Skip to content

Incidents

An incident is a group of related events that Identiguard has linked into one story. Incidents are opened automatically from the events already arriving. You do not create them by hand.

The page title is Incidents, with the description “Correlated attack stories, opened automatically from the events already being ingested.”

Two things must both be true before a workspace has incidents:

  1. The workspace is entitled. Automated Identity Incidents is included in the Super plan. On Pro and Max it is the Automated Identity Incidents add-on. See Plans and billing.
  2. An admin has left Enable automatic incidents for this workspace ticked under Incident settings. It is ticked by default.

When the workspace is not entitled, or an admin has switched automatic incidents off:

  • The Incidents row is hidden from the sidebar.
  • The Needs attention strip is hidden from the dashboard.
  • A direct link to the incidents list or to one incident shows an empty state with the title Automatic incidents are switched off. The text reads: “Automated Identity Incidents is included in Super and available as an add-on on Pro and Max. When it is on, a workspace admin can switch it, or individual detectors, under Incident settings.” The button Open Incident settings takes you there.

Events are still stored and classified. Nothing is correlated and no detector runs.

To switch it back on, see Switch incidents or one detector on or off.

Action Viewer Member Admin
Read the list, an incident, the side panels and the Silenced tab Yes Yes Yes
Use Ask Alex AI on an incident Yes Yes Yes
Close incident, including ticking the silence box No Yes Yes
Reopen incident No Yes Yes
This was real on a silenced incident No Yes Yes
Undo on the silencing message No Yes Yes
Open Incident settings: master switch, detector switches, learned dismissals, Disarm, correlation groups No No Yes
Receive the email sent when a silence rule is armed No No Yes

A viewer sees the text “Viewers can read incidents but not close them” where the close button would be.

Open Incidents in the sidebar.

Tab What it lists The number beside it
Open Open incidents that are not silenced How many are open
Escalated Escalated incidents that are not silenced How many are escalated
Closed Every closed incident that has not yet been removed Only those closed in the last 7 days. Hover the tab to see this. The list behind the tab is not limited to 7 days.
Silenced Incidents hidden by a silence rule, in any status How many are still active
All Open, escalated and closed incidents. It does not include silenced incidents. No number

Silenced incidents appear only on the Silenced tab.

If the counts cannot be loaded, the tabs show no numbers. The list still works.

  • Domain dropdown: All domains, Workforce (Okta and Entra ID) or Customer (Auth0).
  • Tenant dropdown: All tenants, or one tenant.

Changing a tab or a filter returns you to page 1.

There is no search box, sort control, bulk action or export on this page.

  • A count line: “1 incident” or “N incidents”.
  • A change line when the list has changed since you loaded it, for example “+2 new” or “1 no longer matches”.
  • A status dot. Live with a green dot means the page updates on its own. Offline with a grey dot means it does not. The list still works when offline. Reload the page to see new incidents.

On page 1, a new incident appears at the top of the list with a brief highlight. On later pages the list reloads.

Under the tabs, when at least one incident is silenced, you see a line with a dotted underline: “1 incident silenced by a learned rule” or “N incidents silenced by learned rules”. Select it to go to the Silenced tab.

On the Silenced tab the line reads: “Event alerts are not affected by silencing. Rules are managed in Incident settings.”

The list shows 10 incidents per page. The page control is hidden when there is only one page.

The tab and the page number are kept in the address, for example ?status=escalated&page=2. You can share the link to a view. The values for status are open, escalated, closed, silenced and all. An unknown value shows the Open tab, page 1.

The whole row is one link to the incident.

Part Meaning
Colour bar on the left Severity. Red for CRITICAL and ERROR, amber for WARN, blue for INFO, grey otherwise. Hover to read “Severity: X”.
Title Generated from the stages: “First stage on entity”, “First stage then second stage on entity”, or “Activity on entity”
Score N The incident score, from 0 to 100. See How the score is built.
Status pill Open, Escalated or Closed
Silenced chip Shown when a silence rule is hiding the incident. Hover to read the rule.
Entity and kind Who or what the incident is about. The kind is Identity, IP address or Tenant.
Tenant names The tenants the events came from
Age “just now”, “N min ago”, “N h ago”, “N d ago”, then a date
Eight pills on the right The attack story in small. A tinted pill is a stage the incident has reached. Hover a pill for the stage name.
N events How many events are attached
What you see Meaning
No incidents, with a sentence such as “12 of 22 detectors are evaluating across 2 tenants.” Nothing matches the tab and filters
Nothing is silenced No incident is silenced. The text explains: “When the same kind of incident is closed as a false positive three times, or once with Silence ticked, new ones like it are listed here instead of on Open.”
Could not load incidents with a Retry button The list could not be loaded. Nothing is lost.

