Build a workflow
Assumes you have connectors for the systems involved (Connectors and invokers). For a guided end-to-end example including the setup, see Your first workflow.
Open the editor
Alt+W, or Create Workflow in the top bar, or Workflows → Create. To
edit an existing one, use the pencil icon in the workflow list.
The editor replaces the application header with its own; the navigation sidebar stays. Four areas:
the header — title, description, schedules pill, command palette, test run, Save, and the header menu,
the canvas — the graph,
the step drawer — slides in from the right when you add a step,
the Execution Logs panel — collapsed until a test run produces output.
Add steps
Hover a node and use the + handle on its right or bottom edge. The drawer opens with Choose your next step and tells you where the step will land.
Option |
Use it for |
|---|---|
Use Connector |
The normal case. Pick a connector, then one of its operations. |
Add HTTP Request |
An endpoint you do not want to model as an invoker. See Call any API. |
Add Operator |
Branching ( |
Trigger Workflow |
Fire-and-forget another workflow. See Chain workflows. |
The search box at the top searches connectors, operators and methods at once.
Configure a step
Double-click a node, or right-click it for the context menu:
Change Label — a readable name of your own.
Open Configuration — the request and response.
Configure Aggregator — attach a data aggregator. Unassigning removes it from this step only; the aggregator itself stays available elsewhere.
Request → Edit URL / Edit Header / Edit Body
Show Response — the response definition (body, header, status).
For a connector method the HTTP method is read-only — it comes from the invoker. For a simple HTTP request you choose it.
References go directly into the URL field; the separate query-parameter editor of earlier versions is gone.
Map the data
Open Body on the receiving step. The body is a JSON tree (or XML, or GraphQL — see Call any API). Select a field, then insert a reference to an earlier step’s response.
A field must hold either plain text or references — mixed values are rejected.
Prefer plain references over scripts; only add an enhancement where the value
genuinely needs transforming. Data mapping covers the rules,
including arrays, loop iterators and the VAR_0 / RESULT_VAR contract.
Read the canvas
Colour + number badge — steps sharing them call the same method on the same connector.
Aggregator badge — a data aggregator is attached; click it to configure.
Asynchronous badge — a Trigger Workflow step, which does not wait.
Status dot — the connector’s health, see Connectors and invokers.
Red outline — a validation error from the last save or test start. It clears on the next attempt or when you edit the node.
Find things in a large workflow
In the editor’s command palette, type workflow to lock the scope, then
workflow search <term>. The search is fuzzy and matches method names, URLs,
headers, query parameters, request and response bodies, and operator conditions.
Matches are highlighted yellow on the canvas; Esc clears them.
Save
Ctrl+S, or Save. Each save creates a version and asks for a comment.
Leaving with unsaved changes asks for confirmation — including on tab close and
reload.
Where to go next
Debug a workflow — run it and read the logs.
Schedule and notify — run it regularly.
Reuse with templates — reuse it elsewhere.