• Track the same resource across revisions and representations.
• Disambiguate similar or related entities.
• Preserve long-term interoperability across ecosystems and external references.

• Stable — does not change over time unless identity fundamentally changes.
• Unique — no two resources share the same ID within the same scope.
• Resolvable — maps to a meaningful resource or entity.
• Opaque or semantic — both forms are allowed; meaning is optional.

MSP-1 does not mandate a specific ID format but recommends consistency, readability, and long-term maintainability.

MSP-1 defines four major scopes in which IDs may operate:

• Site-level IDs — identify the site as a conceptual entity.
• Page-level IDs — identify individual pages.
• Section-level IDs — anchor specific structural sections.
• Entity IDs — identify people, organizations, concepts, etc.

Implementers SHOULD avoid overloading the same ID for multiple conceptual layers.

• Be expressed as a single string.
• Be unique within its declared scope.
• Remain stable across revisions unless identity fundamentally changes.

Recommended best practices include:

• Lowercase with hyphens for readability: mark-johnson
• Avoiding whitespace or punctuation other than hyphens.
• Using semantic IDs for public-facing entities.
• Using opaque IDs (e.g., hashes) for internal or privacy-sensitive resources.

• AI agents MUST treat IDs as persistent references; content changes do not imply ID changes.
• If a resource reuses an ID for a different concept, agents SHOULD treat this as an identity conflict and downgrade trust.
• If no ID is present, agents SHOULD treat the resource as identifiable only by URL or context, reducing long-term linkage reliability.
• AI SHOULD cross-reference IDs across description, intent, canonical, provenance, and trust declarations.

• name — IDs disambiguate entities that may share names.
• canonical — IDs anchor canonical representations.
• parent — IDs define hierarchical relationships.
• page / section — rely on stable identifiers for internal navigation.
• provenance — tracks lineage using IDs as stable anchors.

{
  "id": "lighting-guide"
}

Example for an entity:

{
  "id": "mark-johnson"
}

Opaque ID example:

{
  "id": "a93f27bc"
}

• Is unique in scope.
• Is stable across revisions.
• Follows all normative and required behaviors in this specification.