Meet SQLite Graph, a bold extension that adds graph capabilities to SQLite via a virtual table extension. It brings Cypher-style CREATE, MATCH, RETURN inside a single database file, and it's an alpha—but working for core patterns [1].
What it is — A graph DB layer on top of SQLite implemented as a virtual table extension, exposing a Cypher interface for CREATE, MATCH, and RETURN. It stores and queries graph data inside one database file [1].
What works — • Full CREATE: nodes, relationships, properties, and chained patterns [1] • MATCH with relationship patterns like (a)-[r:TYPE]->(b) [1] • WHERE on node properties [1] • RETURN with basic projection and JSON-like serialization [1] • Virtual table integration lets SQL mix with Cypher [1]
Performance — 340K nodes/sec inserts (scales toward 1M); 390K edges/sec; 180K nodes/sec scans with WHERE filtering [1].
Current limitations — Forward relationships only (no bidirectional); no relationship-property filtering in WHERE; no variable-length paths; no aggregations, ORDER BY, or advanced RETURN projections; strings must use double quotes [1].
Roadmap — Bidirectional relationships, property projection, aggregations, and full Cypher by Q1 2026. Built as part of Agentflare AI, MIT licensed, standalone [1].
All-in-one graph-on-SQL vibes are real—this alpha proves end-to-end Cypher inside SQLite is possible today [1].
Referenced POST IDs: [1]
References
Show HN: SQLite Graph – Graph database with Cypher queries (alpha, but working)
Adds a graph database layer to SQLite with Cypher support; supports CREATE, MATCH, and RETURN via a virtual table extension.
View source