The SQLite of graph databases. Embeddable, AI-native, Apache 2.0.
An embeddable graph database written in Rust. Single file, zero setup, Cypher queries, native vector search, and built-in MCP support for AI agents.
Status: Pre-release — designing and building the foundation
There's no good embeddable graph database that is truly open source, speaks Cypher, and works natively with AI tools. Neo4j requires a server and is AGPL. KuzuDB was the closest alternative but was abandoned in October 2025. We're building what should exist.
opengraphdb init mydata.ogdb
opengraphdb query mydata.ogdb "MATCH (n:Person)-[:KNOWS]->(m) RETURN n.name, m.name"
One binary. One file. That's it.
- Embeddable — use as a Rust/Python/JS library or standalone CLI
- Cypher / GQL — the query language developers already know
- Graph + Vector + Full-text in one engine, not three
- RDF/TTL import — bring your ontologies, query them with Cypher
- MCP server built-in — plug into Claude, Cursor, or any AI agent
- Rust, columnar storage, MVCC, WAL — the boring, correct foundations
This is a greenfield project. Everything is being built from scratch.
Areas we need help with:
- Storage engines (buffer pools, WAL, crash recovery)
- Query engines (parsing, optimization, execution)
- Rust systems programming
- Vector search / HNSW
- RDF and knowledge graphs
- AI agent tooling / MCP
- Developer experience and documentation
If any of this interests you, please reach out. Open an issue, start a discussion, or just say hello. Every contribution matters.
→ Issues · Discussions
Apache 2.0