Delete
AI-generatedSummary
Delete embeddings from a PGVector collection by ID, external ID, or metadata filter.
Inputs
- collection (required) — Collection name to organize embeddings (partition value)
- schemaMode — How to interact with the database (Default Schema or Custom Schema)
- deleteBy — Delete by UUID, external ID, or metadata filter
- deleteIds — Comma-separated list of UUIDs to delete
- deleteExternalIds — Comma-separated list of external IDs to delete
- deleteMetadataFilter — JSON filter for metadata to delete matching records
- tableName — Database table name (shown when schemaMode is Custom)
- createTable — Create the table if it does not exist (shown when schemaMode is Custom)
- extraReturnColumns — Additional columns to return in query results (shown when schemaMode is Custom)
Output shape
The result of the delete operation, returned as a JSON object from the node's internal delete method (y.delete(i)).
When schemaMode is Custom, the tableName, createTable, columnMapping, and extraReturnColumns parameters become applicable. The deleteBy parameter selects which of deleteIds, deleteExternalIds, or deleteMetadataFilter is used. The collection parameter is always required as it partitions data.