Build the workflow

The workflow document and its revisions, the 52 built-in nodes, the expression syntax that wires them together, how a node's configuration fields work, and the templates you can start from.

Build the workflow

The workflow document

Nodes, edges and ports, variables, canvas organisation, validation, and how a revision is numbered.

Node catalog

All 52 built-in nodes by category, with their slugs and the ports they declare.

Expressions

`${{ }}`, the four reference roots, array auto-mapping, and the two filters.

Node configuration

The configuration field types a node can declare, including credential and secret references.

Templates

The eight shipped workflow templates, the readiness check, and what you adapt after installing one.

Building a workflow is four decisions, and this group takes them one at a time: what the graph is, which nodes go in it, how each node is configured, and where data flows between them.

The pages

  • The workflow document — a workflow is a versioned JSON document. Nodes and their pinned versions, edges between ports, workflow variables, the canvas organisation the engine ignores, and what saving does to the revision number.
  • Node catalog — the 52 nodes that ship with the platform's own package, grouped by category, with the namespace:slug you reference and the output ports you can wire.
  • Expressions — the ${{ … }} syntax: four reference roots, dot paths that map over arrays, two filters, and how a resolved value is typed.
  • Node configuration — the field types a node declares, what the inspector does with each one, and the two reference types that keep access data out of the workflow document.
  • Templates — the eight shipped templates, the readiness check that counts credentials and secrets against your tenant, and what is left to adapt after you install one.

The shortest honest path

If you are starting from nothing, do not read this group front to back. Do this instead:

  1. Check Templates first — a file-shaped integration is probably already there.
  2. Otherwise build the smallest graph that runs, following Getting started.
  3. Reach for the node catalog when you know what the next step has to do.
  4. Reach for expressions when you need one step to read another's output.