Steps and behaviour

Every step is a node on the graph. Triggers start a run; everything downstream receives the previous step's output automatically, so you wire behaviour with connections and with expression variables (tags like {{input.*}}) instead of manual field mapping on each step.

Cross-cutting topics

Read these first: they apply to many step types.

• Expression variables: what tags are, when they resolve, and the main namespaces (input, exe, global, and others).
• Execution settings: the Execution tab in the step sheet: which step types have one and what you configure there.

How these guides are structured

Each step family page follows the same outline so you can skim predictably:

1. Overview: what the family is for on the canvas.
2. Steps in this family: short descriptions and how they differ.
3. Configuration: where settings live in the editor (General, Execution, Output, Gate, and so on).
4. Data flow: how inbound data and tags behave for this group.
5. Related: links back to expression variables, execution settings, or deep dives (for example Code).

Step family guides

Each family has an overview plus a short article for every catalogue template.

Triggers

• Overview
• Invoke
• Webhook
• Schedule

Logic

• Overview
• Decision
• Switch
• Split

Human

• Overview
• Approval

AI

• Overview
• Generate text
• Summarise content
• Classify input
• Extract data
• Chat completion
• Transform data

Code

• Run code
• Random number
• Iteration
• Code helpers (random and iteration in more detail)

Documents

• Overview
• Document from Template
• Generate document (XML)

Actions

• Overview
• Action
• Webhook

Termination

• Overview
• End