Qdrant Search Tool HeliconeInstall

Summary

Search Qdrant vector collections for relevant documents using OpenAI embeddings with Helicone observability integration, returning formatted results with optional metadata and filtering.

Inputs

• Tool Name (required) — Name of the tool used by the agent to identify and call it.
• Tool Description (required) — Description of what this tool does to help the agent decide when to use it.
• Collection Name (required) — Name of the Qdrant collection to search.
• Embedding Model (required) — OpenAI embedding model to generate query embeddings.
• Options — Configuration options including content and metadata payload keys, whether to include metadata in results, minimum similarity score threshold, search filters in Qdrant JSON syntax, and number of top results (Top K) to return.
• Helicone Options — Helicone observability options such as session ID to group requests and user ID for attribution.

Output shape

a single formatted string combining all matched documents or a message when no results found or error occurs

The output is a text block listing matching Qdrant documents with their similarity scores and optionally their metadata. Results are limited by Top K and filtered by score threshold and JSON filters if provided.

Examples

Example 1: Search a Qdrant collection named 'myCollection' for documents relevant to a user's query, returning up to 4 results with metadata included.

Set 'Collection Name' to 'myCollection', use default embedding model, set Top K to 4, enable include metadata, and provide query when the tool is invoked.