standard v0.1

examples/caplifi/BRAND.md BRAND.md · about 500 tokens

# Caplifi BRAND.md v0.2.0
> Caplifi designs and builds community-owned automation, from local AI to verified infrastructure.

## Identity
Name: Caplifi. Legal: Caplifi Technologies. Mission: Building sustainable civilizations for the public good. Founded: 2020. URL: caplifi.com. Contact: matthew@caplifi.com

## Color
| Name | Hex | Role |
| --- | --- | --- |
| Primary blue | #0B3A73 | Structure, links, wordmark |
| Ice blue | #D8EBF9 | Panels, backgrounds |
| Brass gold | #B6872A | Accents, sparingly |
| Signal green | #54A936 | Success, pass states |
| Ink navy | #121724 | Text, dark surfaces |
| Light gray | #E8E9EA | Page ground |

## Type
Display: Michroma, Orbitron, sans-serif. Body: IBM Plex Sans, system-ui. Mono: IBM Plex Mono.

## Voice
Attributes: direct, precise, high-trust, practical, accountable, verifiable. Banned: em dashes, delve, leverage, robust.

## Logo
brand/assets/caplifi-monogram.svg (mark), brand/assets/caplifi-wordmark.svg (primary). Dark: caplifi-lockup-dark.svg. Light: caplifi-lockup-light.svg.

## Load deeper
voice.md · visual.md · strategy.md · usage.md

## Integrity
brand/assets.manifest.json · spec v0.1 · github.com/caplifitechnologies/brandmd

the PDF it replaced

~38,000 tokens

the BRAND.md core

about 500 tokens

Same brand. The worked example currently measures about 500 tokens, read in full on every task. Detail loads only when a task needs it.

Your brand, readable by any AI.

An open standard for portable, token-efficient brand identity. One markdown file. Works everywhere.

Build yours Read the spec

1.1How it works

Write

Curate your brand into one core file and a folder of deeper files. The Builder finds the gaps and helps you fill them.

Export

Download a clean repo: BRAND.md, the brand files, DTCG tokens, an asset manifest with hashes, and a changelog.

Point any AI at it

Paste it, attach it, or point a tool at the repo. Any AI on any platform reads the same brand the same way.

1.2The loading protocol

An AI agent encountering a BRAND.md file MUST read it in full. It SHOULD load files from /brand/ only when the task touches that domain: voice.md for writing tasks, visual.md for design tasks, strategy.md for positioning work, usage.md before publishing. Before embedding any asset file, an agent SHOULD verify the file's SHA-256 hash against assets.manifest.json. If a hash does not match, the agent MUST NOT use the asset and SHOULD report the mismatch.

1.3Who it's for

Founders

Keep every AI on brand without pasting a 60-page guide into each session.

Agencies

Hand a client a portable brand repo that works in whatever tool the client already uses.

AI builders

Read a brand in under 600 tokens and load detail on demand. One format across every client. Site map for agents: llms.txt.