Skip to main content
The download output captures a run’s payload and makes it downloadable from the Runs page. It is a terminal node. Nothing connects after it. Use a download output when you want a copy of the structured result without sending it to an external system. It pairs well with extract, transform, and review actions.

​
Configuration

The download output has no configuration. Drop it onto the canvas, connect an upstream action, and the run output becomes available for download once the run completes.

​
How downloads work

  1. When a run reaches the download output, it captures the upstream payload and stores it against the run
  2. On the Runs page, open the actions menu on a row and choose Download output
  3. The browser downloads a file containing the captured payload
The file format follows the upstream step. A download output placed after a transform action that emits CSV delivers a .csv file; without a transform above it, the payload is delivered as JSON.

​
Downloading a specific node’s output

You can also download from the run detail view. Open a run, select the download output step, and its header shows a Download button. A workflow may contain more than one download output, for example one capturing line items and another capturing a summary. Each node’s Download button serves that node’s own captured payload, and the downloaded filename includes the node’s name so the two files are told apart. The Download output action on the Runs page is run-scoped rather than node-scoped: it serves the last download output the run reached.
Give each download output a distinct name in the editor. The name becomes part of the downloaded filename.

​
When there is nothing to capture

A download output only has something to capture if the step above it produced a payload. When it does not, the step is marked Skipped with a reason instead of completing, and no download is offered for it. This happens when:
  • No upstream step produced output, because the run followed a different branch
  • The step above it was itself skipped, for example by a filter
  • The step above it produced an empty payload
A skipped download output does not fail the run. The run still completes, and the step strip shows the skip and its reason. If the captured payload cannot be read at all, the step fails rather than reporting a successful capture of nothing. Because the download output is terminal, it has no error branch to route that failure into, so a failed download output fails the run.
The Download output action is unavailable when a run never reached a download output (it errored earlier or followed a different branch), when the only download output it reached was skipped, and while the run is still in flight. Output is offered once the run has completed.

​
Inputs and outputs

Allowed inputs: Any action node (extract, review, validation, transform, merge, parse, HTTP, loop, variable). Output: Terminal node, no downstream connections. The captured payload is exposed as a download on the Runs page and in the run detail.

Callback output

Send results to an external webhook instead

Email output

Deliver results via email

Forward output

Hand the document off to another pipeline

Runs

Find and download outputs from completed runs