Case steps
Open or reuse a case, add notes, set severity and status, tag and assign it from a SOAR flow, with no pack or connection.
Case steps work cases natively, with no pack or connection. A flow can open a case, tag it, set its severity and status, add a note that carries a step result, and assign it.
Every case step runs as the flow's owner, the person who saved it, and goes through Cases exactly as that person's own change would. That means the same lifecycle rules, timeline entries, notifications and audit. Each step needs the permission a person needs for the same change, and the flow owner must also hold cases:view and cases:comment. nano checks these permissions when the flow is saved or run manually, and again before each step runs.
Steps
In the flow editor, case steps are in the Cases group of the palette.
All steps except Create case work on the run's case. That is the case that triggered the run, or the one an earlier Create case step opened or reused. A step that has no case fails with case_missing.
| Step | Type | What it does | Needs |
|---|---|---|---|
| Create case | case.create | Opens a case, or reuses one (see Reuse). Sets the title, description and severity (default medium). | cases:create |
| Add case note | case.add_note | Adds a Markdown note to the case timeline, optionally internal. | cases:comment |
| Set case severity | case.set_severity | Sets a fixed severity. | cases:edit |
| Set case status | case.set_status | Moves the case to a fixed status. Closing needs a disposition. | cases:close to resolve or close, otherwise cases:edit |
| Add case tag / Remove case tag | case.add_tag / case.remove_tag | Adds or removes one tag. | cases:edit |
| Assign case | case.assign | Assigns the case to a user, a group, or both. | cases:assign |
Reuse and repeats
Create case never opens a second case for the same trigger event. It uses, in order:
- the run's own case;
- the case the triggering alert is already in;
- the open case automation already opened for this alert, case or hunt escalation. Every flow that the same event triggers shares it.
Manual and scheduled runs open one case per run. The step output says "status": "created" or "reused", with case_id and case_number.
Every case step is safe to repeat. If a run is interrupted in the middle of a step, the step runs again without making a duplicate. Setting a status the case already has changes nothing, and the output says "changed": false.
Text and placeholders
The title, description and note accept:
- the notification placeholders, such as
{{alert.rule_name}},{{case.case_number}}and{{run.id}}; {{bind.<name>}}for explicit bindings;{{results.<step id>.<path>}}for an earlier step's value. For example,{{results.hunt.summary.count}}after a search step with IDhunt.
A value must be a short single value. A whole object, such as search rows, fails the step rather than being pasted in.
Severity, status, disposition, tags and the assignee are always fixed in the flow. Run data can never choose them.
Restricted sources
Case text built from run data is refused with case_source_scope_refused when the run read a restricted source, audit events, or data of unknown origin. Case text means the title, description or note. Run data means placeholders, bindings or step results. Fixed text is always allowed.
Status rules
The statuses are open, in_progress, pending, resolved and closed. Closing records a disposition: true_positive, false_positive, benign or inconclusive. pending takes an optional pending_kind.
Cases decides whether a move is allowed from the current status. For example, a closed case must be reopened before it can be resolved. A refused move fails the step with case_status_transition_invalid.
Tags
A tag is 1 to 48 characters: lowercase letters, digits, _, ., : or -, starting with a letter or digit. A case holds at most 32 tags.
Tags show in the case header, where you can also add and remove them. Through the API, use GET and POST /api/cases/{id}/tags and DELETE /api/cases/{id}/tags/{tag}.
Assignment
The user must be active, hold cases:view and be able to see the case. Otherwise the step fails with case_assignee_not_permitted or case_assignee_cannot_view. The group must exist, or the step fails with case_group_unknown.
Approvals
Case steps follow the flow's approval policy:
- By default they run without a person.
- A run triggered by AI always waits for a person. That covers a hunt escalation, a hunt-finding alert, the AI triage verdict, a case the AI opened (or one seeded from a hunt finding) and a status change the AI made. So does a run requested by AI or started by an API key, and any case step after a Judge step.
- Set
"approval": "required"on a step to always ask.
The approver sees the exact change: the step, its values, the rendered text, and the case. For Create case, the approver sees whether the step opens a new case or which existing case it uses. If the change is different when the run resumes, the step fails with case_change_changed.
Test runs
A test run never changes a case. Case steps record what they would do, and their outputs are placeholders. A later step that reads them is skipped and reported as depending on a placeholder. See Test a flow.
Failure reasons
| Reason | Meaning |
|---|---|
case_missing | The run has no case. Add a Create case step first, or trigger the flow from a case. |
case_not_visible | The flow owner can't see the case. |
case_source_scope_refused | Case text built from run data was refused because the run read a restricted source. See Restricted sources. |
case_status_transition_invalid | Cases doesn't allow that status move from the current status. |
case_change_refused | Cases refused the change, for example because the case already has 32 tags. |
case_change_changed | The change is different from the one the approver approved. |
case_assignee_not_permitted, case_assignee_cannot_view | The user is inactive, lacks cases:view, or can't see the case. |
case_group_unknown | The group doesn't exist. |
case_render_invalid | A placeholder didn't resolve to a short single value. |
case_write_failed, cases_unavailable | The write didn't complete, or Cases isn't available to the executor. |
Every committed case step writes a soar_case_step audit event, plus the case's own audit events.
Example
Open a case on an alert, tag it, raise its severity, add a note with a search result, and assign it to a group:
{"summary": "Triage impossible travel", "value": {"modules": [
{"id": "hunt", "value": {"type": "search", "query": "actor.user.name = {{alert.entities.user}}", "window": "24h", "limit": 100}},
{"id": "open", "value": {"type": "case.create", "title": "{{alert.rule_name}}", "severity": "high"}},
{"id": "tag", "value": {"type": "case.add_tag", "tag": "impossible-travel"}},
{"id": "sev", "value": {"type": "case.set_severity", "severity": "critical"}},
{"id": "note", "value": {"type": "case.add_note", "body": "## Sign-ins\n{{results.hunt.summary.count}} sign-ins for this user in 24h."}},
{"id": "own", "value": {"type": "case.assign", "group": "group_01..."}}
]}}What's next
Actions and connections
Install signed action packs, connect them to your security tools with write-only credentials, run a single action, and control where SOAR may send requests.
Approvals
Make a person sign off before SOAR changes something: when a step waits for approval, who can decide, how long it waits, and what happens after.