Skip to main content
Worlds supports multiple strategies for ingesting data into isolated world memory, ranging from one-time imports to real-time synchronization.

​
Ingestion strategies

​
One-time import

Use the CLI or SDK to perform a bulk import of existing RDF data. This is ideal for bootstrapping a new world with a base ontology or research dataset. 
# Import a Turtle file into a world
worlds import --world <world-id> --file ./ontology.ttl

​
Direct SDK ingestion

For real-time applications, use the worlds-sdk to insert individual triples or chunks as they are generated by your agentic loops. 
await sdk.world("<world-id>").triples.insert({
  subject: "ex:User123",
  predicate: "rdf:type",
  object: "ex:Contributor",
});

​
Data synchronization

To keep your world memory in sync with external sources, utilize our forward-sync proxy.
  1. Map external events: Capture changes in your source system.
  2. Translate to patches: Convert changes into RDF patch operations.
  3. Apply patches: Use the rdf-patch service to update the world graph atomically.
Always skolemize blank nodes before ingestion to ensure deterministic triple IDs.