{
  "$schema": "http://json-schema.org/draft-07/schema#",
  "title": "News Article",
  "description": "Structured news article for LLM pipelines and media monitoring.",
  "type": "object",
  "properties": {
    "headline": {
      "type": "string",
      "description": "Main headline of the article."
    },
    "subheadline": {
      "type": "string",
      "description": "Secondary headline or deck of the article."
    },
    "author_name": {
      "type": "string",
      "description": "Name of the article's author."
    },
    "author_url": {
      "type": "string",
      "description": "URL to the author's profile or bio page."
    },
    "publication_name": {
      "type": "string",
      "description": "Name of the publication or outlet."
    },
    "publication_domain": {
      "type": "string",
      "description": "Domain of the publication (e.g., 'techcrunch.com')."
    },
    "published_at": {
      "type": "string",
      "format": "date-time",
      "description": "ISO 8601 datetime when the article was first published."
    },
    "updated_at": {
      "type": "string",
      "format": "date-time",
      "description": "ISO 8601 datetime when the article was last updated."
    },
    "article_url": {
      "type": "string",
      "description": "Direct URL to the article."
    },
    "canonical_url": {
      "type": "string",
      "description": "Canonical URL as declared by the article page."
    },
    "body_text": {
      "type": "string",
      "description": "Full body text of the article."
    },
    "summary": {
      "type": "string",
      "description": "Short summary or abstract of the article."
    },
    "word_count": {
      "type": "number",
      "description": "Approximate word count of the article body."
    },
    "language": {
      "type": "string",
      "description": "BCP-47 language code of the article (e.g., 'en')."
    },
    "section": {
      "type": "string",
      "description": "Publication section or category the article appears in."
    },
    "tags": {
      "type": "array",
      "description": "List of tags or topic labels applied to the article.",
      "items": {
        "type": "string"
      }
    },
    "entities_mentioned": {
      "type": "array",
      "description": "Named entities mentioned in the article.",
      "items": {
        "$ref": "#/$defs/EntityMention"
      }
    },
    "sentiment": {
      "type": "string",
      "description": "Overall sentiment of the article.",
      "enum": ["positive", "negative", "neutral", "other"]
    },
    "paywall": {
      "type": "boolean",
      "description": "Whether the article is behind a paywall."
    },
    "image_url": {
      "type": "string",
      "description": "URL to the article's featured image."
    },
    "page_title": {
      "type": "string",
      "description": "Title of the source page. Tabstack auto-fills this from page metadata when left empty."
    },
    "favicon": {
      "type": "string",
      "format": "uri",
      "description": "Favicon URL of the source page. Tabstack auto-fills this from page metadata when left empty."
    }
  },
  "$defs": {
    "EntityMention": {
      "type": "object",
      "description": "A named entity referenced in the article.",
      "properties": {
        "name": {
          "type": "string",
          "description": "Name of the entity."
        },
        "type": {
          "type": "string",
          "description": "Category of the entity.",
          "enum": ["company", "person", "product", "location", "other"]
        },
        "url": {
          "type": "string",
          "description": "Optional URL associated with the entity."
        }
      },
      "required": ["name", "type"]
    }
  },
  "required": ["headline", "publication_name", "published_at", "article_url"]
}