MinimemoryInstall

Summary

Batch insert or update multiple vectors in an in-memory vector database using field mappings from all incoming workflow items.

Inputs

• databaseName (required) — The name identifying the target in-memory database instance.
• idField (required) — The property key in incoming item JSON that contains the unique vector ID.
• vectorField (required) — The property key in incoming item JSON that contains the array of vector numbers.
• metadataFields — Comma-separated field names from incoming items to store as vector metadata. If left empty, all incoming properties other than the ID and vector fields are stored as metadata.

Output shape

a single record

Returns a summary object containing the database name, total items processed, successful insert count, error count with item-level error details, and total vector count in the database.

Examples

Example 1: Insert vectors specifying explicit metadata fields

Database Name: 'docs', ID Field: 'id', Vector Field: 'embedding', Metadata Fields: 'title, category'

Example 2: Upsert vectors capturing all remaining item properties as metadata

Database Name: 'default', ID Field: 'doc_id', Vector Field: 'vector', Metadata Fields: ''