Oracle Database Vector Store: ToolInstall

Summary

Creates an Oracle Database vector search tool for an AI agent. It embeds each search query, retrieves up to the configured limit of nearest documents from the specified table, and returns their text with optional metadata.

Inputs

• Name (required) — Tool name exposed to the AI agent.
• Description (required) — Description shown to the AI agent to guide when it should use the tool.
• Table Name (required) — Oracle table containing vector embeddings and documents.
• Limit — Maximum number of matching documents to retrieve; defaults to 4.
• Include Metadata — When enabled (default), return each document's metadata with its text. When disabled, return only page content.
• Distance Strategy — Similarity measure for vector search: EUCLIDEAN_DISTANCE, DOT_PRODUCT, or COSINE.
• Embedding (required) — Connect an embedding model; it embeds the incoming query for searching.
• Input (required) — In direct execution, each input item must have a string json.input containing the search query. As an AI tool, the query is supplied by the agent.
• Oracle Database credentials (required) — Connects to the Oracle database using the configured host, port, service name, user, and password.

Output shape

An AI Tool connection; direct execution returns one output item per input item, each with a response array of text content entries.

Each result contains serialized document content, including pageContent and, by default, metadata. The tool path returns matching documents in its response; direct execution errors if an input item lacks a string input field. Search uses the configured distance strategy and limit.

Examples

Example 1: Let an AI agent search a knowledge table

Connect an embedding model, enter the Oracle table name, set a tool name and a clear description, and choose a result limit. Keep Include Metadata enabled to provide document metadata alongside matching text.

Links

• n8n community nodes installation guide