PGVector Vector StoreInstall

Summary

Query similar embeddings in a PGVector store collection

Inputs

• collection (required) — Collection name to organize embeddings (partition value)
• queryEmbedding (required) — Query embedding vector as JSON array of numbers
• topK — Number of results to return
• offset — Number of results to skip (for pagination)
• distanceMetric — Similarity metric for vector search: cosine, l2, or inner_product
• metadataFilter — JSON filter for metadata using JSONB containment
• includeEmbedding — Whether to include embedding vectors in results

Output shape

An array of result items, each containing a json property with the matched embedding row data retrieved from the database.

The query operation reads the collection name and query embedding vector, constructs a search payload with specified distance metric, offset, and optional metadata filter, then returns matching rows. All parameter defaults and types are as declared in the node's UI properties and source execution flow.