Transform
Run an installed field operation, join or AI transformation
A transform node executes a workspace-installed transformation capability. Each operation declares its parameters, named inputs, outputs and lineage mode. The bundled Transforms package supplies field operations and AI tasks; custom transforms use the same binding model.
Before using it
Create or select the real resources described above in the pipeline’s project/workspace, inspect the upstream data shape and ensure your actor can perform this operation. A configuration fragment cannot create those resources. Read First pipeline for a complete graph and fixture.
Fields and bindings
| Field | Meaning |
|---|---|
capabilityRefId | The installed transform capability UUID, not a package or endpoint ID |
function | The operation implemented by the selected capability |
input | One pool name or a map of named inputs to pool names |
output | Intermediate output name |
params | Operation-specific parameters validated against its manifest |
Configure
Prepare the transform node for the selected test resources. Use the operation and fields shown in this example. Show actual resource bindings, upstream/downstream ports and any write/model effects before applying the draft.Choose the installed capability and use its parameter controls. For select-project, add Fields entries for source order_id → target order_id and source total → target order_total. Multi-input lookup instead needs primary/reference ports and bindings.
Put this section under transform on a full node of kind transform. It is not a standalone API or MCP request. Replace resource placeholders, and provide the full node ports/policy and graph edges.
{
"capabilityRefId": "<SELECT_PROJECT_CAPABILITY_UUID>",
"function": "select-project",
"input": "orders",
"output": "projected_orders",
"params": {
"fields": [
{
"source": "order_id",
"target": "order_id"
},
{
"source": "total",
"target": "order_total"
}
]
}
}Verify a bounded run
For two orders, projection should emit two rows with order_id and order_total. Check lineage and inspect missing-field behavior. For filters/joins/aggregates, test changed cardinality and explicit evidence linkage.
Validate, save a version and run a small known fixture. Inspect the actual node result and downstream consumer, not only the graph preview.
Limits and failure behavior
Installing the bundle does not supply model credentials. AI classification/extraction/semantic mapping require the configured model runtime. Preflight rejects capabilities declared runtimeSupported false. Operation names alone do not install executable code.