Purpose
This block represents a flow input. It supplies a table passed by “Call flow” or “Call flows”. One flow may have several such inputs. Launch parameters are passed separately by name and handled by a “Run parameters” block.
Ports and schema
There are no input ports; out supplies data to later transformations. fields defines expected columns; an empty list keeps all supplied fields. Names must be unique and at most 256 characters; do not use the reserved __bridge_ prefix.
A different launch method
direct=empty supplies an empty table of declared fields when the flow was not started by another flow. direct=refuse stops such a run with an explanation. A subflow call chooses its input; other inputs receive empty data for that call. A previously saved restriction to a specific caller may still exist; the panel displays it and offers acceptance from any flow.
Input setup
- Add the block and name it after the table’s purpose.
- Enter expected fields one per line or accept the list from a detected calling block.
- Choose empty data or refusal for other launch methods.
- Connect processing blocks and publish.
- In the parent, select this input, pass a table and verify its schema.
Contract checks
Save the flow to see other “Call flows” blocks that start it. Check whether they pass data and which fields. Equal parameter and column names do not make them the same object. An empty direct-run result does not prove that the parent has no data.
Shared-processing example
Create a “Sales” input with Product, Period and Amount. Connect product normalization, a period filter and grouping. Different parents supply data from different sources through this contract; the subflow output returns one consistent structure.