Module embedding_store 

Qdrant-backed embedding store for message vector search.

EmbeddingStore owns a VectorStore implementation (Qdrant in production, crate::db_vector_store::DbVectorStore in tests) and exposes typed embed / search / delete operations used by crate::semantic::SemanticMemory.

Message vectors are stored in the zeph_conversations Qdrant collection with a payload that includes message_id, conversation_id, role, and category.

Structs

EmbeddingStore

Typed wrapper over a VectorStore backend for conversation message embeddings.

Filter

SearchFilter

Optional filters applied to a vector similarity search.

SearchResult

A single result returned by EmbeddingStore::search.

Enums

MessageKind

Distinguishes regular messages from summaries when storing embeddings.

Functions

ensure_qdrant_collection

Ensure a Qdrant collection exists with cosine distance vectors.