OxigraphDB
A graph database small enough to live where the agent lives.
An agent that has to ask a cloud for its own memory does not really own it. OxigraphDB is an open-source, embeddable graph database for agentic AI at the edge: what the agent knows lives in a standards-based graph, next to the model, on the device.
RDF for the data, SPARQL for the questions, SHACL for the shape — so the memory is not a proprietary blob. It outlives the agent that wrote it, and any tool that speaks the standards can read it.
What it does
- Time travel
- Queries over versioned graph state, so an agent can ask not only what is true, but what it knew at any point in time.
- Datalog reasoning
- A Datalog engine as a recursive reasoning layer: rule-based inference over the knowledge graph.
- Edge first
- Local-first, on-device agent memory with a lightweight footprint and no mandatory cloud dependency.
- Standards, not formats
- RDF, SPARQL and SHACL. Memory stays portable across agents, models and vendors.