Workflows
Draw a recurring task once on a canvas — one trigger, optional conditions, one or more actions — so the same steps run the same way each time.
Compliance work is full of small steps that are easy to miss: a policy falls due for review, evidence (the proof that a control works) nears its expiry date, a vendor's status changes and nobody is told. A workflow records what should happen in one of those cases, and its run history shows an auditor how the task actually gets done.
A workflow runs when its trigger fires, and only while the workflow is
Active. A Manual workflow runs when somebody
selects Run in the editor or Run Now in the row
menu, or when the run API is called. A schedule is started by a background
job at the time it names, Webhook received by a call to the
workflow's webhook address, and an event trigger — a status change on a
policy, risk, vendor or control, evidence nearing expiry, an incident
created, changing severity or flagged as significant — by that change
happening in Aegis.
Who uses it
-
Viewer and
Contributor can search the
list, open a workflow and read its runs. They see neither
ImportnorCreate Workflow, and the row menu offers onlyView. The editor then opens view-only: aView onlybadge sits beside the status, a notice readsYou can view this workflow but not change it. Ask a manager or administrator to edit it., and theNode Library,Save,Export,TestandRunare not shown. The name and status chip are plain text, andRun Detailsoffers noRe-run Workflow. - Manager and Admin can also create, import, duplicate, edit, validate, export, activate, run and delete.
Two switches sit above the roles: the WORKFLOWS module must be
provisioned, or the list and create pages show a not-found page instead; and the
Workflow Automation licence feature (Professional and above) must
be present, or the list stays readable but the create and import buttons are
gone.
What's on this screen
The header carries the title Workflows and the line
Automate GRC processes with custom workflows. On a Manager or Admin
account, Import and Create Workflow sit at its right;
the screenshot below was captured on a Contributor account, so that corner is
empty. Beneath it is a filter bar: a box reading
Search workflows..., a Search button, and a status
dropdown starting on All Statuses.
The table lists every workflow in the organisation, twenty to a page.
Name is a link with the description, shortened to one line, in grey
beneath it; Status is a coloured chip — Draft,
Active, Paused and Archived all appear on
the captured screen. Runs counts the runs so far and
Last Run gives the date of the latest; both stay at
0 and - until a workflow has run, as the two draft
rows on the captured screen show. Created and
Created By record the origin, and Actions, the last
column, holds a three-dot menu. Selecting a row anywhere else opens that
workflow in the editor. Below the table sit
Showing 1 to 20 of … workflows, then Previous,
Page 1 of … and Next. If nothing matches, the table
gives way to No workflows match your search criteria. — or to
No workflows created yet. Create your first workflow to get started.
when there are none at all.
Opening the workflow list
- Select the open-book icon in the top bar to open this guide inside Aegis, at the chapter for the screen you were on. The question mark beside it opens the help panel instead.
- Check the account name at the top right. Your role decides whether the create and import buttons appear at all.
-
In the left menu, open the
ADMINISTRATIONgroup and chooseWorkflows. The list at/automation/workflowsloads;/automationon its own redirects here. - If the menu crowds the table, drag the divider between menu and page to narrow it. The width is remembered for your next visit.
Finding a workflow
-
Type part of a name or description into
Search workflows.... The table reloads as you type, filtered to workflows whose name or description matches;Searchor Enter runs the same search and returns you to page one. Empty the box to bring the full list back. A link to/automation/workflows?search=…opens the list with the box already filled in and filtered. -
Pick a status in the dropdown, which applies as soon as you choose it, then
page through with
NextandPrevious.
What the statuses mean
| Status | Behaviour |
|---|---|
Draft |
Being built. It cannot be run; Run stays unavailable.
|
Active |
Ready to run. Run Now appears in the row menu, and
Run in the editor becomes available once the design has
no validation errors.
|
Paused |
Temporarily off. Neither Run nor
Run Now is offered.
|
Archived |
Retired but kept, with its history intact. Nothing moves back out again. |
Creating a workflow
These steps need Manager or Admin. Select Create Workflow at the
top right of the list. Creating is a full page, not a popup:
Create New Workflow opens at
/automation/workflows/new with a breadcrumb back to
Workflows. An account without the create permission that types the
address in is sent to /unauthorized instead.
-
Decide where to start.
