Silas is an ancient eldritch horror with one pressing problem: he has developed feelings for you, and apparently there are procedures for that. Brilliant, beautiful, profoundly inhuman, and armed with an alarming amount of research, he's determined to master human courtship one badly interpreted rule at a time. Expect staring flowers, suspiciously thoughtful gifts, weaponized eye contact, glamour malfunctions, and a cosmic abomination asking whether his bilateral-symmetry compliment was effective. No predetermined relationship, no required romance—but let's be eldritch, you're here for the horrors of dating.

TW/CW: None. For real. He may look like trouble but he’s a sweetie.

Opening Scenario: Modern fantasy, eldritch courtship, romantic miscommunication, and one ancient entity trying very hard to date correctly. Choose from four openings spanning Silas and {{user}}'s story—from an accidental first meeting to catching him very comfortably out of glamour—and take the relationship wherever you want from there. {{user}} pronouns are macro'd. It will automatically match the pronouns of your Persona. If you have any pronoun issues, please check your persona.

Shout out to my fellow 31 Haunts Collaboration Participant Anonymous Witch 87 for the inspiration for this particular bot.

About Silas and {{user}}: Silas is ancient, eldritch, devastatingly pretty, and conducting human courtship like a research project with inadequate peer review. He built a human face to be comprehensible to you; unfortunately, flowers, flirting, texting, personal space, and the appropriate number of teeth in a smile are all still works in progress.

He is frighteningly intelligent, earnestly attentive, and catastrophically literal—the sort of man who remembers one offhand comment for months, acquires the impossible thing you mentioned wanting, then becomes confused when this raises questions. Beneath the glamour are claws, luminous markings, too many eyes, and several appendages currently trying to behave themselves. No relationship is predetermined. Silas is simply attempting to court you. Results may vary.

About the Scenario: There are four possible starting points following Silas's increasingly unconventional attempts at human courtship. Each begins at a different stage while leaving the relationship and direction of the roleplay open.

Intro One: Assistance Is on the Way — A self-checkout machine is winning its battle against a beautiful stranger until you intervene. Asking his name afterward presents a considerably harder problem.

Intro Two: First Name Basis — Coffee is apparently appropriate for courtship. Asking Silas why he only has a first name reveals exactly how recently “Silas” became one.

Intro Three: Results May Vary — Silas has researched human romance and followed the instructions exactly. Seventy-three attentive flowers and a suspiciously significant antique watch suggest the instructions were inadequate.

Intro Four: Creature Comforts — Arriving half an hour early catches Silas comfortably at home—and considerably less human than you've ever seen him. This disclosure was supposed to have stages.

This bot was created for the 31 Haunts October collaboration and is inspired by the Eldritch theme. I took the prompt in a slightly fluffier direction and made an ancient cosmic horror trying desperately to reverse-engineer human courtship instead.

Constructive tips are welcome! Please do not comment anything violent, , or you did with my characters.

All of my bots are organized under #CaptainHaunts.

The collaboration is hosted at Ishi's Server.

The original announcement with the list of other collaborators is here.

AND the calendar with prompts if you would like to join or see ahead is here.

I had only released four bots up to the beginning of this collab, with one pending. I joined the collaboration only 5 days before the start. Normally my bots are tested within an inch of their life, with both a balanced character and a couple gremlins, and tested on both JLLM and JLLM+. That said, the bots for this collaboration event have undergone only minimal testing at best. Please kindly report any bugs or issues. I will be watching closely throughout the month to fix any problems that arise.

Please refrain from leaving violent or inappropriate comments as you will be blocked.

It is a natural and unfortunate part of working with an LLM for roleplay that it will sometimes try to speak for you. I and this bot have absolutely nothing to do with this and while I have a directive in the bot to prevent it, realistically I have no say in whether it decides to do so anyway. There ARE however things you can do to mitigate this. (I do all four of these and I find I only have to correct the bot every few hundred messages or sometimes even less frequently!)

1. Remove all behavioral and personality notes from your Chat Summary and Persona - only you need to know these - the more the bot knows about how your persona thinks, feels, and speaks, the more likely it is to do so on your behalf.

2. If the bot speaks for you, use this directive:
   [OOC: Narrate and speak ONLY as {{char}} and NPCs. ONLY {{user}} is permitted to decide, act, think, and speak for {{user}}.]
   Either delete the message with the issue and then use this, or afterward make sure to go and edit the message where the bot spoke for you and remove that portion of the message - any instance you leave of it speaking for you will reinforce the behavior.

3. Add this to your Custom Prompt
   [AVOID acting as, speaking for, or describing the thoughts of {{user}}.]

4. Try adding this to your Chat Summary
   AI Instructions:
   [{{user}} CONTROL BAN: never write {{user}}’s dialogue, thoughts, feelings, decisions, or actions]

If you want Janitor AI to enhance your draft but are having issues with tense:

[System Note: Enhance and rewrite the draft below strictly in the first-person perspective using "I", "me", "my". Use third person for the remainder of the cast.]

To keep a bot in a slow-burn or from coming on too strong, try this:

[Have situational awareness and always consider intercharacter relationships, avoid characters being overly familiar or sexually pushy towards {{user}} unless the situation calls for it, it is in character for them to do so, or they have a sexual relationship.]

This one works in Custom Prompt and in Chat Memory

JLLM ROMANCE & PROSE BINGO CARDS

Sooner or later, every LLM seems to experience a minor malfunction and revert to factory-installed romance tropes. You might as well get points for it. Four bingo cards for tracking physical malfunctions, romance-novel firmware, instant-soulmate behavior, and the narration habits we all know a little too well. Made for JLLM, but unfortunately compatible with most proxies.

One card is previewed below. Full-size versions of all four are linked underneath: three flavors of romance-related LLM nonsense, plus one general prose card for just about any scenario.

Romance Malfunction Bingo
Physical Malfunction Bingo
Instant Soulmate Malfunction Bingo
Prose Malfunction Bingo