Limit Node

Cuts an iterable payload down to its first N items and drops the rest. Two fields, and one of them you will almost never fill. It exists so a feed of two hundred items does not turn into two hundred AI calls because nobody thought to cap it.

Limit

1 to 10,000, five by default. Whatever the upstream emitted, only the first N continue.

⚠It slices, it does not choose. The order is whatever the upstream produced — there is no “top” and no “newest” here. If you want the five biggest orders, sort them in a Transform or a Code node first, then limit.

Array field

Optional, and empty is the right answer nearly always. Left blank, the node reads payload._meta.iterateField — the same marker every iterable node sets — and finds the array by itself:

Upstream	Field it uses
RSS	items
HTTP Action returning an array	results
MongoDB `findAll`	documents
Split	outputs

Fill it in only when the upstream uses a name of its own — a custom HTTP response wrapping its rows in data.rows, say.

Where it belongs in a flow

Right after whatever produces the list, and before anything that costs money per item.

textThe expensive position

RSS (50 articles)
  → Limit (5)          ← here, not after
  → AI Node            5 calls instead of 50
  → Vector Store (insert)

Put it after the AI Node instead and you have already paid for the forty-five items you were about to throw away.

Testing it

The panel carries a Test Limit runner that invents a ten-item sample when nothing is connected upstream, so you can see the slice without building the rest of the flow first.