On a wide screen the list has three panels on the right. They are hidden on narrow screens and phones.

This panel lists up to 10 identities that currently carry risk. It covers the whole workspace. It ignores the tab and filters you have chosen.

  • The subtitle reads, for example, “2 escalated, 5 open, 3 closed in the last 7 days”.
  • Each row shows the name, “N open incidents”, and “ - privileged” when the identity has used a privileged permission.
  • The badge is a risk score. It is red at 60 or above and amber below.
  • The rows are not links.

The risk score rises with the incidents against an identity and is higher for privileged identities. It falls day by day when nothing new happens. The app does not show the calculation.

If nothing is open against any identity, the panel reads “No identity is currently carrying risk.”

This panel is closed until you select it. It is where you see which detectors are working.

The summary line reads “N of M detectors armed.” It may add “Detections on incidents open now or closed in the last N days.”

When you open it you see:

  • A row of small squares, one for each detector. Hover a square, or move to it with the keyboard, to read its title, description, state, recent count and any hint.
  • A list of every detector that applies, each with a state chip, a description and a detections count.

The squares move slowly only while the page is Live.

State chip Meaning
Armed At least one tenant sends what the detector needs. It is working.
Blocked No tenant can feed it. Either a needed event type or Entra ID stream is not subscribed, or an Auth0 log stream masks identities. A hint says which.
Switched off An admin has switched it off. The hint reads “Switched off for this workspace.” with the link Turn on in settings.
Not applicable The detector does not apply here
Fired Shown only inside an incident: this detector matched on this incident

Detectors that can never apply, because the workspace has no tenant of the right platform, are left out of the panel. A footnote says how many, for example “3 detectors not shown: they read telemetry no connected tenant sends”.

The panel follows the tenant and domain filters of the page.

A Blocked detector shows one of these hints:

Hint Meaning
“Subscribe tenant to event type to evaluate detector” An Okta or Auth0 tenant does not subscribe to the event type
“Subscribe tenant to any of a, b, c to evaluate detector” The detector needs at least one of several event types
“Enable the name poll target on tenant to evaluate detector” An Entra ID tenant does not subscribe to that stream
“Masking hides identities on tenant. Set Obscure log stream data to XXHash to evaluate detector” The Auth0 log stream masks identities

This panel is also closed until you select it. It holds one sentence, for example:

“18 of 22 detectors are evaluating across 2 tenants. 3 are blocked by missing telemetry. 1 is switched off for this workspace.”

It may also say “N are blocked by masked identities.”

When an Auth0 tenant hashes identities with XXHash, the panel adds: “Identities on tenant are hashed by the Auth0 log stream. Detection is unaffected, but users appear as pseudonyms rather than names.”

In short, for Auth0: masking blocks the identity detectors. XXHash does not. With XXHash the detectors work and users appear as pseudonyms.

An open incident becomes escalated when any of these is true:

  • A detector hit is recorded on it. This includes the detectors marked “adds to an incident”.
  • It contains a CRITICAL event.
  • It has reached a sensitive stage and also has a risk signal reported by the provider, or five or more credential-access events. The sensitive stages are MFA change, Privilege grant, Persistence, Policy change and Account change. A baseline anomaly does not count as a provider risk signal.

Select a row to open the incident. The page updates on its own while you read.

Back to incidents returns you to the same tab, filters and page you came from.

  • The generated title.
  • A line with the entity kind, the entity and the tenant names, for example “Identity · ana@example.com · Okta production”.
  • Ask Alex AI: every role, viewers included.
  • Close incident or Reopen incident: members and admins.
Chip Meaning
Severity The highest severity among the events
Status Open, Escalated or Closed
Score N Select it to open How this score is built
Opened When the incident opened
Last activity When the newest event was attached
Closed as Resolved, Closed as Benign activity, Closed as False positive, Closed automatically (quiet) On a closed incident, how it was closed

Select Score N. The panel How this score is built lists each part with the points it added.

