Skip to content

select

select is a schema decision. It keeps the columns you name, in the order you name them.

Use it to make the next node cheaper and clearer: trim wide source tables, prepare a report table, or define the exact payload crossing into a language node.

FieldRequiredNotes
inputsyesExactly one upstream table.
columnsyesArray of selected columns. Runtime compiles each entry as an expression, while the schema currently restricts them to identifier-shaped strings.
  • default is the same rows with only the selected columns.
  • Column order is part of the node behavior, so use select when report ordering matters.
  • Selecting a nonexistent column is a hard validation error.
- id: keep_cols
kind: select
inputs: [data]
columns: [a, b, c]