Loops
Run a set of steps once per item in a list.
A loop node runs a body of steps once per item. Point it at an array and it iterates the array. Point it at a number and it iterates that many times. Use it to process every order in a batch, every row in a sheet, or every recipient in a list.
What it iterates
The loop's items is either a reference to an array or a count:
items value | Behavior |
|---|---|
{{list_orders.orders}} | Run the body once per order in the array |
25 | Run the body 25 times, item = index |
Inside the body, each pass sees its own current item, so downstream {{...}} references resolve to that iteration's data.
Concurrency
By default the loop runs one item at a time, in order. Raise concurrency up to 5 to fan items out in parallel. Keep it at 1 when steps depend on order or the target tool rate-limits. Use 2 to 5 for independent items that can run at once.
Errors
By default the loop is forgiving. If one item fails, its error is recorded and the loop keeps going, finishing the rest. The loop still succeeds, and you see per-item results in the run.
Set failFast to abort on the first error instead. Use it when a single failure means the whole batch is invalid.
The body
The steps a loop runs are its body, the nodes tagged to that loop. Build them as an ordinary sub-pipeline. The loop drives them once per item and collects each pass's output. A loop can hold one step or a whole branch.