Line in the panel Points
Highest event severity DEBUG or INFO 10, WARN 20, ERROR 40, CRITICAL 60
How many attack stages it spans 10 for each stage after the first, up to 30
Detector hits 10 for each detector hit, up to 20
Seen across more than one tenant 5
The identity holds privileged access 5

The score cannot go above 100.

Example: the highest event is ERROR (40), the incident spans three stages (20), one detector fired (10), and the identity is privileged (5). The score is 75.

Eight numbered cells in a fixed order. A tinted cell is a stage the incident has reached, with the time it was first seen.

The stages, in order: Risk signal, Credential access, Sign-in, MFA change, Privilege grant, Persistence, Policy change and Account change.

  • A row of detector squares for the incident’s own domain. Filled squares are detectors that fired on this incident. Hover the others to see which were armed but did not fire.
  • One card for each detector hit: the title, the numbers that made it match, “N evidence events” and an Ask Alex AI button.
  • One card for each baseline anomaly: the badge new for this identity, a title, the values, the time and an Ask Alex AI button.

If nothing fired, the text reads: “No detector or baseline hits. This incident is built from vendor signals and direct evidence.” This is normal for an incident opened from a provider risk signal.

The events attached to the incident, oldest first, 10 per page.

Each row shows the time, a severity badge, the event type, the message, the actor, the IP address with its country, the tenant and the system the event came from. A row may also carry these chips:

Chip Meaning
Trigger The event that made a detector fire
Evidence An event a detector counted as supporting evidence
Baseline anomaly The event was unusual for this identity

An incident stores at most 500 events. Two notes can appear:

  • “N routine events not stored (cap reached)”: the incident reached 500 stored events. Later routine events were counted but not stored.
  • “Newest N of M events attached”: more events are attached than the page can return. You see the newest 500.

A row reads “Raw event no longer stored” when the event is older than the workspace’s event retention. The incident keeps its own summary of the event.

Shown only when there are any. These incidents share an identity or an address with this one. Select one to open it.

There are three places to ask on an incident. Each opens the Ask Alex panel and sends a question for you.

Where What it asks
Ask Alex AI in the header To explain the incident: what happened, why it opened, how the score was reached and what to do next
Ask Alex AI on a detector card Why that detector fired, which events it grouped and whether it is a real risk here
Ask Alex AI on a baseline anomaly card The same, for that anomaly

Follow-up questions in the panel stay on this incident. See Ask Alex.

In Event Details on the dashboard, View incident opens the incident an event belongs to. If the event has no incident you see the message “That event is not part of an incident”.

Identiguard learns what is normal for each identity. Once an identity has at least 20 events and 14 days of history, it flags:

  • First activity from a new country
  • A privileged identity from a new IP address
  • A sensitive change outside usual hours
  • A dormant identity active again, after 45 or more quiet days
  • The first privileged action by this identity
  • An identity appearing in another tenant

Members and admins can close an incident.

  1. Choose Close incident.
  2. In the dialog Close incident, choose a reason: Resolved (the default), Benign activity or False positive.
  3. Choose Close incident. You see the message “Incident closed”.
Reason Use it when
Resolved It was real and has been dealt with
Benign activity It happened, and it was legitimate
False positive The detection was wrong

There is no undo for a close. To reverse it, reopen the incident.

  1. Open the closed incident from the Closed tab.
  2. Choose Reopen incident. You see the message “Incident reopened”.

The incident returns to the status it had before it was closed, Open or Escalated. The close reason is cleared.

Silencing is for an incident that keeps opening on one tenant for a known, harmless reason.

A silenced incident still opens. It still collects evidence and can still escalate. It is listed on the Silenced tab and kept off Open, Escalated, Closed and All. Event alerts by email, Teams and other channels are not affected.

There are two ways a silence rule is armed:

  • At once. In the close dialog, choose False positive. The checkbox Silence incidents like this on tenant name appears. Tick it and close.
  • By learning. Three false-positive closes of the same pattern on one tenant within 30 days arm a rule. This needs Learn from false-positive closes to be on under Incident settings. It is on by default.

