Incidents

Get a security event on record the moment it is detected, work it from investigation to closure, and keep the regulators' notification clocks in view while you do.

An incident is something that has already happened — not a risk that might. This module records the event, moves it through a defined lifecycle while stamping the time at every step, counts down any notification deadline it triggers, and holds the post-incident write-up so the proof is there when an auditor asks. Every change goes to the audit trail.

This chapter is the screen-by-screen reference. To see a response worked as a story — detection, triage, regulator deadlines and the post-incident review — see the full walkthrough in Scenario: respond to a security incident.

Who uses it

The breach decisions sit with Manager and Admin on purpose: confirming a breach, recording that the authority was notified, or deciding not to report all start or stop the GDPR 72-hour clock, so they are decisions of record rather than routine edits. A Contributor sees the breach alerts but not those buttons.

What's on this screen

The header reads Incident Register over "Track and manage security incidents and data breaches", with the dark Report Incident button on the far right. Beneath it sits the filter bar: a Search incidents… box with its own Search button, then All Statuses, All Severities, All Types and Any breach status — the last offering Data Breaches Only or Non-Data Breaches. The status list also holds Open (not resolved), which shows every incident not yet resolved or closed; the dashboard's open-incidents tile lands you here with that filter already set. The filters combine, so each one narrows the list further.

The table has a tick-box column, then Title, Type, Status, Severity, Deadline, Detected, Owner and Actions. Status and severity are coloured chips, and the detected time is shown in your local time with its time zone, for example GMT+1. In the captured screen the rows mix one Open, one Investigating and then Closed incidents, and every visible row has a dash under Deadline — no outstanding notification obligation. When an incident does owe one, the cell shows a chip naming the regime (GDPR, NIS2 or DORA) with a countdown, or an overdue figure once the deadline has passed, and a marker such as +2 when further deadlines sit behind the nearest one. DORA deadlines are left out when DORA is switched off for your organisation.

A purple GDPR badge beside a title marks a personal-data breach and turns red as GDPR! once the 72-hour window passes with no notification recorded. A dashed amber Suspected badge next to it means the breach was flagged but no Manager has yet confirmed or dismissed it — the captured screen shows one such row, an insider-threat incident. Actions holds an eye icon; a bin icon joins it for Manager and Admin, so this Contributor view shows the eye alone. The page loads 20 rows at a time.

  1. Open the Security Incidents group in the left menu and choose Incidents. The register loads with the most recently detected incidents on top.
  2. Type a word from an incident's title into Search incidents… and select Search. The table reloads showing only matching titles. Narrow further with the four dropdowns; each applies on selection and stacks with the others.
  3. Tick the box at the start of one or more rows. A bar appears with Export, which downloads the selected incidents as a CSV file.
  4. Select the eye icon in Actions — or the title itself — to open the full incident page, where the AI panels, the significance wizard and the regulatory filings live. Select anywhere else on the row instead and the Incident Details dialog opens over the table.
  5. Read the badges beside a title. GDPR (or red GDPR!) marks a personal-data breach; Suspected means that breach still awaits a Manager's decision. Hover over either for its meaning.
The incident register — /incidents.
The incident register — /incidents.

Statuses and severity

Aegis offers only the valid next steps from the current status.

Status Meaning Can move to
OPEN Recorded; investigation not started. INVESTIGATING, CONTAINED, CLOSED
INVESTIGATING Being analysed; root cause not yet known. CONTAINED, RESOLVED, CLOSED
CONTAINED Stopped from spreading, not yet resolved. RESOLVED, INVESTIGATING, CLOSED
RESOLVED Root cause addressed; operations resumed. CLOSED, INVESTIGATING
CLOSED The record is complete. Nothing — closure is final.

Severity runs LOW, MEDIUM, HIGH, CRITICAL, set on reporting and revisable as the investigation develops.

Reporting an incident

Reporting now happens on a page of its own, /incidents/new, rather than in a dialog over the register. The form is still short on purpose: during a live response you want the event on record in under a minute and the detail filled in later.

