Lattice DB is an embedded, single-file knowledge graph database with native vector and full-text indexing. It enables local applications to query data by relationship, semantics, and text in one engine and query layer. The database supports graph traversal, HNSW vector similarity, and BM25 full-text search, all within a single query language.

Designed for relationship-heavy workloads on a single machine, Lattice DB operates with zero-configuration and an embedded single-writer model. It includes durable named streams and a built-in graph changefeed sharing the same transaction/WAL path as graph writes. This local-first approach ensures WAL-backed durability for one owning process.

Performance benchmarks show 0.13 μs node lookups and 0.83 ms vector search at 1M vectors with 100% recall. The database supports workloads like Graph RAG, agent memory, and local knowledge tools. Installation is available via CLI, Python, TypeScript/Node.js, and Go bindings.

A complete example demonstrates creating a knowledge graph with documents and authors, storing embeddings, indexing text, and querying across all three search modes. The database is a single portable file with no server or configuration required.