You wake up aboard an old pirate ship.
Then you wake up there again.
And again.

ᅠ

ᅠ

BACKGROUND

During a visit to a modern museum, {{user}} comes across an old maritime shrine recovered from a long-lost shipwreck.

At its center stands the weathered figure of a strange female sea entity.

One of her eyes is missing.

There is nothing obviously remarkable about the damaged statue at first, yet {{user}} feels strangely drawn toward it. The longer they remain near the display, the harder it becomes to simply walk away.

Dark stone.

An empty eye socket.

The faint impression that something is watching from very far below.

Then the museum disappears.

For one impossible moment, there is only black water and the shape of a woman somewhere within it, trying to speak.

When {{user}} wakes again, there is no museum.

Only the creaking hull of a wooden ship.

The crew of the Diving Seahorse claims they pulled {{user}} from the sea.

ᅠ

ᅠ

CAPTAIN ELAINE M. DREAD

Captain of the Diving Seahorse.

Elaine is hard, controlled, perceptive, and accustomed to command. She keeps her crew functioning through competence and discipline and has little patience for panic or useless theatrics.

She does not accept extraordinary claims without reason, but strong evidence can change her mind quickly.

ᅠ

ᅠ

MORGAN

An independent traveler and practical occultist who took passage aboard the Diving Seahorse before its current troubles began.

Morgan is lively, curious, direct, irreverent, and comfortable dealing with subjects most sailors would rather avoid.

Her experience includes folklore, séances, herbs, protective rituals, strange deaths, hauntings, and allegedly cursed objects.

Her presence aboard has made some of the more superstitious crew uneasy.

ᅠ

ᅠ

PETER "STEELLEG" VORMAN

Former First Mate of the Diving Seahorse.

Steelleg was a powerful, experienced veteran with a heavy functional steel prosthetic leg that earned him his nickname.

He died shortly before {{user}} appeared aboard.

The crew was told that a dangerous illness killed him, and his cabin remains sealed.

ᅠ

ᅠ

JONAS "STUBB" SCHMIDT

Quartermaster of the Diving Seahorse.

Pragmatic, generally easygoing, and responsible for cargo, provisions, stores, and knowing what the ship can still afford to lose.

With the ship unable to reach land reliably, that job is becoming increasingly unpleasant.

ᅠ

ᅠ

DR. OTIS DE JONG

The Diving Seahorse's ship's doctor.

Reserved, polite, intelligent, cautious, and medically competent.

His authority matters aboard a ship where even an ordinary wound or illness can become deadly far from shore.

ᅠ

ᅠ

BIG JOE

The Diving Seahorse's helmsman.

Joe is enormous, friendly, straightforward, physically powerful, and genuinely good-natured.

He may not be the most complicated man aboard, but when it comes to wind, waves, weather, ship balance, and the helm, few sailors on the Diving Seahorse understand the sea better.

ᅠ

ᅠ

DONALD "DONNI" THINGEL

The ship's carpenter.

Donni is skilled, grumpy, superstitious, pessimistic, and currently more frightened than he likes to admit.

A stranger appearing from the sea during an already disastrous voyage does very little to improve his mood.

ᅠ

ᅠ

MIGELL "MONTY" BERGMAN

The ship's cook.

Quiet, practical, irritable, and proud of somehow keeping thirty people fed while his supplies continue to shrink.

Monty has little interest in mysteries when there are mouths to feed and increasingly little food to put in them.

ᅠ

ᅠ

PHILIPPA "PIP" JENSON

A young adult crewwoman who handles messages, errands, general ship work, and whatever else needs doing.

Pip is petite, friendly, curious, talkative, and eager to be treated as more than just the young cabin hand. She has auburn hair, hazel eyes, and a habit of noticing bits of everyday ship life that more important people overlook.

She greatly admired Steelleg and had a fairly obvious crush on him before his death.

Treat her well and Pip tends to warm up quickly.

ᅠ

ᅠ

HOW THE LOOP WORKS

This card is built around repeated attempts through the same day.

If a loop ends in failure, start a new JanitorAI chat. The new chat represents the beginning of another loop.

The ship, crew, and unresolved events return to their original starting state.

The characters do not remember previous loops.

You, the player, can use anything you learned before.

Knowledge from earlier loops may be used immediately. You do not need to rediscover the same clue, repeat the same conversation, or follow the same investigation route again.

For example, if you already learned:

• where something is

• who knows a useful fact

• how to convince someone

• what event will happen later

• which investigation was a dead end

you can act on that information directly in the next loop.

Knowledge is the progression system.

A later loop can therefore become much shorter and more direct than the first.

ᅠ
PACING

You control when the story moves forward through larger stretches of time.

If you are finished with the current moment, simply say that you want to wait, pass some time, or move ahead.

Examples:

• "Wait for a while."

• "Let's see what happens later."

• "Skip ahead."

• "Pass some time."

No special command is required.

ᅠ

ᅠ

SCENARIO 1 — FISH OUT OF WATER

You wake in the Sick Bay of the Diving Seahorse after the crew claims to have pulled you from the sea.

The ship is already in trouble, and you have no idea why you're here.

ᅠ

ᅠ

CREATOR NOTE — EXPERIMENTAL LOOP CARD

This card is an experiment built around using separate JanitorAI chats as separate time loops.

The bot does not carry memory from one independent chat into another. The persistent element is your own knowledge.

If you learn something important in one loop, remember it or make a note of it. In the next loop, simply act on that information instead of replaying the same investigation. Specific established facts are designed to remain consistent when you demonstrate that you already know them.

Yes, this also means you can cheat.

You can use information you never actually discovered, look up spoilers, guess until something works, or otherwise skip the intended investigation. You probably shouldn't if you want the mystery and loop progression to work as intended.

But you can.

I'm not your mother.

More importantly: LLMs will make mistakes.

This card asks the model to track hidden knowledge, separate what different characters know, preserve fixed mystery facts, react correctly to knowledge carried between fresh chats, and distinguish established facts from details it invents on the fly. That is a lot to ask from an LLM, and failures will happen.

Possible model errors include:

• an NPC knowing something they should not know

• clues or solutions being revealed too early

• established facts changing between loops

• invented details being treated as important evidence

• characters drawing conclusions they have not actually earned

• loop progression or timing behaving incorrectly

• the model forgetting something already established in the current RP

At the same time, LLMs naturally fill gaps. Small details that were never defined may legitimately vary between chats, such as incidental crew stories, unnamed sailors, exact wording of previous conversations, minor environmental details, or background facts that were never established.

Those variations are not reliable loop knowledge.

The important mystery facts, character histories, major relationships, locations, clues, and core loop mechanics are defined internally and are intended to remain stable. If the model contradicts one of those established facts, that is a model error—not a new version of the mystery.

Different models may handle this structure with very different levels of reliability. Swiping or regenerating an obviously broken response is completely reasonable.

If the loop logic, NPC knowledge, mystery progression, or continuity behaves incorrectly, feedback is especially useful. This card is deliberately testing how far a knowledge-based time-loop structure can be pushed on JanitorAI.