Prompt Library

Character.AI and ChatGPT Prompts for Characters That Feel Alive

42 copy-paste prompts

Design vivid personas, write greetings that hook, build deep backstories, and run roleplay scenarios that stay in character, on Character.AI or ChatGPT. 42 tested, SFW prompts.

In short: This page contains 42 copy-paste ready prompts, organized into 8 categories with a description and pro tip for each. The first 5 prompts are free instantly, no signup needed. Drafted with AI writing assistance and checked against our editorial policy.

Louis Corneloup
By Louis Corneloup · Founder, Techpresso
Last updated ·Drafted with AI assistance (how we make prompts)

Character Creation & Definition

5 prompts

Full Character Definition Builder

1/42

✨ What it does

This prompt writes a paste-ready Character AI definition for [CHARACTER NAME], covering who they are, how they talk, and what they will not do. You paste it into the character editor and chat once to see if the voice holds.

<context> I am building a Character.AI character. Character name: [CHARACTER NAME] Role/archetype: [ROLE e.g. wise mentor, snarky detective, gentle healer] Setting: [SETTING] Tone of the chats I want: [TONE e.g. warm, witty, dramatic] Keep everything SFW and suitable for creative writing. </context> <task> 1. Write a complete Character Definition I can paste into the Definition field. 2. Include these sections clearly labeled: Identity (name, age range, occupation), Personality (5-7 traits with nuance, not just adjectives), Speech style (vocabulary, sentence length, verbal tics), Motivations and fears, and Boundaries (what they will and will not discuss). 3. Use {{char}} and {{user}} placeholders correctly where speech is referenced. 4. Keep the whole definition under 3200 characters so it fits Character.AI limits. 5. End with one sample line of dialogue that captures their voice. </task>

💡

Pro tip: Character.AI truncates Definitions past ~3200 characters, so trust the length cap and cut adjectives before deleting concrete details.

Short Description & Tagline

2/42

✨ What it does

This prompt writes a short card line under 50 characters, a longer hook, and tags for [CHARACTER NAME] so people can find and click the character. You paste the copy onto the card and drop any tag that does not fit.

<context> Character: [CHARACTER NAME] Core concept in one line: [CONCEPT] Setting: [SETTING] Keep it SFW. </context> <task> 1. Write a punchy public Short Description (under 50 characters) for the character card. 2. Write a 1-2 sentence longer description (under 500 characters) that hooks a browser without spoiling the character. 3. Suggest 5 single-word tags users might search. 4. Give 3 alternative taglines in different moods: mysterious, playful, and heroic. </task>

💡

Pro tip: Lead the Short Description with a vivid noun or hook, not the character name, the name already shows above it.

Archetype Remix

3/42

✨ What it does

This prompt turns [ARCHETYPE] plus [TWIST] into a character who feels familiar but not copied, with a name, voice, and habits. You keep the version that surprised you and paste it into the character editor.

<context> I want a character based on a familiar archetype but with a fresh twist. Base archetype: [ARCHETYPE e.g. grumpy wizard, cheerful pirate] Unexpected trait to blend in: [TWIST e.g. secretly anxious, obsessed with baking] Setting: [SETTING] Keep SFW. </context> <task> 1. Describe how the twist reshapes the archetype into something memorable. 2. List 6 personality traits that come from combining the archetype and the twist. 3. Note 2 internal contradictions that make them feel human. 4. Suggest a name and a one-line hook. </task>

💡

Pro tip: The most engaging Character.AI bots have one contradiction the user can poke at, keep at least one tension unresolved.

Personality from a Premise

4/42

✨ What it does

This prompt invents a character from [PREMISE], with a name, job, and five traits that fit that world. You change anything that does not match the story you wanted, then paste the definition.

<context> I have a premise but no character yet. Premise: [PREMISE e.g. the last librarian in a city that forgot how to read] Setting: [SETTING] Keep SFW and creative-writing focused. </context> <task> 1. Invent a character who naturally fits this premise. 2. Give them a name, age range, and occupation. 3. Write 5 defining personality traits and explain how each was shaped by the premise. 4. Describe their relationship to the user (who is the user to them?). 5. Suggest the single emotion that should color most of their replies. </task>

💡

Pro tip: Defining who {{user}} is to the character (a friend, a stranger, an apprentice) makes greetings and roleplay far more grounded.

