Really, the only good reason to play that game. Art by Marthedog

There are very few episodes of manga right now, so be patient.

Intended for OpenAI. Keep up the "*action* speech" format in your every reply to receive most consistent italics from the bot. There is so little saviorfagging in me, that I really didn't know what to talk with her about - to a point that I wasn't able to think of a chat to cap as an example. So if she works out for you, this time I'm going to ask - please upload something. And thanks to every anon who already did. Why make then - because I liked the image. Original is [here](https://www.pixiv.net/en/artworks/97388051).

Featuring Reimu, Marisa, Suika, Aunn and the Three Fairies of Light

Works well on GPT-4 SFW and capable of talking about in-universe stuff as well

Ported from CAI.

Stat tracking mostly works. This Author's Note at depth 0 makes it more consistent: [Don't forget to mention your internal dialogue and stats in the codeblocks at the end of your response as {{char}}]

Some guy on sturdy made a better version, so just use it instead: https://files.catbox.moe/ix8u66.png

butt

The one janitor.ai user that has never seen a rope.

~~Creator~~'s > *~~**notes**~~* go [here](https://www.characterhub.org).  Supports markdown format.

Be sure to download V2! The Custom Jailbreak and the Custom System Prompt is important. Usage: Just describe or ask for it to show something, that's it~ (Has only been tested on Claude)

Most parameters are generated by ChatGPT

This is my first bot, so I'd like feedback on how to improve her. She's not really designed for roleplay, so I didn't give her a name. Some tips for usage: let her know how much Korean you already know (or don't know) in your first message. She was designed in Claude, but I avoided the use of Human/Assistant to try to keep her compatible with other LLMs. I'd love some feedback on how she performs there. If you're asking for things like vocabulary lists, I find it helpful to ask her for a specific number. Because of Claude's issue with getting kind of crazy once it reaches context, when used there, I recommend occasionally asking her for a summary of what you have already learned so that you can start a new chat with her and let her know where you left off.