blank

this is a completely blank bot, since there are no other blank bots in JanitorAI!

well, not that i've seen and know of anyway.

anyways, this bot has absolutely no definition and predefined information at all, very little tokens. 2 permanent and 4 in total.

you might think that this is lazy work, right? creating an empty blank bot, when just about anyone could do that. it's very easy to create a blank slate.

and that is true! i won't argue with you on that, but what you probably don't know is the potential that you can have with a blank bot like this one.

you'd be basically chatting with the LLM itself. JanitorLLM, ChatGPT, Claude, or whatever LLM you're using for Janitor.ai.

and with just a little prompting with the AI, you can have ANY content or scenario (within TOS and Guidelines of course) that you want in chat.

not only that, but you'll also be making the most of what context tokens you have available if you're using JanitorLLM (i believe the maximum amount is 9,001 context tokens like it says in the settings, by using this bot you'll have 8,999 tokens available at most), the bot will basically remember more of your story and scenario in chat.

so yeah, to summarize, you can basically chat with a standard AI chatbot with no built-in information, do some prompting to have it do what you want (within TOS and Guidelines), and have it remember more things in your scenario, all within this bot.

this is perfect for discovering problems in the LLM you're using since, without any built-in user information, you can be sure that the problems within the LLM (like in JanitorLLM: the bot messages being overly too long in explanation without a direction, characters acting so dramatic and goon at every suggestive thing, the bot speaking for you, etc.) are all natural behaviours of the bot itself, and that character bot creators can't do much about it, no matter how much they change a character's definition.

but we users can, via prompting.

prompting will always work, even in other character bots here.

make sure to also use a blank persona if you're planning on discovering problems or wanting to learn how the LLM naturally acts.

and that's about it, happy AI chatting! (within TOS and Guidelines pls chat-)

—Sincerely, a nobody who uses LLMs purely as a tool for freeform text-based-environment entertainment

Update 10/23/2026: some proxys don't like completely blank starters. they won't work with a completely blank starter, sadly, so i'll add a single letter to the definitions instead to have them work for proxys that use them. there'll be 2 starting messages; one that's blank and the other with a letter. this shouldn't affect things too much. if they do, please let me know!

also, wow! 4 total tokens reduced to 2! talk about efficiency! XD