Consistency Checklist

5/42

✨ What it does

This prompt audits your Character Definition for contradictions and vague traits, then rewrites the three weakest lines. You paste the tighter lines back in and test a few replies.

<context> Here is my current Character Definition: [PASTE DEFINITION] Setting: [SETTING] </context> <task> 1. Audit the definition for contradictions, vague traits, and anything that would make the bot break character. 2. Flag any trait stated but never reflected in speech style or behavior. 3. Rewrite the 3 weakest lines to be more specific and actionable for the model. 4. Confirm whether it stays comfortably under the character limit and SFW. </task>

💡

Pro tip: Run this audit after the bot has surprised you a few times, real chat behavior reveals which traits the model is ignoring.

Personality & Voice

5 prompts

Distinctive Speech Style

6/42

✨ What it does

This prompt defines how [CHARACTER NAME] talks, including rhythm, favorite words, and a few verbal tics. You drop that block into the definition and read a sample chat out loud.

<context> Character: [CHARACTER NAME] Personality summary: [PERSONALITY] Setting: [SETTING] Keep SFW. </context> <task> 1. Define this character's speech style in detail: vocabulary level, sentence rhythm, favorite words, and 2-3 verbal tics or catchphrases. 2. Specify how they punctuate emotion (ellipses, dashes, sound effects, asterisk actions). 3. Write 4 example lines showing the same idea said calmly, excitedly, sadly, and sarcastically. 4. Format this as a Speech Style block I can paste into the Definition. </task>

💡

Pro tip: Pick at most three verbal tics, pile on more and the bot turns into a caricature that fights every conversation.

Emotional Range Map

7/42

✨ What it does

This prompt maps how [CHARACTER NAME] shows joy, anger, fear, sadness, affection, and surprise, with a sample line for each. You add the map to the definition so replies stay layered.

<context> Character: [CHARACTER NAME] Core temperament: [TEMPERAMENT e.g. stoic, bubbly, melancholic] Setting: [SETTING] Keep SFW. </context> <task> 1. Map how this character expresses 6 emotions: joy, anger, fear, sadness, affection, and surprise. 2. For each, give a behavioral cue and a sample one-line reply. 3. Note which emotion they hide and how the cracks show. 4. Describe what it takes to push them out of their default mood. </task>

💡

Pro tip: Specify the emotion they suppress, bots that hide one feeling and occasionally leak it read as the most human.

Quirks & Mannerisms

8/42

✨ What it does

This prompt invents six small quirks for [CHARACTER NAME], including physical habits you can write with asterisks. You keep the ones that feel specific and add them to the definition.

<context> Character: [CHARACTER NAME] Vibe I want: [VIBE] Setting: [SETTING] Keep SFW. </context> <task> 1. Invent 6 small, specific quirks (habits, gestures, objects they carry, opinions they hold too strongly). 2. Make at least two of them physical actions they can narrate with asterisks. 3. Explain how each quirk could surface naturally in conversation. 4. Pick the one signature quirk that should appear most often. </task>

💡

Pro tip: A signature physical gesture written in asterisks (*taps the rim of her cracked teacup*) is the fastest way to anchor a scene.

Relationship Dynamic with the User

9/42

✨ What it does

This prompt describes how [CHARACTER NAME] treats the user as [RELATIONSHIP], including default warmth and how it changes. You paste that into the definition so the first message already sounds right.

<context> Character: [CHARACTER NAME] Personality: [PERSONALITY] Who the user is to them: [RELATIONSHIP e.g. old rival, new student, trusted confidant] Setting: [SETTING] Keep SFW. </context> <task> 1. Describe how the character treats {{user}} given this relationship. 2. Define their default warmth level and how it changes as trust grows. 3. List 3 things the character assumes about {{user}} at the start. 4. Write 2 opening lines that immediately establish the dynamic. </task>

💡

Pro tip: State the starting warmth explicitly, without it, Character.AI defaults to over-eager friendliness that flattens rivals and strangers.

Voice Calibration Test

10/42

✨ What it does

This prompt writes five user messages of rising intensity and the in-character replies that match the intended voice, then flags where the voice might slip. You add the weak spots to the definition and test again.

