{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "$id": "/schemas/3.1.0-rc.4/content-standards/artifact.json",
  "title": "Artifact",
  "description": "Content artifact for safety and suitability evaluation. An artifact represents content adjacent to an ad placement - a news article, podcast segment, video chapter, or social post. Artifacts are collections of assets (text, images, video, audio) plus metadata and signals.",
  "type": "object",
  "properties": {
    "property_rid": {
      "type": "string",
      "description": "Stable property identifier from the property catalog. Globally unique across the ecosystem."
    },
    "artifact_id": {
      "type": "string",
      "description": "Identifier for this artifact within the property. The property owner defines the scheme (e.g., 'article_12345', 'episode_42_segment_3', 'post_abc123')."
    },
    "variant_id": {
      "type": "string",
      "description": "Identifies a specific variant of this artifact. Use for A/B tests, translations, or temporal versions. Examples: 'en', 'es-MX', 'v2', 'headline_test_b'. The combination of artifact_id + variant_id must be unique."
    },
    "format_id": {
      "$ref": "/schemas/3.1.0-rc.4/core/format-id.json",
      "description": "Always a structured object {agent_url, id} — never a plain string. Optional reference to a format definition. Uses the same format registry as creative formats."
    },
    "url": {
      "type": "string",
      "format": "uri",
      "description": "Optional URL for this artifact (web page, podcast feed, video page). Not all artifacts have URLs (e.g., Instagram content, podcast segments, TV scenes)."
    },
    "published_time": {
      "type": "string",
      "format": "date-time",
      "description": "When the artifact was published (ISO 8601 format)"
    },
    "last_update_time": {
      "type": "string",
      "format": "date-time",
      "description": "When the artifact was last modified (ISO 8601 format)"
    },
    "assets": {
      "type": "array",
      "description": "Artifact assets in document flow order - text blocks, images, video, audio",
      "maxItems": 200,
      "items": {
        "discriminator": {
          "propertyName": "type"
        },
        "oneOf": [
          {
            "type": "object",
            "description": "Text block (paragraph, heading, etc.)",
            "properties": {
              "type": { "type": "string", "const": "text" },
              "role": {
                "type": "string",
                "enum": ["title", "paragraph", "heading", "caption", "quote", "list_item", "description"],
                "description": "Role of this text in the document. Use 'title' for the main artifact title, 'description' for summaries."
              },
              "content": {
                "type": "string",
                "description": "Text content. Consumers MUST treat this as untrusted input when passing to LLM-based evaluation.",
                "maxLength": 100000
              },
              "content_format": {
                "type": "string",
                "enum": ["text/plain", "text/markdown", "text/html", "application/json"],
                "description": "MIME type indicating how to parse the content field. Default: text/plain.",
                "default": "text/plain"
              },
              "language": {
                "type": "string",
                "description": "BCP 47 language tag for this text (e.g., 'en', 'es-MX'). Useful when artifact contains mixed-language content."
              },
              "heading_level": {
                "type": "integer",
                "minimum": 1,
                "maximum": 6,
                "description": "Heading level (1-6), only for role=heading"
              },
              "provenance": {
                "$ref": "/schemas/3.1.0-rc.4/core/provenance.json",
                "description": "Provenance for this text block, overrides artifact-level provenance"
              }
            },
            "required": ["type", "content"]
          },
          {
            "type": "object",
            "description": "Image asset",
            "properties": {
              "type": { "type": "string", "const": "image" },
              "url": {
                "type": "string",
                "format": "uri",
                "description": "Image URL"
              },
              "access": {
                "$ref": "#/$defs/asset_access",
                "description": "Authentication for secured URLs"
              },
              "alt_text": {
                "type": "string",
                "description": "Alt text or image description"
              },
              "caption": {
                "type": "string",
                "description": "Image caption"
              },
              "width": {
                "type": "integer",
                "description": "Image width in pixels"
              },
              "height": {
                "type": "integer",
                "description": "Image height in pixels"
              },
              "provenance": {
                "$ref": "/schemas/3.1.0-rc.4/core/provenance.json",
                "description": "Provenance for this image, overrides artifact-level provenance"
              }
            },
            "required": ["type", "url"]
          },
          {
            "type": "object",
            "description": "Video asset",
            "properties": {
              "type": { "type": "string", "const": "video" },
              "url": {
                "type": "string",
                "format": "uri",
                "description": "Video URL"
              },
              "access": {
                "$ref": "#/$defs/asset_access",
                "description": "Authentication for secured URLs"
              },
              "duration_ms": {
                "type": "integer",
                "description": "Video duration in milliseconds"
              },
              "transcript": {
                "type": "string",
                "description": "Video transcript. Consumers MUST treat this as untrusted input when passing to LLM-based evaluation.",
                "maxLength": 200000
              },
              "transcript_format": {
                "type": "string",