The checkbox appears only when both are true:

  • The reason is False positive. It is not offered for Benign activity or Resolved.
  • The incident involves exactly one tenant.

The help text under the checkbox reads: “Future incidents that open the same way, at or below this severity, move to the Silenced tab instead of Open. Event alerts are not affected. You can undo this in Incident settings.”

A rule belongs to one tenant. It matches incidents that open the same way: the same platform, opening event type, stage and kind of entity, and the same detectors. It has a ceiling: the severity and score of the incidents it learned from. It also remembers the stages, detectors, countries, browser families and actors it saw.

The rule has a description such as “Okta user.session.start (Sign-in) opening on an identity in Okta production, up to WARN”. You see it in the message, in the Silenced chip and in the rules table.

An incident that is already escalated is never silenced.

Message Meaning
“Incident closed” Closed. No rule was armed.
“Incident closed” with “Incidents like this were already being silenced.” A rule for this pattern was already armed. This close kept it going.
“Incidents like this will now be silenced on tenant.” with the rule and Undo This close armed a new rule. The message stays for 15 seconds.

Undo disarms the rule and shows “Silencing undone”. It does not reopen the incident.

When a rule is armed, every workspace admin receives an email with the subject “Identiguard is now silencing an incident pattern on tenant” and the button Review in Incident settings.

Open the Silenced tab to review what the rules are hiding. Hover the Silenced chip on a row to read the rule.

An open silenced incident shows a banner: “Silenced by a learned rule: rule. It stays off the Open tab until it grows beyond what the rule learned.”

Members and admins see This was real on each row of the Silenced tab and on the banner inside the incident. Use it when a silenced incident is a real problem. It does two things:

  • The incident returns to its normal tab.
  • The rule is retired for good. Every other incident that the rule was hiding is released too.

You see “Moved to Open and the rule was retired” or “Moved to Closed and the rule was retired” on the list, or “No longer silenced, and the rule was retired” inside the incident. In Incident settings the rule then reads “Disarmed by email (this was real)”.

A silenced incident leaves the Silenced tab on its own when it grows beyond what the rule learned. This happens when the incident:

  • escalates
  • goes above the rule’s severity or score ceiling
  • reaches a stage, or is matched by a detector, that the rule never saw
  • gains a risk signal from the provider
  • spreads to a second tenant
  • shows a country, browser family or actor that the rule never saw

Each one is counted in the Break-outs column of the rules table. A rule with many break-outs is probably too broad.

  • An open incident with no new activity closes after 7 days. An escalated one closes after 30 days. It is marked Closed automatically (quiet).
  • An incident closed this way is reopened when new activity arrives on the same entity within 24 hours of its events. An incident closed by a person is not reopened. New activity opens a new incident.
  • Closed incidents are removed 90 days after closing.
  • Three false-positive closes of the same pattern on one tenant within 30 days arm a silence rule.
  • A silence rule expires 90 days after it last matched.

Admins manage the master switch, each detector, silence rules and correlation groups under Incident settings.

