Qdrant Search Tool LangfuseInstall

Summary

Provide a configurable Qdrant vector search tool with Langfuse tracing integration, enabling retrieval of relevant documents from a specified Qdrant collection using an AI embedding input.

Inputs

• Tool Name (required) — Specify the name of the tool, used by agents to identify and invoke it.
• Tool Description (required) — Describe the tool's functionality to guide agent usage decisions.
• Collection Name (required) — Set the Qdrant collection name to perform the search in.
• Options — Configure search options including content payload key, metadata inclusion, metadata key, score threshold for filtering by similarity, a JSON filter using Qdrant syntax, and number of results (Top K) to return.
• Langfuse Options — Set Langfuse tracing details such as session ID, user ID, and comma-separated tags to group and filter traces.

Output shape

a string containing the search results formatted as a numbered list with scores and optionally metadata, or an error/no results message

Returns a single string with formatted search results combining content and optional metadata for each matched document. The tool traces embedding and search operations with Langfuse. If no embedding input is connected or invalid JSON filter is provided, it throws an error. The embedding input is expected via a connected AI embedding node.