brand.md

An open standard for AI-executable brand identity.

brand.md is a structured Markdown file that gives AI tools the context they need to write, design, generate, and build on brand. It is intended to live at the root of a project, next to files like llms.txt, design.md, CLAUDE.md, or .cursorrules.

Most brand guidelines describe a brand for humans. brand.md is built for AI systems that need to act on brand guidance.

Why It Exists

AI tools can read a brand PDF, website, or portal, but they still miss the hidden decisions that make a brand coherent:

• why a color was chosen, not just its hex value
• what a voice trait means in behavior, not just as an adjective
• which tokens are allowed to pair, inherit, or override each other
• how imagery should be prompted and what it must avoid
• where brand identity ends and UI implementation begins

brand.md solves this by making brand identity portable, structured, and executable.

The Core Idea

Every meaningful brand element should carry three layers:

Layer	Purpose	Examples
Precision Layer	Exact values AI and code can verify	Hex values, token IDs, font families, motion durations, minimum logo sizes
Semantic Layer	Meaning AI can use while generating	Color descriptors, voice behavior, mood keywords, image prompts, comparable fonts
Relationship Layer	Governance and selection logic	Depends-on rules, approved pairings, channel deltas, inheritance, misuse rules

The Semantic Layer is the key difference between a brand file AI can read and one it can execute.

Quick Example

Hex-only brand guidance:

- Primary green: #184F35
- Usage: headings and CTAs

AI-executable brand guidance:

- Token: color.primary.deep-forest
- Hex: #184F35
- AI-optimized name: dark British racing green
- Image prompt: dark British racing green, deep forest tone, muted and natural
- Avoid confusing with: hunter green, olive, bright forest green
- Mood: grounded, natural, premium, quiet confidence
- Usage: primary CTAs, brand headlines, key UI emphasis
- Accessibility: use with white text only after contrast check

The first version names a value. The second version prevents AI drift.

Where brand.md Fits

File	Layer	Answers	Consumed by
llms.txt	Discovery	Where is this brand's AI-readable data?	Crawlers, agents, LLMs
brand.md	Identity	What is this brand and how does it think, speak, and look?	LLMs, image generators, content tools, coding agents
design.md	UI execution	How should this product interface be built?	Coding assistants, UI tools
skills.md	Procedures	How should specific tasks be performed?	Agents and workflow tools
CLAUDE.md / .cursorrules	Project overrides	What should this assistant know in this repo?	Coding assistants

brand.md is upstream of design.md. Brand identity defines intent and constraints. Design systems execute those constraints in components, layouts, and code.

Repository Structure

brand-md/
├── README.md
├── brand.md
├── spec/
│   └── brand-md.md
├── brands/
│   ├── OFFICIAL.md
│   └── examples/
│       ├── acme-saas/
│       │   └── brand.md
│       ├── nova-agency/
│       │   └── brand.md
│       └── fold-consumer/
│           └── brand.md
├── SEMANTIC-LAYER.md
├── COMPARISON.md
├── CONTRIBUTING.md
└── LICENSE

What This Repo Is

This repo is a standard and reference implementation for brand.md.

It includes:

• the brand.md specification
• the Semantic Layer model
• context for how brand.md relates to llms.txt, design.md, and agent instruction files
• fictional examples that demonstrate the format
• a registry model for official brand-owned files

What This Repo Is Not

This repo is not a scraped library of famous brands.

A website is one public channel. It can show a brand's surface execution, but it cannot reveal the full brand intelligence underneath: positioning, voice governance, channel rules, campaign direction, internal guardrails, or strategic rationale.

Legitimate brand.md files should come from inside the brand or from a brand system built with the brand team's input.

Generate A brand.md

Sameness generates brand.md from a structured brand graph, including precision, semantic, and relationship layers for identity, voice, color, typography, logo, motion, imagery, video, and audio.

Further reading: What is brand.md?