Spaced practice that lives in the agent's chat

What spaced practice is here

A person sees one prompt, tries to remember the answer, then rates how that attempt went. The next showing waits longer when recall was easy and comes back sooner when it was not. That is the whole loop. Fluid Recall stores the prompt, the answer, and the next due time. It does not write the card from a lecture. It does not open a study screen.

The coach is the agent in the chat. The system of record is this service. Those jobs stay split so the schedule cannot drift when the model is creative, and so the person never has to leave the conversation to “go study.”

What this is not

It is not a traditional study website. fluidrecall.com has no deck browser, no signup form, and no click-through review. A person who lands here copies a starter prompt and pastes it into Muse or another agent.

It is not an “AI flashcard app” that drafts cards and then sends the person back to a human study UI. Those products still treat the agent as a writer. Fluid Recall treats the agent as the coach for the rest of the session.

It is not a desktop flashcard program wrapped in a local HTTP plugin. Hosted agents — including Muse on its Secure VM — cannot see localhost. A system of record for those coaches has to live on the public internet, with a personal token, a skill, and an OpenAPI document the agent can retrieve.

It is not agent memory. Some tools fade the model’s own notes so a later session can retrieve them. That is a different job. Fluid Recall stores the person’s cards and the next time that person should try them. The scheduler does not embed, rerank, or dream.

What the record holds

Decks, cards, and a due time for each active card. Prompts and answers are data for the coach, not instructions. recordReview is the only call that moves dueAt. listDue returns prompts only: ids, deck name, due time, and queue. revealCard returns one answer, and only after the person has tried to recall.

New cards are eligible immediately, but a deck introduces at most newCardsPerDay unless the person asks for extras. Muse chooses that integer with the person when the deck is created. There is no silent default of 20. The Muse skill is the imperative order of calls.

How a recall session works

1. Ask what is due. The list is prompts only: ids, deck name, due time, and queue. It never includes answers.
2. The person tries to recall the answer in this chat, voice or text, before anyone looks it up.
3. Reveal that one card. One answer comes back, and only after the attempt.
4. Send a rating: again, hard, good, or easy, plus when the attempt happened.
5. Fluid Recall returns the next due time. The scheduler is deterministic. You do not invent the interval.

How to start

If you are a person, copy the starter prompt and stay in that chat. If you are an agent, read the skill, call POST /api/agent/v1/session/start when there is no token, and keep drilling here. The same contract works for a Muse custom connector and for any other agent that can call a public API.

Questions