vector-assist-define-spec

The “vector-assist-define-spec” tool defines a new vector specification by capturing the user’s intent and requirements for a vector search workload, generating SQL recommendations for setting up database, embeddings, and vector indexes.

3 minute read

About

The vector-assist-define-spec tool defines a new vector specification by capturing the user’s intent and requirements for a vector search workload. It generates a complete, ordered set of SQL recommendations required to set up the database, embeddings, and vector indexes.

Use this tool at the very beginning of the vector setup process when an agent or user first wants to configure a table for vector search, generate embeddings, or create a new vector index. Under the hood, this tool connects to the target database and executes the vector_assist.define_spec function to generate the necessary specifications.

Compatible Sources

This tool can be used with the following database sources:

Source Name
Cloud SQL for PostgreSQL Source

Requirements

Tip

Ensure that your target PostgreSQL database has the required vector_assist extension installed, in order for this tool to execute successfully.

Parameters

The tool takes the following input parameters:

Parameter	Type	Description	Required
`table_name`	string	Target table name for setting up the vector workload.	Yes
`schema_name`	string	Name of the schema containing the target table.	No
`spec_id`	string	Unique ID for the vector specification; auto-generated if omitted.	No
`vector_column_name`	string	Name of the column containing the vector embeddings.	No
`text_column_name`	string	Name of the text column for setting up vector search.	No
`vector_index_type`	string	Type of vector index (‘hnsw’, ‘ivfflat’, or ‘scann’).	No
`embeddings_available`	boolean	Indicates if vector embeddings already exist in the table.	No
`num_vectors`	integer	Expected total number of vectors in the dataset.	No
`dimensionality`	integer	Dimension of existing vectors or the chosen embedding model.	No
`embedding_model`	string	Model to be used for generating vector embeddings.	No
`prefilter_column_names`	array	List of columns to use for prefiltering vector queries.	No
`distance_func`	string	Distance function for comparing vectors (‘cosine’, ‘ip’, ’l2’, ’l1’).	No
`quantization`	string	Quantization method for vector indexes (’none’, ‘halfvec’, ‘bit’).	No
`memory_budget_kb`	integer	Maximum memory (in KB) the index can use during build.	No
`target_recall`	float	Target recall rate for standard vector queries using this index.	No
`target_top_k`	integer	Number of top results (top-K) to retrieve per query.	No
`tune_vector_index`	boolean	Indicates whether automatic tuning is required for the index.	No

Note Parameters are marked as required or optional based on the vector assist function definitions. The function may perform further validation on optional parameters to ensure all necessary data is available before returning a response.

Example

kind: tool
name: define_spec
type: vector-assist-define-spec
source: my-database-source
description: "This tool defines a new vector specification by capturing the user's intent and requirements for a vector search workload. This generates a complete, ordered set of SQL recommendations required to set up the database, embeddings, and vector indexes. Use this tool at the very beginning of the vector setup process when a user first wants to configure a table for vector search, generate embeddings, or create a new vector index."

Reference

field	type	required	description
type	string	true	Must be “vector-assist-define-spec”.
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.

Feedback

Was this page helpful?

Glad to hear it! Please tell us how we can improve.

Sorry to hear that. Please tell us how we can improve.

Last modified April 13, 2026: chore(main): release 1.1.0 (#3024) (da6f5f8)