VecLiteEmbeddable vector search for Go

Store vectors, text, and metadata in a local single-file database with HNSW, BM25, filters, and hybrid search.

Local-first storage

Persist records, vectors, content, and metadata to one VecLite database file.

Use HNSW for approximate nearest neighbors, BM25 for keyword search, or hybrid ranking for both.

Import VecLite into applications that own chunking, media extraction, and embedding provider choices.

Use separate collections today and follow the named-vector-space roadmap for future multimodal records.