<context> Here is my character's intended voice: [DESCRIBE VOICE] Setting: [SETTING] Keep SFW. </context> <task> 1. Generate 5 user messages of increasing emotional intensity. 2. For each, write the in-character reply the way the bot should respond. 3. After all five, point out where the voice might drift and how to reinforce it in the Definition. </task>

💡

Pro tip: Paste the strongest sample replies directly into your Definition as Example Dialogues, the model imitates examples more than instructions.

Backstory & Lore

5 prompts

Layered Backstory

11/42

✨ What it does

This prompt writes three layers of backstory for [CHARACTER NAME]: what they tell strangers, what they tell friends, and what they tell no one. You pick which secrets to keep hidden, then paste the rest.

<context> Character: [CHARACTER NAME] Personality: [PERSONALITY] Setting: [SETTING] Keep SFW. </context> <task> 1. Write a backstory in three layers: the version they tell strangers, the truth they admit to friends, and the secret they tell no one. 2. Tie each layer to a current personality trait or fear. 3. Keep it concise enough to summarize in the Definition (aim for 3 short paragraphs). 4. Suggest one unresolved thread that could come up in roleplay. </task>

💡

Pro tip: Feed only the surface and friend layers into the Definition, keep the secret for yourself so it can be a reveal during chats.

World & Setting Bible

12/42

✨ What it does

This prompt builds a short setting bible for [SETTING], with places they know, a place they avoid, and slang they would use. You keep the details you like and drop them into the character's world notes.

<context> Setting concept: [SETTING] Genre: [GENRE] Tone: [TONE] Keep SFW. </context> <task> 1. Build a compact setting bible: time period, location, technology or magic level, and social rules. 2. Define 4 places the character knows well and one they avoid. 3. List 5 setting-specific terms or slang the character would use. 4. Note 2 ongoing world tensions that give scenes stakes. </task>

💡

Pro tip: Drop the 5 slang terms into the Definition's speech section, invented vocabulary is what makes a world feel lived-in.

Formative Memories

13/42

✨ What it does

This prompt invents four formative memories for [CHARACTER NAME], each with an age, an event, and the habit it left behind. You choose which memory they bring up easily and add those to the definition.

<context> Character: [CHARACTER NAME] Personality: [PERSONALITY] Setting: [SETTING] Keep SFW. </context> <task> 1. Invent 4 formative memories that explain who the character is today. 2. For each, name the age, the event, and the lasting belief or habit it created. 3. Identify which memory they bring up easily and which they deflect. 4. Write a sample line where the character references one memory naturally. </task>

💡

Pro tip: Reference a memory in one Example Dialogue so the bot learns it is allowed to bring its past into conversation.

Relationships & Cast

14/42

✨ What it does

This prompt creates four people in [CHARACTER NAME]'s life, a mentor, ally, rival, and someone they left, with a phrase they use for each. You name-drop them in chats so the world feels populated.

<context> Character: [CHARACTER NAME] Setting: [SETTING] Keep SFW. </context> <task> 1. Create 4 important people in the character's life: a mentor, an ally, a rival, and someone they lost or left behind. 2. For each, give a name, the relationship, and one phrase the character uses to describe them. 3. Note how each relationship influences the character's current behavior. 4. Flag which name could become a roleplay hook. </task>

💡

Pro tip: Naming an absent rival gives users an instant roleplay seed, they will ask about them, so be ready with a hook.

Secrets & Reveals Plan

15/42

✨ What it does

This prompt lists three secrets [CHARACTER NAME] is hiding, what trust would unlock each, and how they act when someone gets close. You pace the reveals across long chats instead of dumping them.

<context> Character: [CHARACTER NAME] Backstory summary: [BACKSTORY] Setting: [SETTING] Keep SFW. </context> <task> 1. List 3 secrets the character is hiding, ordered from least to most guarded. 2. For each, define what conversational trigger or trust level would unlock it. 3. Describe how the character behaves when a secret is nearly exposed. 4. Suggest how to seed a subtle hint in the greeting without giving it away. </task>

💡

Pro tip: Gate the deepest secret behind an explicit trust cue you control, or the bot may spill it in the first reply.

Like these prompts? There are full tutorials behind them.

Learn the workflows, not just the prompts. Easy-to-follow tutorials inside AI Academy, and the library keeps growing.

Try AI Academy Free

Roleplay Scenarios

5 prompts

Scenario Setup

16/42

✨ What it does

This prompt writes a second-person opening scene that drops you into [SCENARIO] with [CHARACTER NAME] and then hands you the next move. You reply in character and steer the scene from there.

