Search the documentation corpus by meaning — the query is embedded with the local nomic-embed-text model and cosine-ranked against the stored chunk vectors in embeddings. Rendered through the <<<SemanticSearch>>> tag.
Vector store empty — no embeddings ingested yet.