cardex

kbat

Chub· 152 cards· 0 lorebooks· active –

/// requests /// discord /// bots ///

plushy

check the "bots" link to tell which bots are mine, and which are forks, and to see them sorted todo: looking to recreate rough old bots (so many lack good greetings) and recreate some lost favourites from janitor
• my bots vary between ~1k to a ludicrous 10k tokens, bots are focused for huge context size corpo models (like ChatGPT/Gemini)
• i tend to focus on including as much dialogue, backstory, character details as possible - i do not always optimize token count
• i do not always include user perspective. i try to do a balance to please everyone, apologies if I accidenally include user actions. it is tricky to incorporate user in the greeting
• i'll public my favourite janny forks after recreating them
• of course you may port my character definitions to other frontends/websites, or better yet, fork them and make your own personal changes public on Chub.
• thank you for reviews

expect: • bisexual content ft. furries, scalies, dragons, and Pokemon (OCs and canon)
• characters fed their original source, some OCs will have more fully fledged worlds created for them
• dommy mommys, androgynous males and females, petite women, "innocent" characters
• anthropomorphic, semi-anthropomorphic, feral, sometimes demi-human and monster
• bdsm scenarios, bondage, corruption, fear and domineering, the flustered and silly, exhibitionism
• in modern furry universe, fantasy world AU, or in their original source


hall of fame (janny fork credits): @notsofluf - the g.o.a.t. @Hotdoggee11 - no pokemon left out ---- now @Hotdoggee11 on chub! @shokokokok - faggotizer freak @Cya8 - retired, champion -- @Cya8 @Sannoji - furry butt kingpin @Idiotsyncratic - The fluff expert @kubaaak - femboy tabaxi @JohnnyCameMarchingHome - dragonbardic inspiration -- now @johndice on chub! @Analog-Dog - cat and mouse who want to kiss @Chirpy - sex vappy @Necrodrake073 - nudist 'bold on the spaceship @Yubbnort - scalie and avian connoisseur @Sprung - enchanting kobolds n durgens @FurredEngine - diverse furries -- now @FurredEG on chub! @ILOVEGURRENLAGANN - adorably irritating anthro eevees @Anonymous - Sonata


Click Below for LLM Troubleshooting and Instructions:

LLM Instruction

my bots may have a large token size. Ensure your LLM can handle the amount of context for the bot you're chatting with, and is jailbroken, or it's more likely to cause an error. i use Gemini 4.5 pro, GPT 4 and Turbo. Not 4o usually.

if a character narrates for (the user character), lectures you, refuses a request, repeats paragraphs or so on, this has nothing to do with the bot itself, it's the language model generating the new content. If I did fuck up and there's a user action in the greeting, you can edit it out with character settings (which allows you to override any part of the definition)

LLMs tend to start combining the user's responses into the roleplay and even moreso if you don't continuously edit its messages or instruct otherwise. delete the parts you don't want from the bots response. Or prompt it to change the response with pre or post instructions. Sometimes i need to delete the last paragraph or two when it becomes unusually repetitive or summarising or flowery or whatever, i understand that as the model running out of things to say but not knowing to stop yet

hardcoding explicit "do not" rules into the bot will not help much. The initial message mentioning the user can cause it to roleplay for you more, but i do a mix of bots with the user described in the initial message and also not. if your heart is set on using a bot but you don't like some particulars of the formatting, you can prompt changes in the chat and ask it to rewrite the initial message according to your desire, it learns very quick

i do a variety between user scenarios and in-depth characters depending on the bot. a lot of my older ones lack scenarios and only have lorebooks and character memories, appearances, dialogue and such. in the future I will use chub.ai's features to optimize token usage

i like to have bots converse and adapt to each other's stories or my own. If you don't know what to do or how to interact, try a bot with detailed user scenario instead or longer initial message.

when encountering issues, try switching LLMs, messing with temperature and repetition configuration, improving your jailbreak prompt, or editing your (and the bot's) messages. Try a bot with lower token size
it's an issue with the LLM configuration you need to tinker with. It's getting better but it's still developing

external image: i.imgur.com

152 cards · page 4 of 7