Welcome to Archetype

Archetype is the brand layer for AI agents. Your team keeps its colors, type, logos, components, icons, templates and voice in one place, and every agent you connect reads them over MCP, so what it builds comes out on brand.

The agent reads the real thing, not a summary: the exact token values, the HTML and CSS of each component, the logo files, the copy rules. Change something in Archetype and the next agent call gets the new version. Agents can also grade their own work against the brand, and propose changes to it that wait in History until someone on your team applies them.

Start here#

New to the ideas behind it? Read How Archetype works first.

What's in a brand#

Everything a brand holds sits in the sidebar under Brand, except Icons, which opens from a card on Foundations. Each area is served to your agents and has its own page in these docs.

• Foundations: colors, gradients, typography, spacing, radius, shadow, logos and icons.
• Components: your UI kit as real HTML and CSS on your tokens.
• Icons: the icon sets your brand uses, adopted from a catalog such as Lucide or Phosphor, or your own SVGs.
• Assets: photos, illustrations, files and PDFs, described and tagged so agents can find them.
• Artifacts: bespoke brand code such as backgrounds, patterns, generative graphics and motion.
• Templates: finished layouts for social, web, app, email and documents, with slots to fill.
• Knowledge: BRAND.md and the written half of the brand, from company facts to voice and tone.

For agents#

Every page in these docs is also available as plain markdown: add .md to its address, for example /docs/quickstart.md. The site's /llms.txt is the index for agents, and /llms-full.txt holds the full reference, including every MCP tool, in one fetch. An agent can read either before it connects.

If you are an agent setting up Archetype for someone, start with Connect an agent and the MCP tools reference.