vector-assist-improve-query-recall - Genai Toolbox

About

The vector-assist-improve-query-recall tool is designed to troubleshoot and optimize existing vector search workloads when a user reports irrelevant results, poor accuracy, or degraded recall.

It determines the optimal tuning parameter (such as hnsw.ef_search) for an active vector index to improve the search results. The tool evaluates the target recall and outputs an actionable SQL query recommendation (e.g., SET hnsw.ef_search TO ...) that must be executed as a next action using the execute_sql tool.

Compatible Sources

Requirements

{{< notice tip >}} Ensure that your target PostgreSQL database has the required vector_assist extension installed. This tool requires an existing HNSW index to function properly; if the table lacks an existing vector setup, use the define_spec tool instead. {{< /notice >}}

Parameters

The tool takes the following input parameters:

Parameter	Type	Description	Required
`table_name`	string	Table name experiencing degraded vector search recall.	Yes
`vector_column_name`	string	Column name containing the vector embeddings.	Yes
`index_name`	string	Name of the vector index to tune.	Yes
`schema_name`	string	Schema name of the table (default is `public`).	No
`top_k`	integer	Top k value for the vector search (default is `10`).	No
`target_recall`	float	Target recall value for search results (default is `0.95`).	No
`distance_func`	string	Distance function used for the vector search similarity (default is `cosine`).	No

Note : Parameters are marked as required or optional based on the tool's parameter definitions. The underlying function may perform further validation on optional parameters to ensure all necessary data is available before returning a response.

Example

yaml

kind: tool
name: improve_query_recall
type: vector-assist-improve-query-recall
source: my-database-source
description: "Use this tool to troubleshoot and optimize existing vector search workloads when a user reports irrelevant results, poor accuracy, or degraded recall. It determines the optimal tuning parameter (such as ef_search) for an active vector index to improve the search results. The tool outputs an actionable SQL query recommendation to be executed as a next action using the 'execute_sql' tool."

Reference

field	type	required	description
type	string	true	Must be "vector-assist-improve-query-recall".
source	string	true	Name of the source the SQL should execute on.
description	string	false	Description of the tool that is passed to the agent.