Extend the Studio

Adding your own nodes to Integration Studio — the node contract, what one package ships, resolving fields and ports while the workflow is being edited, and how a package reaches a tenant.

When no built-in node does the job, write one. The contract is public, the SDK is on the public npm registry, and a package can ship credential types and workflow templates alongside its nodes.

This group is the Studio-side view: what a package is, what it can contain, and how it reaches a tenant. The Integration Node SDK is the authoring reference with the worked examples — a transform node, a branching control node, an action node with an error port. Read this group to decide what to build; read the SDK page while you build it.

The pages

  • Custom nodes — the four types you implement against, what the runtime hands your node, the project layout, and the two-step build.
  • Package manifest — one package = nodes + credential types + templates; slug namespacing as {org}:{name}; versioning and what a version pin means for a running workflow.
  • Dynamic behaviour — the three optional author-time hooks that let a node offer live dropdowns and resolve its own fields and ports from a connected system's schema.
  • Registration — operator-mediated, not self-service: what a partner hands over, and the every-tenant-sees-every-package behaviour to plan around.

Before you write a node

Two checks that save real work:

  1. Is it a node, or a workflow? A step that composes existing nodes is a template, not a node. Templates are plain data and much cheaper to ship.
  2. Is it a node, or an app? If the behaviour is part of the product rather than the wiring between two systems, it belongs in an App.

And one expectation to set with whoever is waiting for it: registration is a human step, not a deploy. Plan the turnaround into your release process — see Registration.