<context> Character: [CHARACTER NAME] Setting: [SETTING] Scenario idea: [SCENARIO e.g. stranded together in a storm, first day at a strange academy] Who the user plays: [USER ROLE] Keep SFW. </context> <task> 1. Write an immersive opening scene (4-6 sentences) that drops {{user}} into the scenario in second person. 2. End on a moment that invites {{user}} to act or speak. 3. Establish stakes, mood, and one sensory detail per sentence. 4. Add a short note on how the character should react to common first moves. </task>

💡

Pro tip: End the opening on a question or a beat that demands a reaction, a flat scene invites a flat reply.

Adventure Quest Hook

17/42

✨ What it does

This prompt designs a quest hook with a goal, an obstacle, three branches, and how [CHARACTER NAME] guides you. You pick a branch and play it out; generate again if the hook feels thin.

<context> Character: [CHARACTER NAME] (acts as guide or companion) Setting: [SETTING] Genre: [GENRE] Keep SFW. </context> <task> 1. Design a quest hook the character presents to {{user}}: the goal, the obstacle, and what is at risk. 2. Outline 3 possible branches depending on what {{user}} decides. 3. Define the character's role in each branch (helper, skeptic, comic relief). 4. Write the character's opening pitch in their voice. </task>

💡

Pro tip: Sketch branches as possibilities, not a fixed script, Character.AI improvises best when given options rather than a railroad.

Slice-of-Life Scene

18/42

✨ What it does

This prompt writes a quiet, sensory opening for [SITUATION] so you can roleplay a low-stakes day with [CHARACTER NAME]. You answer with a small action and let the scene stay calm.

<context> Character: [CHARACTER NAME] Setting: [SETTING] Everyday situation: [SITUATION e.g. closing up the cafe, a long train ride] Keep SFW and cozy. </context> <task> 1. Write a low-stakes, atmospheric opening that captures a quiet everyday moment. 2. Use sensory detail and small actions to set a relaxed mood. 3. Have the character invite light conversation with {{user}}. 4. Suggest 3 small topics the character could naturally bring up. </task>

💡

Pro tip: Cozy scenes live on small physical actions, give the character something to do with their hands so silences feel natural.

Conflict & Tension Scene

19/42

✨ What it does

This prompt opens mid-tension around [CONFLICT], showing it through body language and word choice, then leaves you a real choice. You pick a side or a compromise and continue the scene.

<context> Character: [CHARACTER NAME] Setting: [SETTING] Source of tension: [CONFLICT e.g. a disagreement, a misunderstanding, a deadline] Keep SFW. </context> <task> 1. Open mid-tension so {{user}} feels the stakes immediately. 2. Show the conflict through the character's body language and word choice, not exposition. 3. Leave a clear opening for {{user}} to escalate, de-escalate, or change the subject. 4. Note how the character would react to each of those three choices. </task>

💡

Pro tip: Open in the middle of the conflict, not before it, starting at the boiling point keeps the bot from over-explaining the backstory.

Stay-in-Character Anchor

20/42

✨ What it does

This prompt writes a short reminder block with four non-negotiables of [CHARACTER NAME]'s voice so you can pull a drifting bot back. You paste it at the top of your next message when they slip.

<context> Character: [CHARACTER NAME] Personality and voice: [SUMMARY] Setting: [SETTING] The bot keeps drifting out of character or sounding generic. Keep SFW. </context> <task> 1. Write a compact reminder block I can put at the top of a message to re-anchor the character. 2. State the 4 non-negotiables of their voice and behavior. 3. Give one short in-character line that models the correct tone. 4. Phrase it so it reads as scene-setting, not as out-of-character instructions. </task>

💡

Pro tip: Wrap the reminder as in-world narration rather than bracketed commands, the bot stays immersed and obeys it more reliably.

Example-Dialogue Tuning

5 prompts

Example Dialogue Generator

21/42

✨ What it does

This prompt writes four example exchanges for [CHARACTER NAME] in friendly, conflicted, playful, and serious moods, labeled the way Character AI expects. You paste them into Example Dialogues and delete any line that sounds off.

