Postgres Vector Store Tool icon

Postgres Vector Store Tool

Vector similarity search tool for AI agents with pgvector support

Actions2

Vector Search

AI-generated

Summary

Perform a vector similarity search to find documents in a PostgreSQL table that are most relevant to a given search query embedding.

Inputs

  • agentDescription — Optional description guiding how an AI agent should use this tool.
  • Table Name — Name of the PostgreSQL table containing vectors (supports schema.table format). Defaults to 'n8n_vectors'.
  • Limit — Maximum number of matching results to return (1-1000). Defaults to 4.
  • Include Metadata — Whether to include the metadata column in the search results. Defaults to true.
  • Options — Additional optional settings including debug mode and custom column mappings for ID, vector, content, and metadata columns.

Output shape

a single JSON string representing an array of matching documents, each including at least the text content and optionally metadata fields.

The output is a JSON string with an array of search results ordered by vector similarity. The query parameter must be provided and is transformed into an embedding by the connected embedding input node. The node requires a PostgreSQL credential with appropriate access to the specified table. Debug option enables detailed logs for troubleshooting.

Discussion