Templates and expressions: pass data between nodes

• {{ trigger.<field> }}: the inputs that started the run, such as a test input on the Manual Trigger, a schedule's inputs or an API call's input.
• {{ nodes.<id>.output }}: what an upstream node returned on the port that fired.
• {{ nodes.<id>.ports.<port> }} and {{ nodes.<id>.status }}: per-port output and the node's status.
• {{ run.id }} and {{ run.started_at }}.
• {{ args.<field> }}: inside an HTTP Request node attached as a tool, the arguments the agent called it with.

Under every template field, Insert variable lists the run inputs the workflow already knows and drops {{ trigger.<name> }} where your cursor is. Choose New variable to add one: the name is selected, so you can type over it straight away. Typing {{ also suggests run inputs and upstream nodes.

Each variable a prompt reads then shows up on the Manual Trigger's Test inputs and in the Playground's Inputs form, as a row waiting for a value.

A saved endpoint on a custom API uses single braces, such as /items/{item_id}. Each placeholder becomes an input on the API node. Fill that input with a value or a template:

item_id: {{ trigger.item_id }}
title:   {{ nodes.summarise.output.text }}

The endpoint defines the shape of the request once; the workflow supplies the values with templates.

Fields such as an agent's user prompt or an HTTP node's URL are rendered as text. A field that is exactly one expression keeps the raw value instead, so an object stays an object:

{
  "summary": "Ticket from {{ trigger.customer }}: {{ trigger.message }}",
  "ticket": "{{ trigger }}"
}

If conditions and Switch cases are bare expressions without braces, evaluated to a value:

nodes.classify.output.structured.urgency == 'high'

An Agent node returns text, and structured when an output schema is set. Branch on structured fields rather than parsing text.