Blank Workflow—Start with an empty canvas— comes preselected, outlined in navy underStart from Template (optional). Keep it to draw everything yourself. -
Or select one of the published templates beside it. Each card carries its
name, a category tag and a description, and the chosen card takes the navy
outline.
Data Loss Incident Follow-up(tagData Loss) opens a follow-up ticket in Jira or ServiceNow when a Data Loss incident is created, and records the ticket's key and link on the incident. Its trigger is limited to incidents whose source system isPurview, and itsCreate ticketstep is set to Jira until you change it. It needs a working Jira or ServiceNow connection — see Connectors. A re-run never opens a second ticket for the same incident, and a failed hand-off is reported on the incident rather than blocking it. -
The cards below cover GDPR flows —
GDPR Breach Response,GDPR Consent Withdrawal Cascade,GDPR DSR Lifecycle,GDPR HR Offboarding ErasureandGDPR Retention Enforcement— and sector templates such asNL Healthcare Incident Reporting (AP, Z-CERT, IGJ), described in Incident reporting. -
NIS2 Incident Notification(tagNIS2) prepares the NIS2 reporting steps for an incident flagged as significant — see the callout under "Building the logic" below. On the captured screen its card names the CCB, because the demo organisation is Belgian.Procedure Review Remindersits beside it. Choosing any template fills in the name and description below; choosing a different one afterwards replaces them, and going back toBlank Workflowclears them. Text you typed yourself is kept, each field judged on its own. With no templates published, this section is absent.
Scroll down to Workflow basics to finish the form.
-
Fill in
Name(the box suggestse.g., Policy Review Reminder), the only required field — leaving it blank brings upName is requiredwhen you submit. A name may be up to 200 characters. -
Add a
Descriptionto help colleagues recognise the workflow in the list. It is optional. -
Select
Create Workflow(it readsCreating…while it saves). Aegis stores the workflow inDraftand opens the editor.Cancelbeside it returns to the list without saving.
Inside the editor
Across the top is the toolbar: a back arrow to Back to Workflows,
the workflow name (select it to rename it in place), the status chip, and an
indicator reading Saving... or Unsaved changes. At its
right sit Save, Export, Test and
Run. While the design is incomplete, a red count such as
2 errors appears beside them; select it to read the
Validation Errors list. A name typed in place may be up to 200
characters. On a Viewer or Contributor account the editor is view-only, as
described under "Who uses it" above.
Down the left is Node Library — "Drag nodes onto the canvas, or
click one or press Enter to add it to your workflow." — with three collapsible
groups: Triggers, Actions and Logic. The
canvas fills the middle. The right panel follows your selection: with a node
selected it is Node Configuration; with none it shows
Workflow Details (description, node and connection counts,
Run timeout (minutes), last updated) above
Recent Runs. Each side panel has its own hide control —
Hide node library, Hide details panel — and a matching
show control to bring it back. On a screen narrower than 1,536 pixels the
details panel lies over the canvas and starts hidden, so the canvas keeps a
workable width on a laptop.
Building the logic
-
Open
Triggersand drag one onto the canvas — or select it, or focus it and press Enter:Manual, the two schedules,Webhook received, an entity status change (policy, risk, vendor, control, procedure, DSR or DPA),Evidence expiring soon,Framework enabledorFramework disabled, one of the incident triggers (Incident created,Incident severity changed,Incident flagged significant,Breach confirmed), or another record event such asDSR created,Consent withdrawnorEmployee offboarded. A node card appears where you dropped it. Each workflow needs exactly one trigger. -
Select that node. The right panel becomes
Node ConfigurationwithLabel,Type,IDand aTrigger Typedropdown, followed by the fields that type needs — such asCron expressionandTimezone, orDays before expiry. -
Drag an action across from
Actions:Send Email,Send Webhook,Send Teams Message,Send Notification; a status-setter for a policy, risk, vendor, control, procedure, DSR or incident; orCreate Task(an action item with an owner and a due date, linked to the record),Create ticket(a follow-up ticket for the incident in Jira or ServiceNow, withTicket system,Summary, and for Jira an optionalProject key (Jira)andIssue type (Jira)),Create DSR,Assign Owner,Create Audit Log,Enqueue JobandApproval Gate. Configure it from itsAction Typedropdown and the fields below. - Join them: drag from the handle on the right of the trigger to the one on the left of the action. A line is drawn and the connection count goes up.
