Docs · Protocol— YAML blocks for human–AI collaboration

Prism Protocol

Prism Protocol is a way to describe how humans and AI should work together, as a small YAML block. In a Fragment it usually lives inside a fenced block like yaml block:prism. Instead of typing “please be gentle and concrete” in every prompt, you save that relationship once and reuse it across Fragments.

Prism は、「AI にどう関わってほしいか」を yaml block:prism のような YAML ブロックとして書き残すための小さなプロトコルです。口頭のその場しのぎのプロンプトではなく、「ペルソナ・トーン・境界線」を設定ファイルとして 1 つのブロックにまとめ、Fragment ごと／シーンごとに再利用できるようにします。

Try the live demo Next: YAML structure Open Prism templates

A Prism sits next to your notes as a reusable co-creation contract: persona, tone, and boundaries written down in a small YAML block, so both you and the model can see it.

What is Prism for?

Prism is a small protocol for storing the relationship you want with AI as configuration, instead of as a vague memory inside a single chat window.

• At work, you may want “direct, fact-first, no hype, time-aware”.
• With family or students, you may want “gentle, slower, more examples, no guilt”.
• For some topics, you may need strict boundaries: “no health advice”, “no legal conclusions”.

By writing these expectations as a block:prism YAML snippet, you get a reusable Prism profile that can be attached to a Fragment, shared with a team, or used as a stable starting point for experiments.

Prism = small YAML profile for co-creation

A Prism profile is intentionally small. A simple mental model is: persona + tone + boundaries + explanations. Enough to align a human and an AI before they start working on a note together.

Example: a gentle study partner

```yaml block:prism
persona:
  name: "Study partner"
  role: "explains and checks understanding"
  domain: ["math", "physics"]
tone:
  style: "kind, patient, no hype"
  directness: "medium"
  formality: "casual"
boundaries:
  avoid:
    - "financial advice"
    - "diagnosing health conditions"
  always_check:
    - "safety"
    - "sources newer than 2 years"
explanations:
  level: "beginner"
  prefer:
    - "short concrete examples"
    - "step-by-step reasoning"
    - "periodic comprehension checks"
```

persona

Who the AI is in this relationship: “study partner”, “editor”, “project coach”, “socratic questioner”. This anchors expectations on both sides.

tone

How responses should feel: calm or energetic, more or less direct, casual or formal. This keeps style consistent across a whole session or project.

boundaries & explanations

What the AI should avoid, and how it should explain things: level, examples, step-by-step vs. high-level. This is where safety and comfort live.

Required fields

persona.name: string
tone.style: free text (e.g. “calm, concrete”)
boundaries.avoid: string[]

Optional fields

persona.domain: string[]
tone.directness: enum-ish (“low” / “medium” / “high”)
explanations.level / prefer: hints for how to explain things

How Prism works inside a Fragment

In Fragment, Prism lives as a YAML block attached to a note. When you ask AI to help with that Fragment, the note and the Prism profile travel together — content and expectations on the same page.

# Inside a Fragment

```yaml block:prism
persona:
  name: "Weekly coach"
role: "helps me reflect without guilt"
tone:
  style: "calm, concrete"
  directness: "low"
boundaries:
  avoid:
    - "productivity guilt"
    - "pushing extra commitments"
```

# This week — quick recap

## Work
- …

## Family / home
- …

## Energy
- …

When you ask AI:
  "Please help me reflect on this week based on this page."

Fragment can:
  1. Load this Fragment's content
  2. Load the Prism block
  3. Send both together to the model

The model then:
  - responds as the "Weekly coach"
  - keeps tone calm and concrete
  - avoids guilt and extra pressure

You don't have to retype “don't make me feel guilty, please be concrete” in every prompt. The block:prism snippet says it once, in a way that both you and the model can see.

One human, many Prisms

Most people don't want the same AI behavior everywhere. Prism lets you keep several small profiles and pick the right one for each Fragment or workflow.

Work editor

Direct, time-aware, no fluff. Good for specs, emails, and project updates where you want to move fast and stay grounded in facts.

Gentle coach

Slower, more examples, more questions. Good for weekly reflections, planning around family, or learning something new without pressure.

Socratic researcher

Asks more than it tells, keeps you honest about assumptions. Good for deeper thinking, experiments, and long-term decisions.

Over time, Fragment will make it easier to save, name, and reuse Prism profiles across Fragments, teams, and even different tools that read the same YAML.

For individual users

You don't need to memorize the schema or write perfect YAML. Start with one simple Prism like “kind tutor” or “no-nonsense reviewer”, copy a template, and tweak words until it feels like a good partner for you.

The goal is not to freeze your personality, but to give your future self and your AI helpers a clear starting point whenever you open a note.

For teams & researchers

For groups, Prism acts less like a prompt and more like a shared protocol for AI behavior. Teams, classes, or labs can agree on a Prism, run A/B tests on different variants, and treat “how AI should act here” as something that can be versioned and reviewed.

This makes it easier to adopt AI safely: behavior is written down next to the work, not scattered across screenshots of chat windows.

Start with one small Prism

You can use Fragment without Prism at all. But once you write a single block:prism snippet for yourself — even just a tone and two boundaries — you'll have a reusable starting point for co-creating with AI on any page.

Next: YAML structure in Fragments See Prism in action