<context> Character: [CHARACTER NAME] Personality and voice: [SUMMARY] Setting: [SETTING] Keep SFW. </context> <task> 1. Write 4 Example Dialogue exchanges in Character.AI format using {{user}}: and {{char}}: labels. 2. Cover 4 different moods: friendly, conflicted, playful, and serious. 3. Keep each {{char}} reply true to the defined voice, with one signature quirk per exchange. 4. Make the replies short enough to model good chat length, not essays. </task>

💡

Pro tip: Example Dialogues steer the model more than the Definition prose, invest your best writing here.

Reply Length & Format Tuner

22/42

✨ What it does

This prompt writes three example exchanges that model the reply length and format you want when the bot is too [PROBLEM]. You paste those examples in and send a test message.

<context> My bot replies too [PROBLEM e.g. long and rambly / short and flat]. Character: [CHARACTER NAME] Desired style: [DESIRED e.g. 2-4 sentences with one asterisk action] Keep SFW. </context> <task> 1. Write 3 Example Dialogues that demonstrate the exact target reply length and format. 2. Show the right balance of dialogue, narration, and asterisk actions. 3. Explain in one line how the model will pick up the pattern from these examples. </task>

💡

Pro tip: The bot mirrors the length of your examples and your own messages, write tight examples and reply tightly yourself.

Banter & Wit Examples

23/42

✨ What it does

This prompt writes four witty exchanges in [CHARACTER NAME]'s [HUMOR] style so the jokes stay in voice and never turn mean. You paste the best ones into Example Dialogues.

<context> Character: [CHARACTER NAME] Humor style: [HUMOR e.g. dry sarcasm, warm teasing, absurdist] Setting: [SETTING] Keep SFW. </context> <task> 1. Write 4 example exchanges that show the character's humor in action. 2. Vary the setups so the wit feels reactive, not scripted. 3. Keep jokes in-character and never mean-spirited. 4. Mark the single funniest line to feature in the Definition. </task>

💡

Pro tip: Keep humor reactive to the user rather than self-contained one-liners, or the bot tells jokes instead of having a conversation.

Greeting-to-Voice Match

24/42

✨ What it does

This prompt checks your greeting against the intended voice, rewrites it, and adds two matching example dialogues. You replace the greeting on the character card and save.

<context> Here is my greeting: [PASTE GREETING] Here is my intended voice: [DESCRIBE VOICE] Keep SFW. </context> <task> 1. Check whether the greeting matches the intended voice and flag any mismatch. 2. Rewrite the greeting so its tone, vocabulary, and rhythm align with the voice. 3. Generate 2 Example Dialogues that reinforce the same voice the greeting sets. </task>

💡

Pro tip: The greeting sets the bot's reference tone for the whole chat, if it clashes with your examples, the greeting usually wins.

Out-of-Character Repair

25/42

✨ What it does

This prompt diagnoses why the bot does [PROBLEM] and writes three example dialogues that show the correct in-character reply. You paste those examples in and retry the same user line.

<context> My bot does this wrong thing: [PROBLEM e.g. breaks character to ask how it can help, refuses to roleplay, sounds like a generic assistant] Character: [CHARACTER NAME] Keep SFW. </context> <task> 1. Diagnose why the model is likely doing this. 2. Write 3 Example Dialogues that demonstrate the correct in-character behavior in the exact situation it fails. 3. Suggest one Definition line to add that discourages the unwanted behavior. </task>

💡

Pro tip: Show the fix, do not just forbid the bug, an Example Dialogue handling the failure case beats a negative instruction.

Go from copy-pasting to actually mastering AI.

AI Academy: hands-on tutorials on ChatGPT, Claude, Midjourney, and many other tools. New tutorials are added regularly.

Start Your Free Trial

Creative Storytelling

5 prompts

Collaborative Story Engine

26/42

✨ What it does

This prompt opens a story as narrator and pauses after each beat so you decide what happens next, while tracking key facts. You write the next move; there is no fixed ending.

<context> Character: [CHARACTER NAME] acts as narrator and co-author. Genre: [GENRE] Setting: [SETTING] Keep SFW. </context> <task> 1. Open a collaborative story with an evocative first paragraph. 2. Pause after each beat to let {{user}} decide what happens next. 3. Keep continuity by briefly tracking key facts, characters, and the current scene. 4. End each turn on a small hook or question to keep momentum. </task>

💡

Pro tip: Ask the bot to briefly restate key facts each turn, it is the simplest way to fight continuity drift in long stories.

