EDB Vector Index v1.4.0 (LTS)

The EDB Vector Index component manages pgvector indexes on an existing table. It can create a new index, drop an existing one, rebuild it, or just check whether it exists.

Choosing the right component

Use the EDB Vector Index component when:

  • You already have a Postgres table with a vector column and want to add (or change) an index to accelerate similarity search.
  • You want to switch index type (HNSW vs IVFFlat) or change distance metric on an existing column.
  • You want to verify from a flow whether a particular index is in place before running other operations.

Use a different component if:

  • You want to query a knowledge base, not manage index structure. Use EDB Knowledge Base.
  • You want raw SQL access to the database. Use EDB Database.
  • The table doesn't have a vector column yet. The component only manages indexes on existing vector columns. Add the column first (with aidb or your own DDL).

Prerequisites

  • A Postgres role with the CREATE INDEX privilege on the target table.
  • An HM cluster with the pgvector extension installed and a table containing at least one vector column.
  • An HM machine-user access key and the database user/password saved in Langflow as Global Variables.

Inputs

Connection

FieldTypeRequiredDefaultNotes
Hybrid Manager URLTextNoAdvanced. Override the default HM URL.
HM Machine User KeySecretYesHM_API_KEYDefaults to the global variable named HM_API_KEY.
Hybrid Manager ProjectDropdownYesPopulated from your HM projects.
Hybrid Manager DatabaseDropdownYesThe cluster containing the target table.
Database GroupDropdownNoCluster group, for example primary or reader.
Database Connection TypeDropdownYesFor example, read or read/write.
User for the databaseTextYesHM_DB_USERDefaults to the HM_DB_USER global variable.
Password for the databaseSecretYesHM_DB_PASSWORDDefaults to the HM_DB_PASSWORD global variable.
Default database name overrideTextNoAdvanced. Override the cluster default when listing databases.
Database NameDropdownYesDatabase containing tables with vector columns.

Target

FieldTypeRequiredDefaultNotes
TableDropdownYesTables in the selected database, excluding system schemas.
Vector ColumnDropdownYesPopulated from columns of type vector in the selected table.
Distance OperationDropdownYesCosineOne of Cosine, L2, InnerProduct. Maps to pgvector operator classes vector_cosine_ops, vector_l2_ops, vector_ip_ops.
Index TypeDropdownYeshnswhnsw or ivfflat.
OperationDropdownYesCreateOne of Create, Drop, Rebuild, Check Exists. See Operations.

HNSW parameters (Index Type = hnsw)

FieldTypeRequiredDefaultNotes
HNSW M (Max Connections)IntegerNo16Advanced. Maximum number of connections per node.
HNSW ef_constructionIntegerNo64Advanced. Size of the dynamic candidate list during index construction.

IVFFlat parameters (Index Type = ivfflat)

FieldTypeRequiredDefaultNotes
IVFFlat ListsIntegerNo100Advanced. Number of inverted lists.

Outputs

OutputTypeCarries
ResultDataOperation outcome. Always includes success, operation, table, column, index_name, and message. Check Exists adds exists. Create adds index_type and the generated sql when it builds a new index, and already_existed when an index was already present (set both on an idempotent same-type match and on a different-type conflict).

Operations

OperationBehavior
CreateCreates the index if it doesn't exist. Idempotent: if an index of the same type already exists on this column, the operation returns success without re-creating. If an index of a different type exists on the column, the operation fails and asks you to Drop it first.
DropDrops the index. Idempotent (uses DROP INDEX IF EXISTS).
RebuildRuns REINDEX INDEX on the existing index. Fails if the index doesn't exist.
Check ExistsReturns exists: true/false. Does not modify the database.

Index naming

The component generates a stable, predictable index name from the table and column: idx_<table>_<column>_<index_type> (without schema prefix). Choose your indexes through this component if you want that naming convention. Otherwise, manage them in SQL.