{
  "name": "Base2026 public data dictionary",
  "version": "2026-06-14",
  "public_boundary": {
    "allowed": [
      "attributed public source records",
      "reviewed public source text",
      "public passages",
      "public insight cards",
      "public topic and creator metadata",
      "deterministic topic signal briefs"
    ],
    "not_public": [
      "raw captions",
      "raw ASR",
      "private QA notes",
      "local databases",
      "credentials",
      "audio files",
      "video files",
      "unreviewed source vaults"
    ]
  },
  "datasets": [
    {
      "file": "/knowledge/static/manifest.json",
      "format": "json",
      "description": "Release-level public export counts, timestamps, and policy flags."
    },
    {
      "file": "/knowledge/static/documents.jsonl",
      "format": "jsonl",
      "description": "Search-ready public documents used by the Base2026 workspace.",
      "primary_keys": ["item_id"],
      "common_fields": [
        "item_id",
        "source_id",
        "creator_handle",
        "published_date",
        "source_type",
        "language",
        "source_url",
        "public_source_text",
        "source_summary_short",
        "source_summary_long",
        "topics",
        "topic_labels"
      ]
    },
    {
      "file": "/knowledge/static/passages.jsonl",
      "format": "jsonl",
      "description": "Public evidence passages linked to source records for search and citation.",
      "common_fields": [
        "source_id",
        "item_id",
        "body",
        "topics",
        "creator_handle"
      ]
    },
    {
      "file": "/knowledge/static/insight_cards.jsonl",
      "format": "jsonl",
      "description": "Reviewed public insight cards derived from public source evidence.",
      "common_fields": [
        "source_id",
        "topic_id",
        "topic",
        "creator_handle",
        "claim_text",
        "stance",
        "evidence_excerpt",
        "suggested_action",
        "public"
      ]
    },
    {
      "file": "/knowledge/static/topic_signal_briefs.jsonl",
      "format": "jsonl",
      "description": "Deterministic compact topic signal summaries generated only for strong public topics.",
      "strong_topic_threshold": {
        "source_count": ">= 5",
        "creator_count": ">= 2",
        "public_insight_count": ">= 3"
      },
      "common_fields": [
        "topic_id",
        "topic_label",
        "status",
        "source_count",
        "creator_count",
        "public_insight_count",
        "creator_angles",
        "repeated_tactics",
        "source_backed_actions",
        "monthly_activity",
        "top_sources"
      ]
    }
  ]
}