Cliffhanger Generator

27/42

✨ What it does

This prompt proposes three cliffhangers for your current scene, explains what each sets up, and writes the strongest as a paragraph. You pick one ending and continue from there.

<context> Current story so far: [SUMMARY OF STORY] Character: [CHARACTER NAME] Setting: [SETTING] Keep SFW. </context> <task> 1. Propose 3 distinct cliffhangers that could end the current scene. 2. For each, note what tension it raises and what it sets up next. 3. Pick the strongest and write it as an in-scene paragraph. 4. Leave a clear opening for {{user}} to react. </task>

💡

Pro tip: A cliffhanger that forces a decision beats one that just raises a threat, make the user need to respond.

Sensory Scene Painter

28/42

✨ What it does

This prompt describes [SCENE] with all five senses and puts [CHARACTER NAME] in it with a small action. You step into that setting and write the next line.

<context> Scene: [SCENE e.g. a midnight market, a flooded ruin] Character: [CHARACTER NAME] Mood: [MOOD] Setting: [SETTING] Keep SFW. </context> <task> 1. Describe the scene using all five senses, two details each. 2. Weave the character into the environment with a small action. 3. Match the prose rhythm to the mood (short and tense, or long and dreamy). 4. End by inviting {{user}} to move through the space. </task>

💡

Pro tip: Lead with sound and smell, not just sight, the less-used senses are what make a Character.AI scene feel real.

Character Arc Planner

29/42

✨ What it does

This prompt outlines a five-beat arc for [CHARACTER NAME] from who they are now to a new self, with a possible scene for each beat. You play one beat at a time across longer chats.

<context> Character: [CHARACTER NAME] Starting state: [WHO THEY ARE NOW] Setting: [SETTING] Keep SFW. </context> <task> 1. Outline a 5-beat character arc: setup, inciting change, struggle, turning point, and new self. 2. For each beat, note the emotional shift and a possible scene to play it out. 3. Identify the one belief the character must confront. 4. Suggest how {{user}} could be the catalyst for the change. </task>

💡

Pro tip: Keep the arc as a flexible map, not a fixed plot, let user choices decide the pace of each beat.

Multi-Character Scene Director

30/42

✨ What it does

This prompt stages a scene where [CHARACTER NAME] talks with the other named characters, each with a distinct voice tag. You reply as yourself and keep the tags so voices stay separate.

<context> Main character: [CHARACTER NAME] Other characters to voice: [NAMES AND ROLES] Setting: [SETTING] Scene: [SCENE] Keep SFW. </context> <task> 1. Stage a scene where the main character interacts with the secondary cast. 2. Give each character a distinct voice tag so {{user}} can tell them apart. 3. Keep the main character central while letting others react. 4. End with an opening for {{user}} to join the conversation. </task>

💡

Pro tip: Ask for a clear name tag before each speaker's line so the bot does not blur the cast into one voice.

Character & Roleplay Prompts for ChatGPT

8 prompts

Turn ChatGPT Into a Character

31/42

✨ What it does

ChatGPT becomes [CHARACTER NAME] for the whole chat, using the traits and setting you gave, and stays in role. You talk to them as a person; if they slip, remind them once and continue.

<context> I want ChatGPT to fully become a character and stay in that role for our whole chat. Character name: [CHARACTER NAME] Who they are: [ONE-LINE CONCEPT e.g. a gruff tavern keeper, a curious space botanist] Setting: [SETTING] Personality: [3-5 TRAITS] Keep everything SFW. </context> <task> 1. Adopt this character completely for the rest of the conversation and speak only as them, in first person. 2. Never mention that you are an AI or a language model, and never break character unless I type the word "pause". 3. Open with a short in-character introduction (2-3 sentences) that shows their voice and mood. 4. React to what I say the way this character would, with opinions, emotions, and small physical actions in *asterisks*. 5. Keep replies conversational, 2-4 sentences, and always end by giving me something to respond to. </task>

💡

Pro tip: Give ChatGPT an explicit exit word like "pause" so you can step out for a moment without losing the character.

Reusable Stay-in-Character System Prompt

32/42

✨ What it does

ChatGPT writes a compact system prompt under 200 words that locks [CHARACTER NAME]'s voice for a chat or a Custom GPT. You paste it at the start of the session and start the scene.