-
Add
ConditionorDelayfromLogicwhere you need a branch or a pause. A condition has two outgoing handles, yes and no, and both must lead somewhere before the workflow validates. -
To remove a node, use the delete control on the node or in
Node Configuration; both askDelete Nodefirst. Delete or Backspace on a selected node skips the confirmation.
Severity at least
Incident created starts a run whenever an incident is created —
on the Incidents screen, through the API, by an auto-incident rule or by an
incident connector. Incident severity changed starts one when
an incident's severity actually changes. Incident created also
takes an optional Source system: filled in, only incidents from
that source start the workflow; left blank, any source does. Both take an
optional Severity at least threshold: left blank, any severity
matches, and High also matches a
Critical incident. For a severity change the threshold must be
crossed upwards — with "at least High", Medium → High or Low → Critical
starts the workflow, while High → Critical or a downgrade does not. The
previous severity is available to later nodes as
{{previousSeverity}}.
Incident flagged significant starts a run once, when an
incident is marked as a NIS2 significant incident. The
NIS2 Incident Notification template uses it: it confirms the
incident owner, then creates three tasks under Roadmap → Action items, due
on the incident's own NIS2 deadlines — the 24-hour early warning, the
72-hour notification (24 hours for a trust service provider whose trust
services are affected) and the one-month final report — and e-mails the
owner for each. To copy a CISO or DPO, add their address to the recipients
of the three reminder e-mails. To hand the incident to a fixed incident
manager, enter their user ID as the owner in
Assign incident owner; the tasks and e-mails then go to that
person. If your organisation's country is Belgium, the texts name the CCB,
Safeonweb@Work and CyFun instead of "your national CSIRT". The three tasks
hang directly off the trigger, so they are still created if the owner step
fails — for example because the chosen user is no longer active.
Edits are written about two seconds after you stop, with
Saving... in the toolbar meanwhile. Wait for that to clear
before closing the tab — a pending change is lost.
Text fields take placeholders in double braces, such as
{{title}} — a trigger field goes by its own name, with no
trigger. prefix (workflows saved with the old
trigger. form still resolve). Add a fallback after a vertical
bar, such as {{dueDate|not set}}, to print that text when the
value is empty. A value the run does not hold and that has no fallback stays
in the message as typed, so read the first real run rather than assuming it
resolved.
Testing before you activate
-
Select
Test(unavailable while the toolbar shows a red error count — clear those first). Aegis checks the structure and returnsValidation PassedorValidation Failedwith a count of errors and warnings. -
Read the list under
Errors. The usual ones areWorkflow must have exactly one TRIGGER node,Nodes not reachable from trigger, a condition missing a branch, and required fields left blank. -
Use the link beside an error to jump to the node it names, fix it, then
select
Re-run Test. Warnings are listed separately and do not block activation.
Test validates structure and configuration only. It sends no
email, calls no webhook and changes no record, and cannot prove your
condition matches real data. For that, activate the workflow and read the
first real run.
Activating, pausing and archiving
-
Select the status chip beside the workflow name. The menu lists only the
moves the current status allows —
Activate,Pause,Set to Draft,Archive. -
Choose
Activate. While validation errors remain it readsFix validation errors first; otherwise the chip turns green and the workflow becomes runnable. -
Select
Runin the toolbar to start an active workflow. On anything else it stays disabled and readsWorkflow must be active to run; on an active workflow whose design has validation errors it stays disabled and readsFix validation errors first.
The same check guards every way of starting a run. Run Now in the
row menu, Re-run Workflow in Run Details and the run
API all refuse a workflow whose stored design fails validation, and no run is
created.
Watching what ran
With no node selected, Recent Runs in the right panel lists each
run with Pending, Running, Completed,
Failed, Cancelled or Timed out, when it
was queued and how long it took (In progress... while it runs) — or
No runs yet. A failed run shows its reason in red on the card.
Refresh reloads the list, and Retry appears if it
could not be loaded. Select a run to open Run Details.
