Step Outputs and Variables
Besides alert values, a step can use two other kinds of data: the output of an earlier enrichment step in the same run, and stored variables such as tokens, host names or passwords. This page explains both, and how secret variables are protected.
Outputs of earlier steps
Enrichment steps (HTTP call and LLM enrichment) return data that later steps on the same path can insert. Write the step identifier, then the path inside its output:
$(<step id>.<path>)Examples:
| Earlier step | Placeholder | Holds |
|---|---|---|
LLM enrichment step_2 | $(step_2.result) | The AI answer. |
HTTP call step_1 whose answer is {"data":{"verdict":"malicious"}} | $(step_1.data.verdict) | malicious |
How it works:
Only enrichment steps expose an output. Notification, incident, email and conditional steps do not, and neither do shell steps unless you make them enrichment steps in the Code tab (see below).
A step sees the outputs of the enrichment steps above it on its own path (its parents and their parents).
The Insert field menu in the step editor lists the alert fields and the outputs of the enrichment steps above the step. If there are none, it shows "No enrichment ancestors yet — add an enrichment node upstream to expose its output here."
Outputs are quoted when inserted, the same way as alert values. See Placeholders.
A Conditional step can check an output directly, for example field
step_1.data.verdict, operator is, valuemalicious.In the Insert field menu, "whole output (edit the path)" inserts
$(<step id>.); complete the path by hand when the menu does not list the field you need.
Shell output as data (Code tab only)
In the Code tab you can change a Shell step's kind from executor to enrichment. Its output then becomes data for later steps, like an HTTP answer. The command must print one JSON document; any other output fails the step with soar shell enrichment: stdout is not JSON. A command that prints nothing also fails, because the agent then reports the text "Command executed successfully but no output".
Stored variables
Variables keep values you do not want to type into every flow, for example an API token, a firewall address or the path of an SSH key. Write:
$[variables.NAME]For example, a shell step on a jump host:
ssh -i $[variables.FW_SSH_KEY] admin@$[variables.FW_HOST] sudo iptables -I FORWARD -s $(alert.adversary.ip) -j DROPVariables are filled in when the step runs and are quoted the same way as alert values (see Placeholders). Do not put quotes around them.
A reference to a variable that does not exist (for example a typing mistake in the name) is left as written, and the command is sent with the literal text $[variables.NAME]. Names are case-sensitive. If a secret variable cannot be read, the Interactive Console refuses the command with "a variable in this command could not be resolved".
Manage variables
There is no separate variables page in UTMStack v12. Open the Variables button in one of these places:
In the flow editor, in the settings of a Shell step.
In SOAR > Interactive Console.
The panel is titled Command variables and shows how to reference each one ("Reference these in commands as").
| Field | Notes |
|---|---|
| Name | Required. Unique within your tenant. |
| Value | Required for a new variable. |
| Description | What the value is for. |
| Secret (encrypted, masked in output) | Stores the value encrypted and hides it in output. |
Click New variable to add one. Edit, Delete and Insert (puts the reference at the cursor) are on each row.
Who can manage variables: users with soar.write (Administrator, Analyst). Users with only soar.read can list them.
Variables belong to a tenant. Each tenant has its own set.
Secret variables
A secret value is stored encrypted and is shown masked in the list.
When you edit a secret, leave Value blank to keep the stored value ("Leave blank to keep the stored value.").
To turn Secret off, you must type a new value. The stored secret is never revealed.
In Execution history, commands show
****in place of the secret, and step output has the exact secret value replaced by****. The same masking applies to Interactive Console output.
Masking replaces the exact stored value only. If a command changes the value (for example prints it encoded or split in parts), the changed text is not masked. Do not write commands that print secrets.
Changed from UTMStack v11: v11 refused commands that printed a secret with echo, cat and similar commands. UTMStack v12 does not check commands; it masks the output instead.
Related pages
Placeholders
Step Types
Interactive Console