Skip to main content
Data flows from a step to the steps after it. The trigger comes first and hands its data to step 1. Step 1 hands its result to step 2, and so on. A step can use the results of every step before it, not only the one directly before.

An example

1

Trigger: Work item created

The trigger hands over the key, the title and the project of the new work item.
2

Step 1: Send notification

Its message uses the key and the title from the trigger. It returns whether the notification was sent.
3

Step 2: Add comment

Its text uses the key from the trigger and the result of step 1. Because it is the last step, nothing can use its result.

Insert data from an earlier step

Select an input and the data picker opens next to it.
  • Each earlier step is a group. Open it to see the fields of its result.
  • A small arrow shows that a field has fields of its own.
  • Select a field to insert it at the cursor. You can mix typed text and data in one input, for example New: {{trigger.title}}.
  • A reference reads {{step_1.field}}. Collabase replaces it with the value when the step runs.
  • When the output of a step is a list or a single value, the reference reads {{step_1.output}}, and an item of a list reads {{step_1.output[0]}}.
For a step whose output is an object, {{step_1.field}} and {{step_1.output.field}} are the same. Use whichever the picker inserts.
The picker lists the fields of the saved test result of each step. Test every step before the one you are filling in. A step that has not been tested shows no fields.

Test each step

Select a step, open its Test tab and run it. Collabase runs the step once with the saved data of the steps before it and shows the result. The result is saved, so later steps can offer its fields in the picker.
A step test runs the step for real. It can create a work item, send a notification or call another system.

Use a value that is not listed

If the field you need is not in the picker, type its path:
Start from a field the picker offers and add the names of the fields inside it, separated by dots. The data of a webhook call sits under body, for example {{trigger.body.order}}. A reference that points to nothing becomes an empty value, and so does a reference to a step that does not exist or has not run. If the step then needs that value, it fails with a message that names the field, for example Invalid input: taskId. Test the step so the picker offers the fields of the earlier steps. A step that failed has no output, only an error. {{step_1.error.message}} reads its message, which is useful after Continue on failure.

What publishing checks

You do not have to find a wrong reference by running the flow. Publish is refused while a reference has one of these problems, and the issue list names the field: A reference holds only a path. Under a field, small chips show the references it contains and whether the data of the earlier steps has that path. A path that is not found says to test the step first. A dropdown, a multiple choice, a checkbox, a list of values, names and values and rows have the switch Dynamic value next to their label, a pair of braces. Turn it on to change the field into an expression field, where you type text and insert data from earlier steps. Turn it off to go back to the typed control. The value does not carry over in either direction, because a choice and an expression are not the same thing. Dates and numbers have a switch of their own next to the label, and pickers for spaces, projects, work items, pages and people have an input mode for a reference.

Data inside a loop

A Loop step repeats the steps inside it once for each item of a list. Inside the loop, the loop step’s name followed by item is the current item, for example {{loop_1.item}}, and index is its position in the list, starting at 1. In the data picker, a group named after the loop and labelled current item offers both. The loop’s own result is the last item, the count and the number of iterations.

Formulas

For a value you have to calculate, such as joining text, changing a date or counting a list, select Formula next to the data picker. Pick a function from the groups Text, Number, Date, List and Logic, adjust it with data from earlier steps and select Insert formula. The dialog shows the Result with the sample data before you insert it. The preview uses the saved test data, so test the steps before it.

Data that is not part of a flow

Two more kinds of reference exist, and both read from the Automation workspace and not from steps:

Large results

A step result that is too large to keep in the run’s record is stored as a file next to it. The run page shows a download instead of the content, and later steps still read the full result.
Last modified on November 5, 2026