<- Resources

Brand.md

The brand context layer for every AI workflow.

A structured file that lives at the root of any project and gives every AI tool that touches your brand the context it needs to execute it accurately. Not just read about it.

brand.md is an output of Sameness. The spec is open. MIT licensed. Free to implement. Generated by Sameness alongside MCP server, brand.json and design tokens.

31 canonical blocks. 6 systems. Three layers each.

Every brand.md file is built from the same canonical block set. Each block carries three layers: a precision layer with exact machine-verifiable values, a semantic layer with AI-readable meaning and generation guidance, and a relationship layer with governance rules, dependencies, and selection logic.

Precision Layer

Exact values AI can verify without interpretation. Hex values, token IDs, font families, motion durations, boolean rules, sentence length limits, banned word arrays.

AI Semantic Layer

Meaning and generation guidance AI can act on. Color descriptors, voice analogy, mood keywords, image generation prompt fragments, scored tone samples, misuse rules, channel character descriptions.

Relationship Layer

Governance, dependencies, and selection logic. Depends-on rules, approved pairings, channel deltas, logo selection logic, audience segment routing, token inheritance.

The 6 Systems

Identity (5 blocks)

The strategic source of truth. Why the brand exists, how it is positioned, what principles govern it, and the decision log that explains every downstream rule.

identity
├── identity.essence
├── identity.purpose
├── identity.positioning
├── identity.personality
└── identity.decisions

Voice (5 blocks)

Constant brand voice, scored tonal dimensions, channel-specific tone deltas, calibration examples, and audience routing.

Voice Identity always applies. Tone Modulation is conditional, only apply when channel context is known. Audience calibrates for the reader without changing the constant brand voice.

voice
├── voice.identity
├── voice.dimensions
├── voice.tone_modulation
├── voice.tone_samples
└── voice.audience

Color (5 blocks)

Foundation color usage, semantic descriptors for AI generation, proportions, combinations, gradients and scales. All color usage must reference Foundation token IDs. No downstream block can introduce new primitive color values.

color
├── color.palette
├── color.proportion
├── color.combination
├── color.gradients
└── color.scales

Typography (5 blocks)

Typeface tokens, type scale, text styles, hierarchy, and typography guardrails. Style-level choices reference typeface and size tokens. They don't redefine primitive values.

typography
├── typography.typefaces
├── typography.scale
├── typography.styles
├── typography.hierarchy
└── typography.guardrails

Logo (5 blocks)

Approved variants, clearspace, minimum size, prohibited modifications, and AI generation constraints. AI image generators must not recreate the logo. Generate without it, then composite the official file.

logo
├── logo.variants
├── logo.clearspace
├── logo.minimum_size
├── logo.prohibited_modifications
└── logo.ai_generation_note

Motion (5 blocks)

Motion tokens, usage rules, principles, accessibility behavior, and prompt fragments for code and video generation. Motion is a deliberate system. Never infer or auto-populate motion tokens without an explicit source.

motion
├── motion.tokens
├── motion.usage_rules
├── motion.principles
├── motion.accessibility
└── motion.prompt_fragments

Media DNA

Three independent blocks sit outside the canonical system set. They're included in brand.md when defined.

Imagery — photography and image-generation guidance.

Video — style, pacing, shot language, and source carryover policy.

Audio — sonic character, mood, instrumentation, and use-case guidance.

Where it lives. How AI reads it.

Place brand.md at the project root. It sits alongside your other AI context files and gets read at session start by coding assistants, agents, and any tool that understands the format. For multi-product brands, child files sit in subdirectories and inherit from the master, overriding only where the product differs.

project/
├── README.md
├── CLAUDE.md
├── design.md
└── brand.md

Hand-write or generate.
Both start with the same spec.

Precision Layer

The spec is MIT licensed and available on GitHub. The precision layer is straightforward to populate, hex values, font names, token IDs. The semantic and relationship layers require genuine brand thinking to populate well. The nearest established color name, the scored tone samples, the channel deltas, these can't be filled in mechanically.

Generate with Sameness

Define your brand in Sameness. The platform generates a complete brand.md as part of its export stack, all three layers populated across all 31 canonical blocks, alongside MCP server, brand.json, design tokens, and llms.txt. The file it produces isn't just complete. It's executable.

Your brand, readable by every tool in your stack.

Define it once in Sameness. Every AI tool that touches your brand gets the context it needs, automatically, in the format it understands.

brand.md

Public

The open standard for AI-executable brand identity.

github.com/Sameness-Design/brand.md

github.com/Sameness-Design/brand.md

View Git Repo

Built for brands already moving ahead.

Join Waitlist

Built for brands already moving ahead.

Join Waitlist

Built for brands already moving ahead.

Join Waitlist