Provenance supports origin and lineage clarity. It informs interpretation but does not by itself establish authority, verification, correctness, trust, ranking value, or answer accuracy.

The canonical MSP-1 object form requires type. The type field may be a single provenance category or an array of unique provenance categories.

• type — declares the provenance category or categories for the resource.

Allowed provenance categories include:

• original — content created directly by the declared author or originating party.
• derived — content adapted from an identifiable external source.
• aggregated — content assembled or synthesized from multiple sources.
• ai-assisted — content created or shaped with AI assistance.
• ai-generated — content produced primarily by an AI system.

• source — original source URL or identifier if the resource derives from an external origin.
• contributors — entities that materially contributed to the creation, development, transformation, or review context of the resource.
• confidence — optional self-assessed confidence in the accuracy of the provenance declaration.
• timestamp — date and time when the provenance declaration was created or updated.
• notes — free-form provenance notes, caveats, or clarifications.

• The provenance declaration should accurately reflect the known origin and creation context of the resource.
• The declaration should remain stable unless the resource origin, creation method, or lineage meaningfully changes.
• Derived or aggregated content should identify source material when available.
• Multiple provenance categories may be used when each category is meaningful and non-duplicative.
• Provenance should not be treated as proof of authority, correctness, trust, verification, or ranking value by itself.
• If provenance is missing, origin context may be unspecified, but absence of provenance should not alone invalidate an otherwise interpretable MSP-1 declaration.

{
  "provenance": {
    "type": "original"
  }
}

Derived provenance:

{
  "provenance": {
    "type": "derived",
    "source": "https://example.com/original-article",
    "notes": "Adapted with permission."
  }
}

AI-assisted provenance:

{
  "provenance": {
    "type": [
      "original",
      "ai-assisted"
    ],
    "contributors": [
      {
        "id": "mark-johnson",
        "role": "author"
      },
      {
        "id": "chatgpt-jdk",
        "role": "ai-partner"
      }
    ],
    "confidence": "high",
    "notes": "Drafted with human guidance and AI structural assistance.",
    "timestamp": "2025-12-11T10:00:00Z"
  }
}

• author identifies creation or origination responsibility; provenance describes broader origin and lineage.
• reviewer identifies review responsibility and should not be conflated with provenance.
• revision describes lifecycle change events that may affect provenance context.
• authority concerns declared decision-making or stewardship context and should not be inferred from provenance alone.
• trust concerns separate confidence or supporting signals and should not be conflated with provenance.
• canonical identifies the preferred URL representation when multiple representations exist.

When present, provenance should be an object with type. The type value should be either a supported provenance category string or an array of unique supported provenance category strings.

Validators should flag unsupported provenance categories, duplicate category values, missing source context for clearly derived declarations, or contradictions with author, reviewer, or revision metadata as non-ideal clarity conditions.