Skip to content

In-Memory Store

Type: inmemory

An in-process BM25 lexical store with no external dependencies. Scores documents using term frequency / inverse document frequency (BM25, k1=1.2, b=0.75). All data is lost when the sidecar restarts.

Best for: development, unit tests, small static corpora where exact-keyword retrieval is sufficient.

Configuration

- name: docs
  type: inmemory

No metadata keys are required. No external service needed.

Notes

  • Retrieval is lexical, not semantic — there is no embedding model. The query text is tokenised and BM25-scored against stored documents.
  • For semantic similarity search, use chroma, qdrant, redis, or pgvector with a configured embedder.
  • IDs are UUID v4 strings when the caller omits the id field.