The knowledge store object
A knowledge store contains your videos plus the understanding the platform derives from them: spatiotemporal context, a typed ontology, and embeddings for semantic retrieval. It is the corpus agents reason over — the full video collection, not just individual clips.
The object is composed of the following fields:
_id: A string representing the unique identifier of the knowledge store.name: A string representing the name of the knowledge store.description: A string representing the description of the knowledge store.ingestion_config: An object that controls how content added to the knowledge store is processed. You can shape extraction either with a JSON Schema for precise structured fields, or with a natural-language description. Immutable after creation. For the full structure, see theingestion_configparameter on the Create a knowledge store page.item_count: An integer representing the number of items in the knowledge store.metadata: An object containing custom metadata for the knowledge store. Both keys and values are strings.created_at: A string representing the date and time, in the RFC 3339 format, when the knowledge store was created.updated_at: A string representing the date and time, in the RFC 3339 format, when the knowledge store was last updated.
Examples
Each example shows a populated knowledge store. The first uses a JSON Schema to define the metadata to extract from each video; the second uses a natural-language description instead.