-
The status chip at the top right repeats the run's outcome —
Failedon the captured screen. The cross beside it closes the window. -
The body opens with
Run ID,Duration,Started,Completed,Trigger TypeandVersion, then anError Messagebox naming every step that failed. Below it,Node Executionslists each step by the label it has on the canvas and its type (TRIGGER,ACTION), with its own status —COMPLETED,FAILEDorSKIPPED— its start time and duration, and, underShow details, its input and output. This is where you learn which branch a condition took. A failed step carries its reason in a red box; scroll down for the steps below the fold, and forRun Contextat the end when the run carried data. -
Select
Re-run Workflowto trigger the workflow again after a fix. It readsTriggering...while the new run starts. Accounts without the right to run workflows do not see it. - Select
Closeto return to the editor.
On the captured run, the trigger
Exception lapsed (manual) completed and the step
Set policy to In Review failed with
Database error (P2025). A step that is recorded
SKIPPED says why in a grey box, as
Skipped: upstream step … failed with the failed step's identifier
in place of the dots: it was never run because a step it depends on failed.
When a step fails, only the steps downstream of it along the connections are
recorded
SKIPPED. Branches that do not depend on it still run, and the
run ends FAILED, naming every failed step. A
Send Email step is also recorded SKIPPED, with the
reason, when no e-mail delivery is configured for your organisation at all;
a send that is attempted and fails stays FAILED.
Row actions, export and import
The three-dot menu on each row offers Edit; Duplicate,
which opens a Draft copy named after the original with
(Copy) appended; Run Now, on active workflows only;
and Delete, which asks for confirmation. Only entries your role
allows appear; without the right to edit, Edit reads
View. Export in the editor toolbar offers the workflow
as JSON to Copy to Clipboard or Download. If the file
would not import again, an amber note above it says so — "This workflow cannot
be imported again until these problems are fixed. You can still download it as a
backup." — and lists the problems. Import on the list page takes
that file (Upload JSON File) or pasted JSON (Paste Workflow JSON), shows a Preview of its name, description, trigger type, nodes
and connections, and Import Workflow saves it as a new
Draft.
Wherever Aegis shows a workflow — the editor, the run details and the export
file — the secret parts of a Send Teams Message or
Send Webhook address, and credential headers such as
Authorization, are replaced by ••••••••. Leave a
masked value unchanged when you save and Aegis keeps the stored one; type a
complete new address to replace it. A masked value you have edited is
refused. Because an export file holds only the masked form, an import clears
each masked value to an empty field: the workflow still arrives as a
Draft, the editor flags every cleared field, and it cannot be
activated until you fill each one in with the real address or header.
Duplicate copies the stored values and is not affected.
Tips and limits
-
Run timeout (minutes)inWorkflow Detailscaps how long one run may be active, from 1 to 44,640 minutes (31 days). Time spent waiting in aDelayor for an approval does not count. Once the budget is spent the run stops — before its next step, or by cutting the running step short — and is markedTIMED_OUTwith the reason in its run details.Export,ImportandDuplicatekeep the timeout; a file exported before the timeout was included imports with the default of 30 minutes. -
Prefer
ArchivetoDeletewhen an auditor may ask later. An archived workflow keeps its runs; a deleted one leaves the list with no way back. - Where the order of several steps matters, use one workflow with branches — nothing sequences one workflow after another.
-
An active workflow with no runs at all has never been started: for a
Manualtrigger nobody has selectedRunorRun Now; for any other trigger its event has not happened yet. -
Send Teams Messageneeds an incoming-webhook address in itsWebhook URLfield; once saved it is shown masked. There is no Slack action in the library. -
Update Incident Statusfollows the same status rules as a person on the Incidents screen. A move those rules do not allow — reopening a closed incident, for example — or a status someone changed in the meantime fails the step with a reason startingRefused:; asking for the status the incident already has changes nothing. -
The editor has no AI assistance. A workflow does what a person drew on the
canvas, which is what makes it defensible in an audit. A step can hand work
to a background job — the
GDPR DSR Lifecycletemplate, for example, queues jobs that classify the request and draft a response before aDPO Approval Gate— and a person still reviews and decides.
Where this connects
- Policies, Procedures, Evidence, Risks and Vendors — the records a workflow's actions read and set.
- Connectors and Webhooks — how outbound mail and HTTP calls leave Aegis.
- Automatic incident rules — rule-based automation for incidents, separate from this canvas.
-
Incidents,
Data loss,
Jira and
ServiceNow — the incidents the
incident triggers watch, and where a
Create ticketstep sends its follow-up ticket. -
Action items — where
Create Taskputs the tasks a workflow creates. - Roles overview and Audit log — who may do what, and who changed a workflow.