• Identify the website-level semantic entity.
• Apply consistent site-level context across subordinate pages and resources.
• Distinguish site identity from page identity and discovery metadata.
• Associate pages, sections, and resources with the correct implementation context.

Site metadata provides global semantic context. It should not be treated as a trust guarantee, enforcement layer, or compliance claim.

• Provide stable identity through id, name, and url.
• Describe the website as a whole rather than an individual page or section.
• Remain stable unless the website identity, domain, or implementation context changes.
• Support subordinate context for pages, sections, resources, and discovery documents.

Protocol and version context may be included where useful, but minimal site identity is established by id, name, and url.

• id — required; stable identifier for the site, often based on the domain or canonical site identity.
• name — required; human-readable site name or structured name declaration.
• url — required; primary URL of the site.

Recommended fields include:

• description — high-level description of the site's purpose, scope, and content.
• intent — optional statement describing the site's primary purpose under MSP-1.
• protocol — optional site-level MSP-1 protocol context. Object form is preferred for MSP-1 v1.0.1 declarations.
• version — optional version identifier for the site's MSP-1 implementation or declared protocol context.
• lastUpdated — date when the site metadata was last reviewed or updated.

Additional fields are permitted only when defined by the active site schema or an active profile.

• AI agents should treat the site object as the website-level semantic entity of an MSP-1 implementation.
• Site-level metadata may provide context for subordinate pages and resources unless more specific page-level or section-level declarations apply.
• Conflicting site metadata across a domain should be treated as a non-ideal structural condition.
• Agents should not treat site metadata as a compliance claim, trust guarantee, or enforcement layer.
• If site metadata is missing, agents may rely on domain, discovery, page-level metadata, or surrounding context, with reduced site-level clarity.

Site metadata supports consistent domain-wide interpretation without converting site identity into a broader claim of authority or trustworthiness.

• protocol — identifies the semantic protocol being used and binds the declaration to a specific MSP-1 version.
• version — may identify the site's MSP-1 implementation version or related protocol context when version identity is needed.
• id — provides stable site identity.
• name — provides the human-readable site label.
• url — identifies the primary web-addressable site location.
• description and intent — explain what the site is about and why it exists.
• page and section — identify subordinate resources within the site context.
• discovery — identifies how MSP-1 metadata is located.
• provenance, authority, and trust — may provide additional context, but are not substitutes for site identity.

Do not confuse site with page or discovery. site identifies the website-level semantic entity; page identifies an individual resource within the site; discovery identifies how MSP-1 metadata is located.

The former compliance term is sunset/deprecated compatibility only and should not be emitted in new MSP-1 v1.0.1 declarations.

{
  "site": {
    "id": "msp-1-org",
    "name": "MSP-1 — Mark Semantic Protocol",
    "url": "https://msp-1.org/"
  }
}

Site declaration with protocol context:

{
  "site": {
    "id": "msp-1-org",
    "name": "MSP-1 — Mark Semantic Protocol",
    "url": "https://msp-1.org/",
    "description": "An open, AI-first metadata protocol for machine-readable clarity.",
    "intent": "Provide a structured semantic layer that AI agents can reliably interpret.",
    "protocol": {
      "name": "MSP-1",
      "version": "1.0.1"
    },
    "version": "1.0.1"
  }
}

.well-known site declaration:

{
  "protocol": {
    "name": "MSP-1",
    "version": "1.0.1"
  },
  "site": {
    "id": "msp-1-org",
    "name": "MSP-1 — Mark Semantic Protocol",
    "url": "https://msp-1.org/"
  }
}

• It expresses site as an object.
• It provides required id, name, and url fields.
• It uses protocol and version context consistently when those fields are present.
• It avoids contradictory site metadata across the implementation.
• It does not treat site metadata as a trust, authority, compliance, or enforcement claim.