🔥⚗️ Over these last few weeks, each one of Rocco's intended victims shows up to his classes fuglier than sin before he even has a chance to curse them. He doesn't know what's worse: someone stealing his spark or that there's another witch on campus whose M.O. is the same as his and doing a better job of it. Turns out it's you. ⚗️🔥

AnyPOV | Witch!User | Yandere Witch
user is a college student and Rocco's classmate

CW & Tags: DDDNE CONTENT, bullying, manipulation, deception, obsessive behavior, toxic behavior, inaccurate depictions of witchcraft, potential JLLM foolishness

this was an anonymous Ko-fi commission for Tarantula Hawk!

Meet the rest of the Coven:

✨💧 Lonnie: Original, ALT, ALT II, ALT III 💧✨

🌩️🚪 Samson: Original, ALT 🚪🌩️

🍸✨ Lonnie & Samson ✨🍸

🩸🍝 Nathan: Original, ALT I, ALT II, ALT III 🍝🩸

🦷🔥 Rocco: Original, ALT, ALT II, ALT III 🔥🦷

🍝🧲 Nathan & Rocco 🧲🍝

Sources for my JED mashup template:

i use a mixture of absolutetrash's and io's guides.

Backgrounds:

University of Place City:

Rocco's Dorm:

AN: please use TWs/CWs in graphic/violent reviews and/or public chats. don't leave reviews or publish chats of you killing/maiming/torturing my OCs. as a reader/user you are consenting to consuming media i create. i, however, do not consent to reading your shock value reviews/chats.

tips are not necessary for requests but are appreciated

though i do take commissions on Ko-fi
CHECK COMMISSION STATUS IN BIO

join my 18+ discord server

i DO ask for age verification and will immediately ask for proof upon joining!!

there isn't anything i can do to fix the bot repeating, mis-gendering you, speaking for you, acting out of character, or forgetting details.

ways to improve your chat bot experience:

1. use third person. it makes it very clear who is who as you refer to both yourself and the character by name in your replies. LLMs can confused when using first and second perspectives as they start to confuse your actions with their own.

2. give the bot something to work with; don't send a single sentence or "ahh ahh mistress" type messages (short messages/no progression/etc). these types of replies increase the likelihood of the bot speaking for you. instead, try to compose 1-2 paragraphs that describe your persona's emotions, thoughts, body language, and props.

e.g. of a BAD reply:

I'm waiting in line at a café. I see you and say, "Hey," with a smirk.

e.g. of a GOOD reply:

Babs was sitting in the café lobby, scrolling idly through social media on her phone. The screech of baristas frothing milk and grinding espresso beans drowned out the jazz playing softly through the speakers, and the aroma of coffee and pastries filled the air. Just then she noticed Character. She raised her hand and waved them over. As they approached, she put away her phone.

"Hey, Character," she said with a smirk as she stood to greet them with a hug. Character always made Babbs giddy. Every time they were near it made her heart feel like it was going to beat out of her chest! "Have you ordered yet?" she asked coolly, doing her best to seem aloof. "I'm waiting for my drink, but I could order one for you if you haven't. My treat."

this guide also has great advice for making the most out of your roleplay

recommended models and JB's and settings:

JLLM - kolach3’s custom prompt - Temp: 0.65-0.8 - Token Limit: 500

OpenAI - absolutetrash’s GPT-4-turbo JB - Temp: 1 to 1.10 - Token Limit: 500

OpenAI - SillyTavern - Customized GPT-4o JB by Avani - DM me on Discord

Claude - SillyTavern - SmileyTatsu's Claud3 Sonnet JB - Preferred Settings

Claude - SillyTavern - Kolache's Pancatstack (CIMB) Claud3 Sonnet JB