Patterns

A pattern is the artefact a pattern detection produces — a named, clustered view of a dataset that groups posts by recurring narrative or angle. Patterns are read-only on the API; create them via pattern detections or via a tracking agent.

Endpoints

curl -X GET "https://example.com/v1/patterns/string"

null

{
  "detail": [
    {
      "loc": [
        "string"
      ],
      "msg": "string",
      "type": "string"
    }
  ]
}

curl -X GET "https://example.com/v1/patterns/string/items/string"

null

{
  "detail": [
    {
      "loc": [
        "string"
      ],
      "msg": "string",
      "type": "string"
    }
  ]
}

Shape

GET /v1/patterns/{id} returns a top-level summary plus an items[] array of PatternItem summaries. Each item carries:

• A short title and synopsis.
• A post_count (how many posts cluster into it).
• An item_id you can use to drill into the underlying posts.

GET /v1/patterns/{pattern_id}/items/{item_id} expands one item with the full list of constituent posts. Use the mentions endpoint with the returned post_refs to fetch the post bodies.

Pattern in references[]

Assistant turns sometimes cite a pattern under references[] with { "type": "pattern", "id": "pt_..." }. Resolve via the endpoint above.

Next steps

• Pattern detections — how to produce a pattern from scratch.
• Tracking agents — patterns produced on a schedule.
• Mentions — fetch the underlying posts via post_refs.