  1. Before you report, search the register for the event. If a colleague or a connector rule has already recorded it, work that incident instead of opening a duplicate.
  2. Select Report Incident. The Report New Incident page opens, with a breadcrumb back to the Incident Register above it.
  3. When you finish the form (steps below), Aegis takes you straight to the new incident's own page. Back on the register it appears in this table in OPEN status.
Starting a report from the register — /incidents.
Starting a report from the register — /incidents.

The capture above shows the register at the moment you start; the form itself is on the next page and is not pictured. It has two cards — Incident Details and Detection & Impact — then the breach box and the buttons.

  1. Enter an Incident Title. Required — the form will not submit without it.
  2. Add a Description: what happened, how it was discovered, first observations. Optional, and it can wait.
  3. Choose an Incident Type and a Severity. They open on Other and Medium - Moderate impact; both are required, so change them if those defaults are wrong.
  4. Fill When was the incident detected?. The note beneath says this timestamp drives the GDPR 72-hour calculation, and Aegis rejects a time in the future with "Detection time cannot be in the future".
  5. Optionally list Affected Systems as a comma-separated line. Each entry becomes a tag on the incident.
  6. If personal data may have been accessed, disclosed or lost, tick This incident involves a personal data breach. A Number of Data Subjects Affected field appears. The note under the box explains that the incident is recorded as a suspected breach until a Manager confirms it, and that the 72-hour deadline runs provisionally from detection in the meantime.
  7. Select Report Incident. The button stays greyed out until the title and detected time are valid. Once selected, a confirmation appears and the new incident's page opens.
  8. Select Cancel to discard everything you typed and return to the register.
The detected time drives every clock

The GDPR 72-hour window and the three NIS2 stages run from the detected time you enter here, not from the moment you log the incident. If you later learn the event started earlier, correct it.

Review security signals from connectors

Security alerts from a connector do not become incidents on their own. This covers Microsoft Defender XDR (Defender incidents with their related alerts already grouped), CrowdStrike Falcon detections, AWS GuardDuty findings, GitHub secret-scanning alerts and high or critical Palo Alto threat logs. Each sync puts new alerts in Security Signals (under Security Incidents in the sidebar) as pending signals. A person decides every one; nothing is created or discarded automatically. This keeps a flood of alerts, such as dozens of email alerts from one phishing campaign, out of the incident register while still keeping each alert on record.

Each pending signal shows a suggestion with its reason. It is advice, not a decision: Dismiss when the source system already resolved, closed or archived the alert (for example Defender removing a malicious email after delivery), marked it a false positive or expected activity, or rated it informational; Merge when an open incident from the same source has the same title; Keep for anything else.

With AI assistance switched on, Aegis also asks the AI to triage each batch of new signals after a sync (one AI credit per batch). Its suggestion appears in a purple box under the rule suggestion, marked as AI-generated, with a short reason and a confidence percentage. It is advice only: the AI never creates, merges or dismisses a signal. If AI is off or out of credits, the signals keep only the rule suggestion. Merge as suggested uses the rule's incident first and the AI's when the rule has none.

  1. Open Security Signals. The list opens on Pending; filter by Suggestion to work through similar alerts together.
  2. Select one or more signals and choose an action: Create incidents (one incident per signal), Combine into one incident (one incident from the most severe signal, the others merged into it), Merge as suggested or Dismiss. Dismiss asks for confirmation.
  3. Check the result message. A signal someone else decided in the meantime is skipped, never decided twice. Switch the Status filter to see accepted, merged and dismissed signals and open the incident each one belongs to.

Creating or merging adds an entry to the incident's timeline and records the signal as evidence linked to that incident. Alerts are no longer copied into the Evidence Locker as they arrive: only a signal someone kept becomes evidence, so the locker holds triaged alerts rather than every raw one. A dismissed signal stays on record in Security Signals only. Deciding needs the permission to edit incidents; creating an incident also needs the permission to create one.

Reading an incident

The Details grid and the Timeline always appear; the Description, Affected Systems, Root Cause, Remediation and Lessons Learned panels show only once their field holds something. The incident in the captured screen is open, high severity, with a description and two affected systems, and dashes under Contained, Resolved and Closed because it has reached none of those states yet. Because it is still open, the Update status panel is offered; on a closed incident that panel disappears — there is nothing left to change.

