Formats

Every format your AI tools need

Brand DNA in structures agents load, not loose PDFs. Each export is diffable, scoped, and built to drop into repos, IDEs, and model system prompts.

CLAUDE.md

Repo-local instructions for Claude and other LLM agents: voice, visual constraints, and do/don't rules shaped from your mythos.

Example

1## Voice

2Clear, confident; evidence over hype.

4## Visual

5Primary #9C4221 - hero CTAs only.

6Display: Instrument Serif; body: DM Sans.

8## Rules

9DO: short paragraphs, numbered steps.

10DON'T: ALL CAPS emphasis or vague claims.

Generate yours →

.cursorrules

Cursor-native constraints: naming, tokens, and voice. Completions and refactors stay on-brand without opening a separate doc.

Example

1# Cursor - Acme

2voice: see CLAUDE.md

3components: PascalCase in src/components/

4colors: var(--color-*) only, never raw hex

5spacing: --space-2 | --space-4 | --space-8

6a11y: landmark tags, focus rings on interactives

Generate yours →

AGENTS.md

Governance for autonomous agents: scope, safety rails, and brand voice rules. Reviewable in git like any other policy file.

Example

1## Scope

2Agents may: suggest tests, document APIs.

3Agents must NOT: change secrets or CI config.

5## Brand

6All user-facing copy follows CLAUDE.md voice.

8## Escalation

9Stop if token or voice rules conflict with a ticket.

Generate yours →

Design tokens (CSS)

Custom properties your design system and codegen share. Color roles, type stacks, and spacing so AI-written UI doesn't invent one-off values.

Example

1:root {

2 --color-primary: #9c4221;

3 --color-accent: #d97706;

4 --color-surface: #fffbf5;

5 --font-display: "Instrument Serif", serif;

6 --font-body: "DM Sans", sans-serif;

7 --radius: 4px;

8}

Generate yours →

System prompts

Portable instructions for ChatGPT, Copilot, Gemini, and other chat surfaces. Same mythos, tuned for conversational length limits.

Example

1You write as Acme: direct, warm, never breathless.

2Use #9C4221 only for primary CTAs.

3Before shipping copy, check claims against approved proof points.

4If unsure, ask for a human review - do not guess.

Generate yours →

Knowledge graph (JSON-LD)

Structured entities and relationships. Queryable facts for RAG, agents, and internal tools without re-parsing prose each time.

Example

1{

2 "@type": "Brand",

3 "name": "Acme",

4 "slogan": "Ship with confidence.",

5 "knowsAbout": ["B2B SaaS", "compliance"],

6 "sameAs": ["https://acme.com"]

7}

Generate yours →

HTML brand guide

A hosted, semantic HTML reference: logo usage, palette roles, type, and voice. Share one URL instead of emailing versioned PDFs.

Example

1<article class="bm-voice">

2 <h1>Voice</h1>

3 <p>Direct, proof-first. No filler.</p>

4</article>

5<section id="colors">

6 <h2>Primary</h2>

7 <p>#9C4221 - CTAs and key actions only.</p>

8</section>

Generate yours →

Export your mythos in every format

Ingest once; generate CLAUDE.md, Cursor rules, tokens, prompts, graphs, and the HTML guide from the same structured source.

Try your brand free View pricing

Read the how it works guide or the blog for format deep dives.