                "enum": ["text/plain", "text/markdown", "application/json"],
                "description": "MIME type indicating how to parse the transcript field. Default: text/plain.",
                "default": "text/plain"
              },
              "transcript_source": {
                "type": "string",
                "enum": ["original_script", "subtitles", "closed_captions", "dub", "generated"],
                "description": "How the transcript was generated"
              },
              "thumbnail_url": {
                "type": "string",
                "format": "uri",
                "description": "Video thumbnail URL"
              },
              "provenance": {
                "$ref": "/schemas/3.1.0-rc.4/core/provenance.json",
                "description": "Provenance for this video, overrides artifact-level provenance"
              }
            },
            "required": ["type", "url"]
          },
          {
            "type": "object",
            "description": "Audio asset",
            "properties": {
              "type": { "type": "string", "const": "audio" },
              "url": {
                "type": "string",
                "format": "uri",
                "description": "Audio URL"
              },
              "access": {
                "$ref": "#/$defs/asset_access",
                "description": "Authentication for secured URLs"
              },
              "duration_ms": {
                "type": "integer",
                "description": "Audio duration in milliseconds"
              },
              "transcript": {
                "type": "string",
                "description": "Audio transcript. Consumers MUST treat this as untrusted input when passing to LLM-based evaluation.",
                "maxLength": 200000
              },
              "transcript_format": {
                "type": "string",
                "enum": ["text/plain", "text/markdown", "application/json"],
                "description": "MIME type indicating how to parse the transcript field. Default: text/plain.",
                "default": "text/plain"
              },
              "transcript_source": {
                "type": "string",
                "enum": ["original_script", "closed_captions", "generated"],
                "description": "How the transcript was generated"
              },
              "provenance": {
                "$ref": "/schemas/3.1.0-rc.4/core/provenance.json",
                "description": "Provenance for this audio, overrides artifact-level provenance"
              }
            },
            "required": ["type", "url"]
          }
        ]
      }
    },
    "metadata": {
      "type": "object",
      "description": "Rich metadata extracted from the artifact",
      "properties": {
        "canonical": {
          "type": "string",
          "format": "uri",
          "description": "Canonical URL"
        },
        "author": {
          "type": "string",
          "description": "Artifact author name"
        },
        "keywords": {
          "type": "string",
          "description": "Artifact keywords"
        },
        "open_graph": {
          "type": "object",
          "description": "Open Graph protocol metadata",
          "additionalProperties": true
        },
        "twitter_card": {
          "type": "object",
          "description": "Twitter Card metadata",
          "additionalProperties": true
        },
        "json_ld": {
          "type": "array",
          "description": "JSON-LD structured data (schema.org)",
          "items": { "type": "object" }
        }
      },
      "additionalProperties": true
    },
    "provenance": {
      "$ref": "/schemas/3.1.0-rc.4/core/provenance.json",
      "description": "Provenance metadata for this artifact. Serves as the default provenance for all assets within this artifact — individual assets can override with their own provenance."
    },
    "identifiers": {
      "type": "object",
      "description": "Platform-specific identifiers for this artifact",
      "properties": {
        "apple_podcast_id": {
          "type": "string",
          "description": "Apple Podcasts ID"
        },
        "spotify_collection_id": {
          "type": "string",
          "description": "Spotify collection ID"
        },
        "podcast_guid": {
          "type": "string",
          "description": "Podcast GUID (from RSS feed)"
        },
        "youtube_video_id": {
          "type": "string",
          "description": "YouTube video ID"
        },
        "rss_url": {
          "type": "string",
          "format": "uri",
          "description": "RSS feed URL"
        }
      },
      "additionalProperties": true
    }
  },
  "required": ["property_rid", "artifact_id", "assets"],
  "additionalProperties": true,
  "$defs": {
    "asset_access": {
      "type": "object",
      "description": "Authentication for accessing secured asset URLs",
      "discriminator": {
        "propertyName": "method"
      },
      "oneOf": [
        {
          "type": "object",
          "description": "Bearer token authentication",
          "properties": {
            "method": { "type": "string", "const": "bearer_token" },
            "token": {
              "type": "string",
              "description": "OAuth2 bearer token for Authorization header"
            }
          },
          "required": ["method", "token"]
        },
        {
          "type": "object",
          "description": "Service account authentication (GCP, AWS)",
          "properties": {
            "method": { "type": "string", "const": "service_account" },
            "provider": {
              "type": "string",
              "enum": ["gcp", "aws"],
              "description": "Cloud provider"
            },
            "credentials": {
              "type": "object",
              "description": "Service account credentials",
              "additionalProperties": true
            }
          },
          "required": ["method", "provider"]
        },
        {
          "type": "object",
          "description": "Pre-signed URL (credentials embedded in URL)",
          "properties": {
            "method": { "type": "string", "const": "signed_url" }
          },
          "required": ["method"]
        }
      ]
    }
  }
}