابة الجليد.
ابدأ الدردشة→


Janitor AI Persona Template (2026): Examples & Setup
People searching for a Janitor AI persona template want a reusable way to describe themselves inside a chat. They are not looking for another character card or Lorebook. A persona should tell the bot who the user is in the current roleplay while leaving enough room for the conversation to develop.
This guide follows the format used by strong ranking tutorials: a direct answer, a field table, a copy-ready template, an annotated example, common mistakes, a test checklist and concise FAQs.
A useful persona contains only facts the character needs repeatedly: a fictional name, clearly adult age, role, relationship, communication style, boundaries and two or three durable preferences. Temporary events belong in Chat Memory; places and factions that activate only when mentioned belong in a Lorebook.
| Field | Include | Leave out |
|---|---|---|
| Identity | Fictional name, adult age and role | Legal identity or private details |
| Relationship | How the character knows you | The entire relationship history |
| Voice | Brief tone and reply preference | Rules for every possible scene |
| Boundaries | Clear behavior to avoid | Vague warnings |
| Stable traits | Two or three durable facts | Today’s mood or location |
Name: [fictional name]
Age: [clearly adult age]
Role: [job, archetype or story function]
Relationship: [how you know the character]
Appearance cues: [only scene-relevant details]
Communication style: [direct, playful, reserved]
Stable preferences: [two or three facts]
Boundaries: [behavior to avoid]
Current goal: [one durable reason to interact]
Keep the first version under roughly 120 words. Every sentence should change how the character understands or addresses the user.
Rowan Hale is a 29-year-old museum archivist and the character’s longtime research partner. Rowan speaks calmly, dislikes being rushed and prefers choices explained before action. They wear a green field jacket on expeditions. The character must not humiliate Rowan or decide Rowan’s actions. Their current goal is to recover a missing observatory map.
The example works because every detail affects dialogue or action. It does not prescribe the character’s feelings, write the ending or repeat world history.
| Layer | Main job | Update frequency |
|---|---|---|
| Persona | Defines the user role | Rarely |
| Character Definition | Defines the bot | Rarely |
| Chat Memory | Summarizes current events | After meaningful changes |
| Lorebook | Inserts keyed world facts | When world design changes |
| First Message | Starts one scene | At the opening |
The common error is repeating one fact in every layer. Put each fact in the smallest layer that needs it, then use the Chat Memory guide when story state changes.
Use observable traits instead of flattering labels. “Pauses before answering and asks one clarifying question” gives more direction than “intelligent and kind.” Describe relationship dynamics without controlling the other character. State boundaries as behavior, such as “do not narrate Rowan’s private thoughts.”
Avoid a negative list longer than the useful description. The model still needs positive direction.
Start a clean fictional chat. Ask the character to address you, offer two choices, move locations, introduce a wrong assumption, correct it once and request a recap. A good persona influences address, boundaries and stable preferences without forcing identical wording into every reply.
For a lower-setup comparison, choose a role in the LumiChat character library and repeat the same six turns.
Use fictional details and remove names, addresses, employers, health information and account data. Do not copy a creator’s private persona. Public screenshots are not permission to reproduce hidden configuration. Review the privacy guide before storing sensitive roleplay details.
It is the user’s persistent fictional identity and relationship context inside a chat.
Start below about 120 words and add only details that repeatedly affect replies.
No. Persona is stable identity; Chat Memory records what has happened.
Only appearance cues that matter to scenes or forms of address.
Use separate personas for genuinely different roles instead of mixing contradictory identities.
No direct import is promised. Recreate only details you own and are comfortable storing.
تابع في LumiChat
تصفح الشخصيات المتاحة واختر الأسلوب والمشهد المناسبين ثم ابدأ الدردشة فوراً.
دليل عملي
A field-by-field workflow that turns vague character traits into stable, observable behavior.
دليل عملي
Six scene openers that show character voice, create momentum and preserve the user’s agency.
دليل ذاكرة الدردشة
افصل ما ينتمي إلى Chat Memory وPersona والسجل الحديث وLorebook ثم اختبر الاسترجاع فعلياً.
دليل ذاكرة العالم
Lorebook معرفة عالم مشروطة وليست ذاكرة طويلة بلا حدود. يدخل الواقع المناسب عند ظهور مفتاح دقيق.