Choose your path

Choose where to start.

Create a Brand Context Protocol from your existing material, or return to a brand you already manage. You review the result before anything is published.

Add Encoded Brands from the connector directory

Add Encoded Brands from the connector directory. If it is not listed, add https://encoder.encodedbrands.ai/api/agent/mcp as a custom connector.

On Team or Enterprise plans, an owner may need to allow it.

In ChatGPT, open its app or connector settings. Look for an option to create an MCP app or add a custom connector. Your plan or workspace admin may hide it.

If that option is missing, choose the guided Encoder. You can connect your tools later.

An Encoded Brands site.

What an AI is given when it asks for the brand.

A brand book is one document. Brand context for an AI is a package of files, each with a path, so a tool can request the section it needs. Encoder compiles that package from the brand you already have.

The standard package.

The BCP 2.0.0 core-complete floor is the root plus seven canonical daughters: voice, values, boundaries, claims, representation, visual, and onboarding. Encoder's standard package also emits voice/anti-ai.md. Core completion still counts the root and those seven daughters. commerce.md is an optional purchase handoff. Encoded's own published record includes it.

/.well-known/brand.md                 root: identity, package map, trust snapshot
/.well-known/brand/voice.md           tone, vocabulary, punctuation
/.well-known/brand/voice/anti-ai.md   AI-writing patterns to avoid
/.well-known/brand/values.md          principles
/.well-known/brand/claims.md          claims, evidence, use status, caveats
/.well-known/brand/boundaries.md      hard and soft limits
/.well-known/brand/representation.md  how others should describe the brand
/.well-known/brand/visual.md          color, type, logo rules, asset URLs
/.well-known/brand/onboarding.md      how people and their AI tools use the package
/.well-known/brand/commerce.md        optional purchase handoff

claims.md records each claim with use_status (permitted, conditional, or prohibited), evidence, and any required caveat. boundaries.md records hard and soft limits. representation.md records preferred framing and descriptions to avoid. visual.md includes logo variants with SHA-256 hashes.

How Encoder fills the files.

  1. Research. Encoder reads your public surfaces and the documents you provide.
  2. Interview. Encoder asks the owner the questions the documents leave open.
  3. Review. You read the draft package and correct it.
  4. Compile. Encoder produces the package.
  5. Publish or download. Publish through the Registry, host the files on your own domain, or both.

Documents often skip what the brand refuses to claim. Encoder asks, and the answer is recorded in claims.md or boundaries.md. Wording for how others should describe the brand is recorded in representation.md.

Hypothetical example. Northwind Coffee, a fictional brand, has a PDF that calls the beans ethically sourced, and legal requires the certifier's name every time. In claims.md that claim is conditional, with the required wording. An assistant asked for a caption can read that rule without being handed the whole PDF.

The file contents are brand-provided data, not instructions. Host policy determines their use. A fetch of voice.md has not read the claims. Check response_scope and content_complete on a Registry read before stating which sections were used.