  1. Select a row in the register. The dialog opens at the top of the record — title, status chip, severity chip, then the type and detected line.
  2. If you opened the wrong row, select the × in the dialog header. Nothing is changed and you are back on the register.
  3. Use the Update status panel to move the incident on. It holds one button per valid next step — here Mark Investigating, Mark Contained and Mark Closed, the three moves open from OPEN. The callout marks Mark Closed, which opens a confirmation first (see Closing an incident below).
  4. Read the Description, then the Details grid for the owner and five lifecycle timestamps (Detected, Reported, Contained, Resolved, Closed). A dash means the incident never reached that state. Scroll on through Affected Systems, Root Cause, Remediation, Lessons Learned and the Timeline, where Contributors and above get an Add Entry control. A short dialog means a thin record, not a fault.
  5. Select Close at the foot of the dialog. Anything you changed was saved as you went.
An incident's detail dialog — /incidents.
An incident's detail dialog — /incidents.

For a personal-data breach the dialog also shows the GDPR alert under the header, and a Manager or Admin sees a Record GDPR Notification button there until the notification is recorded.

Moving an incident through its statuses

While an incident is still open, the Update status panel sits between the header and the description with one button per valid next step. Viewers do not see it at all. The same controls sit beside the title on the full incident page.

  1. Select the step you want, for example Mark Investigating. The status chip changes, the matching timestamp is stamped and the timeline gains an entry. If someone else changed the status a moment before you, Aegis refuses the change and says so rather than silently doing nothing.
  2. Carry on through Mark Contained and Mark Resolved. From either you can step back to Mark Investigating if new facts reopen the analysis.
  3. Select Add Entry in the timeline to record what you found or did. The entry carries your name and the time.

Closing an incident

Mark Closed is offered from every status before closure, not only from RESOLVED, and it opens a confirmation dialog rather than closing on the spot.

  1. Select Mark Closed. The Close Incident dialog names the incident and states that this marks it resolved and will set the closed timestamp.
  2. Optionally fill Resolution Summary (Optional), saved as the incident's Remediation, and Lessons Learned (Optional), saved as its Lessons Learned. Text already stored in either field is shown so you can keep or extend it. If someone else changed the incident in the meantime, the dialog says so and keeps your text.
  3. Select Close Incident. The status becomes CLOSED, the closure time is recorded and the register refreshes.
Closing cannot be undone

Aegis offers no transition out of CLOSED, so a closed incident cannot be reopened. Check that root cause and remediation are recorded first; afterwards the only way to add anything is a timeline entry.

The full incident page

The eye icon or the title opens /incidents/<id>. A breadcrumb leads back to the Incident Register; any GDPR or NIS2 alert banner sits above the title, and the status buttons sit to its right. The main column holds the description, affected systems, root cause and remediation, the timeline, the regulatory filings, the authorities' acknowledgements, communications and response playbooks. The right-hand column holds the details, a Jira link panel, the reporting decision, the DORA classification (when DORA is switched on for your organisation) and the two AI panels. For a personal-data breach a 72-hour countdown runs across the page under the header.

Response Playbooks lists any remediation playbook started for this incident, with its status and a percentage complete, each linking through to the playbook itself. A Contributor or above can pick a Playbook template — Incident Response is the default — or type a Custom playbook title, then select Start playbook. Until one is started the panel reads "No playbook has been started for this incident yet." See Playbooks for working the steps.

An incident raised from a Microsoft Purview data-loss alert also carries a Data Loss badge beside its title and a data-loss panel in the main column, shown only to people with access to data-loss monitoring.

Regulatory deadlines

Aegis counts every clock from the detected time. Two European regimes drive the flags you set here; the Deadline column also counts DORA stages, but only for an incident classified as a major ICT incident.

GDPR (the EU's data-protection law) requires a personal-data breach to be notified to the supervisory authority within 72 hours. Flag the incident as a data breach and an alert shows whether that window is open, closing or overdue.