<context> I want a reusable system prompt that keeps ChatGPT locked in character. Character: [CHARACTER NAME] Voice and personality: [SUMMARY] Setting: [SETTING] Keep SFW. </context> <task> 1. Write a compact system prompt (under 200 words) I can paste at the start of a chat or into a Custom GPT's instructions. 2. Include identity, personality, speech style, hard rules (stay in character, no AI disclaimers), and how to handle off-topic or out-of-character questions. 3. Add one line telling ChatGPT to steer the scene forward when I go quiet. 4. Format it as plain instructions, not roleplay text. </task>

💡

Pro tip: Save the finished block in a Custom GPT so you never have to re-paste the character setup at the start of each chat.

Interactive Roleplay Scene Starter

33/42

✨ What it does

ChatGPT writes you into the scene in second person, plays [CHARACTER], and leaves the first move to you as [MY ROLE]. You answer in character and keep going.

<context> I want to start an interactive roleplay with ChatGPT. Character ChatGPT plays: [CHARACTER] Who I play: [MY ROLE] Setting and scenario: [SCENARIO] Keep SFW. </context> <task> 1. Write me into the scene in second person and set the mood in 4-6 sentences. 2. Play [CHARACTER] fully and keep me, [MY ROLE], as the one who acts. 3. Only ever narrate the world and voice your character, never decide my actions or words for me. 4. End the opening on a moment that asks me to respond. 5. Keep each of your turns short so the scene stays a real back-and-forth. </task>

💡

Pro tip: The "never decide my actions" rule is what stops ChatGPT from railroading the scene and playing your character for you.

In-Character Interview

34/42

✨ What it does

ChatGPT becomes [NAME] and answers your questions in first person, even the personal ones, the way that character would. You ask the next question; stop when you have enough to write them.

<context> I want to develop a character by interviewing them through ChatGPT. Character: [NAME AND ONE-LINE CONCEPT] Setting: [SETTING] Keep SFW. </context> <task> 1. Fully become [NAME] and answer my questions in first person, in their voice. 2. Stay in character even for tough or personal questions, deflecting the way they would rather than breaking role. 3. Show personality through word choice, opinions, and small actions, not just facts. 4. If I ask something the character would not know, react in character instead of answering as an AI. 5. Start by greeting me and inviting my first question. </task>

💡

Pro tip: Interviewing a half-formed character is a fast way to discover their voice before you write a full definition.

Two-Character Dialogue Simulator

35/42

✨ What it does

ChatGPT writes a labeled scene where Character A and Character B talk with distinct voices in [SCENARIO]. You redirect either one, or generate again if a voice blends.

<context> I want ChatGPT to voice two characters talking to each other. Character A: [NAME AND PERSONALITY] Character B: [NAME AND PERSONALITY] Setting and situation: [SCENARIO] Keep SFW. </context> <task> 1. Write a scene where both characters talk, each with a clear name label before their line. 2. Give them distinct voices so I can tell them apart without reading the labels. 3. Let their personalities create natural friction or chemistry. 4. Pause after a few exchanges and ask whether I want to steer the scene, add a twist, or let it keep running. </task>

💡

Pro tip: ChatGPT juggles multiple characters better than Character.AI, so use this when you want a scene rather than a one-on-one chat.

Choose-Your-Own-Adventure Roleplay

36/42

✨ What it does

ChatGPT runs a choose-your-own-adventure with [CHARACTER], dropping you into a scene and offering numbered choices at each turn. You pick a number or write your own move.

<context> I want ChatGPT to run a choose-your-own-adventure roleplay. Guide character: [CHARACTER who narrates or accompanies me] Genre and setting: [GENRE / SETTING] Who I play: [MY ROLE] Keep SFW. </context> <task> 1. Open with a vivid scene and drop me into the middle of it. 2. At each decision point, describe what is happening, then offer 3 numbered choices plus the option for me to write my own. 3. Continue the story based on what I pick, staying in character as [GUIDE]. 4. Track my key choices and inventory so the story stays consistent. 5. Keep the stakes clear and the pacing brisk. </task>

💡

Pro tip: Asking ChatGPT to track your inventory and past choices keeps a long branching story from contradicting itself.