What you see Why What to do
Incidents is missing from the sidebar The plan is Pro or Max without the Automated Identity Incidents add-on, or an admin has switched automatic incidents off An admin opens Incident settings, ticks Enable automatic incidents for this workspace and chooses Save detectors. If the card says “Not available on the current plan.”, choose Manage plan.
Automatic incidents are switched off, or the message “Automatic incidents are switched off for this workspace.” The same The same
Could not load incidents: “The incident engine did not answer. Nothing has been lost - the list rebuilds itself when it does.” The request for the list failed Choose Retry. Check your connection. If it continues, report it to the platform team.
Offline beside the count The live connection is not open The list still works but does not update on its own. Reload the page. Check whether a proxy or firewall blocks WebSocket connections.
The tabs show no numbers The summary could not be loaded Reload the page
“Risk scores are unavailable right now.” The summary could not be loaded Reload the page
“No identity is currently carrying risk.” Nothing is open against any identity Nothing. This is a healthy state.
No incidents with “N of M detectors are evaluating across N tenants.” Nothing matches the tab and filters Check the other tabs and the silenced line. Read the sentence for blocked detectors.
“No tenants are connected yet, so no detectors are evaluating.” The workspace has no tenants Connect a tenant
“N are blocked by missing telemetry.”, or a Blocked chip with “Subscribe tenant to …” or “Enable the … poll target on tenant …” The tenant does not subscribe to the event type (Okta, Auth0) or the stream (Entra ID) that the detector needs Edit the tenant and add the named event type or stream. See Fix a Blocked detector.
“N are blocked by masked identities.”, or “Masking hides identities on tenant. Set Obscure log stream data to XXHash to evaluate detector” The Auth0 log stream masks identities In Auth0, change Obscure log stream data on the stream to XXHash or none. Agree this with the client first.
“Identities on tenant are hashed by the Auth0 log stream. Detection is unaffected, but users appear as pseudonyms rather than names.” The Auth0 stream uses XXHash Nothing. This is information.
A Switched off chip with “Switched off for this workspace.” An admin switched that detector off Choose Turn on in settings, tick the detector and choose Save detectors
“N detectors not shown: they read telemetry no connected tenant sends” The workspace has no tenant of the platform those detectors need Nothing. This is information.
“1 incident silenced by a learned rule” or “N incidents silenced by learned rules” A silence rule is keeping incidents off Open Select the line and review the Silenced tab
Nothing is silenced No incident is silenced Nothing
“That event is not part of an incident” You followed a link from an event that has no incident Use Explore correlations on the event
“Could not check whether that event belongs to an incident.” The lookup failed Try again
What you see Why What to do
Incident not found: “It may have been removed with its workspace data, or the link is wrong.” The link is wrong, or the incident was removed 90 days after it closed Choose Back to incidents
Could not load this incident: “The incident engine did not answer. Nothing has been lost - this page rebuilds itself when it does.” The request failed Choose Retry
“Viewers can read incidents but not close them” You have the viewer role Ask an admin to change your role to member
“Raw event no longer stored” on an evidence row The event is older than the workspace’s event retention. The incident keeps its own summary. If you need raw events for longer, an admin raises retention under Policies
“Newest N of M events attached” More than 500 events are attached. You see the newest 500. Use the event feed or Correlations for the rest
“N routine events not stored (cap reached)” The incident reached its limit of 500 stored events Nothing. This is information.
“No detector or baseline hits. This incident is built from vendor signals and direct evidence.” The incident opened from a provider risk signal or direct evidence Nothing. This is normal.
“No events are attached to this incident yet.” The evidence is not attached yet Wait a moment and reload
“Silenced by a learned rule: … It stays off the Open tab until it grows beyond what the rule learned.” A silence rule matched this incident If the incident is real, choose This was real
What you see Why What to do
“Could not close this incident.” The close failed Try again. Check your role and your connection.
“Request failed (400)” when you close with the silence box ticked The incident cannot teach a rule. It opened before silence rules existed in the product. Close it without the tick. Silence a newer incident of the same kind, or switch the detector off.
“Only a false positive can silence incidents like it” Silencing was requested with another reason Choose False positive
The silence checkbox is not there The reason is not False positive, or the incident involves more than one tenant Choose False positive. An incident across several tenants cannot be silenced.
“Forbidden: Insufficient permissions for this workspace” You are a viewer, or your membership changed You need the member or admin role
“Could not reopen this incident.” The reopen failed Try again
“Incident closed” with “Incidents like this were already being silenced.” A rule for this pattern was already armed Nothing. This is information.
“Incidents like this will now be silenced on tenant.” This close armed a new rule, by the tick or as the third false positive in 30 days If you did not intend it, choose Undo within 15 seconds. Later, an admin can disarm the rule in Incident settings.
“Could not undo the silencing. Disarm the rule in Incident settings.” The undo failed An admin opens Incident settings, Learned dismissals, and chooses Disarm on the rule
“Could not release this incident.” This was real failed Try again. Check your role.
“Cannot reach Identiguard. Check your connection and try again.” You are offline, or there is a network error Restore the connection and try again
An email “Identiguard is now silencing an incident pattern on tenant” A silence rule was armed. Every admin receives it. Review the rule in Incident settings. Disarm it if it is wrong.

For problems on the settings page, see Incident settings.