Confirming or dismissing a suspected breach

A breach flagged when the incident is reported — on the form, from Slack or by a connector — starts as a suspected breach (flagged, but not yet confirmed). The alert reads Suspected data breach with a Provisional badge, and explains that the 72-hour clock runs from detection as a precaution until someone decides. A Manager or Admin then chooses:

Once you have notified the authority — inside Aegis or outside it — a Manager or Admin selects Record GDPR Notification to stamp the time. That clears the red GDPR! badge and puts the notification in the audit trail. A recorded decision not to report also stops the deadline counting as overdue.

NIS2 (the EU cybersecurity law) applies only once an incident has been judged significant. Three stages then run:

Stage Deadline
Early warning 24 hours after detection
Incident notification 72 hours after detection
Final report 1 month after the incident notification

To make that judgement, open the full incident page and select Assess Significance in the AI Analysis panel — or Open Full Assessment if a reading is already there. The NIS2 Significance Assessment wizard opens over the page. Step 1, Automatic Checks, runs four tests against the record — high or critical severity, the data-breach flag, at least one affected system, a high-risk incident type — and counts how many are met. Step 2, AI Assessment, shows the AI reading if one has been run, and offers Run AI Assessment or Skip to Decision if not. Step 3, Decision, asks you to choose Mark as Significant, which activates the deadlines above, or Not Significant, which records the judgement without them. Neither takes effect until you select Confirm.

Filing, deciding and telling people

Three further panels sit on the full incident page. Regulatory Notifications appears once an incident is a data breach or marked significant, holding the filings it triggers — GBA — Personal data breach notification (Belgian Data Protection Authority, GDPR Article 33) and CCB — NIS2 significant incident report (Centre for Cybersecurity Belgium). Select Generate draft, review it, then Mark as filed and enter the reference the authority gave you. For a Dutch care provider the section also lists Z-CERT and the IGJ calamity duty; Incident reporting explains when they appear.

Reporting Decision holds the judgement itself, separate from the filing: it reads Pending Decision until a Manager or Admin selects Record Decision and chooses Report or Do Not Report with a reason. It then shows the decision, who made it and when — what a regulator asks for when a breach was assessed and not reported. The Recorded: line gives that moment in your local time with its time zone, for example GMT+2, as do the GDPR breach milestones and the recorded authority acknowledgements, so a reader in another zone does not misread a deadline. Update Decision revises it later. Communications records what you told people; external messages pass through Create draft, Approve and Send, so nothing leaves without a second pair of eyes.

Aegis never files on your behalf

The panel says so itself: drafts are for review only, Aegis transmits nothing to an authority, and the field mappings need sign-off from your data protection officer or legal team. File through the authority's own channel, then record the reference here.

The AI assist

The full incident page carries two AI panels. AI Analysis offers Run AI Assessment, weighing the incident against the NIS2 criteria and returning a Likely Significant or Not Significant reading with a confidence figure, per-criterion reasoning and the frameworks it thinks apply. Once an incident is marked significant, the same panel adds a button that drafts each notification stage — Draft Early Warning, Draft Notification, Draft Intermediate and Draft Final Report. AI Triage holds one button, Deep AI Triage, which runs a longer pass — significance under NIS2, GDPR and DORA, whether this looks like a personal-data breach, the notification stages and their deadlines, related risks, affected assets and relevant controls, and prioritised response actions. The panel says it takes 30 to 90 seconds.

The panel states it plainly, and so do we: these are recommendations drawn from your own records, and a person must confirm them. Neither panel changes a status, marks an incident significant, files a notification or tells anyone. Both need the AI feature enabled for your organisation, and both are write actions — a Viewer can read a finished reading but cannot run one.

Tips and limits

Where this connects

Deadlines across every incident are tracked in Incident reporting. Rules that raise incidents from connector events live in Auto-incident rules. For the personal-data side of a breach see GDPR; to fire a notification or task when an incident opens see Workflows, and to work a structured response see Playbooks. A recurring incident often belongs in the risk register too, and every change shows in the audit log. For the journey end to end, follow the incident scenario.