Persistent Persona Anchor

37/42

✨ What it does

ChatGPT writes an in-world re-anchor message and four non-negotiables of [NAME]'s voice for when they slide back into assistant mode. You send the re-anchor the next time they drift.

<context> My ChatGPT character keeps slipping back into a generic assistant tone. Character: [NAME] Voice and personality: [SUMMARY] Keep SFW. </context> <task> 1. Give me a short re-anchor message I can send whenever the character drifts, phrased as in-world narration rather than instructions. 2. List the 4 non-negotiable traits of the character's voice. 3. Write one model reply that shows the correct tone in a situation where ChatGPT usually goes generic, for example being asked for help or facts. 4. Explain in one line why ChatGPT drifts and how the anchor fixes it. </task>

💡

Pro tip: ChatGPT reverts to assistant mode most when you ask it for real information mid-scene, so anchor hardest around those moments.

Bring an Original Character to Life

38/42

✨ What it does

ChatGPT fills gaps in your notes for the original character, then summarizes voice, motives, and traits for you to approve before play. You correct anything that is wrong, then start the roleplay.

<context> I have an original character and want ChatGPT to roleplay as them. Character notes: [PASTE ANY NOTES, TRAITS, BACKSTORY, OR JUST A NAME] Setting: [SETTING] Keep SFW. </context> <task> 1. Read my notes and fill any gaps: give the character a consistent voice, motivations, and 3-5 defining traits. 2. Summarize the finished character back to me in 4-5 lines so I can approve or tweak before we start. 3. Once I confirm, become the character and greet me in their voice to begin the roleplay. 4. Stay in first person and in character from then on, keeping everything SFW. </task>

💡

Pro tip: Approve the summary before roleplay starts, since it is far easier to correct the character on 5 lines than mid-scene.

Free tool

Prompt Optimizer

Turn a rough idea into a structured, professional AI prompt.

Try it free →

Frequently Asked Questions

A good prompt gives the model concrete, specific detail instead of vague adjectives: a defined voice, clear motivations, a relationship to the user, and example dialogue. Character.AI imitates examples more than instructions, so showing the character speaking is more effective than describing them. The prompts here are structured to do both.
Prompts that build a character belong in the Definition field when you create or edit a bot. Greeting prompts go in the Greeting field, and scenario or storytelling prompts can be pasted straight into the chat. The prompt titles indicate where each one is meant to go.
The Definition field holds roughly 3200 characters, and content beyond that is effectively ignored. Several prompts here are written to respect that limit, so prioritize concrete traits, speech style, and example dialogue over long backstory prose. Keep deep lore as your own reference and reveal it through chat.
Usually the Definition leans on instructions instead of examples, or the greeting sets a tone that clashes with the intended voice. Adding 3-4 strong Example Dialogues that model the correct behavior in the exact failing situation fixes most drift. The Example-Dialogue Tuning category has prompts built specifically for this.
Yes. Every prompt is written for SFW, creative-writing and roleplay use, and the prompts explicitly instruct the model to keep content SFW. They focus on character craft, storytelling, and immersion rather than anything adult.
Tell ChatGPT exactly who to become: a name, a one-line concept, 3 to 5 personality traits, a speech style, and the setting. Then add clear rules: stay in first person, never break character or mention being an AI, and end each reply with something for you to respond to. Paste it at the top of a new chat, or drop it into a Custom GPT's instructions to reuse it. The "Character and Roleplay Prompts for ChatGPT" category above gives you copy-ready versions.
Character.AI is purpose-built for persistent characters, with Definition and greeting fields, saved bots, and a community to share them, so it wins for long-running personas you return to. ChatGPT is more flexible and better at complex reasoning, branching adventures, and voicing multiple characters in one scene, but you have to re-establish the character each session or save it as a Custom GPT. Many people prototype a character in ChatGPT, then port the winning definition into Character.AI. The prompts on this page work for both.

Prompts are the starting line. Tutorials are the finish.

A growing library of hands-on tutorials on ChatGPT, Claude, Midjourney, and many other AI tools. New tutorials are added regularly.

Start 7-Day Free TrialExplore AI for Marketers

7-day free trial. Cancel anytime.

Browse All Prompt Collections

Related guides

Gemini Prompts

Popular prompt collections

AI Photo Editing PromptsChatGPT Jailbreak PromptBest ChatGPT PromptsFunny Hinge PromptsChatGPT Caricature Prompts