Alert channels, policies, quiet hours and digests
Every alert is in the Portal inbox. Channels put an alert where your team already looks, such as email, Teams, Slack, a ticket in your PSA, a signed webhook or your SIEM. A policy says which alerts go…
Written By Christopher Scaminaci
Last updated 4 days ago
Every alert is in the Portal inbox. Channels put an alert where your team already looks, such as email, Teams, Slack, a ticket in your PSA, a signed webhook or your SIEM. A policy says which alerts go to which channels. The schedule sets quiet hours and digests. This page covers all three.
At a glance
- What it does: sends alerts to your team's tools, on your rules.
- Plan: Free has the inbox, email and text messages (a Free organization sends texts only with SMS pay as you go). Pro adds Microsoft Teams, Slack, a ticket in your PSA and a signed webhook. Business adds the SIEM forwards: Microsoft Sentinel, Splunk HTTP Event Collector, syslog over TLS and OpenTelemetry export.
- Who: an Administrator changes settings. Reviewers and auditors read them.
- Where: AI Governance → Alerts → Alert settings, with the tabs Channels, Policies and Schedule.
- Needs: for a ticket channel, your PSA already connected to StackJack.
A channel above your plan cannot be added. If you save a channel and later move to a lower plan, its settings are kept and it shows "Needs a higher plan" until you upgrade.
Start with email
A new organization gets one email channel, "Governance administrators", and three starting policies, so email alerts work with no setup:
- Serious findings
- Source and coverage problems
- Incidents and approvals (an incident, a notification deadline that is due or missed, and a request for approval)
You can change or remove any of them. One you remove does not come back.
Add a channel
- Open Alerts → Alert settings and the Channels tab.
- Select Add channel and choose the kind.
- Fill in the form, save, and use Send test to see what the channel receives.
Every web address, token and key you give is stored in your region's key vault and is never shown again. To change one, use Rotate secret.
Choose your governance administrators (looked up when each email is sent) or up to ten addresses of your own. This channel needs no setup on your side.
Text message
Pick the people who should get a text. The list shows only people who turned alert texts on for their own phone, so you cannot type a number for someone else. Anyone who replies STOP is removed from every text channel, and only they can turn texts back on. A text carries the alert title, the severity and your organization name, and no link. Every plan can add the channel, and Pro and Business include a monthly allowance of texts. The opt-in steps, the STOP rules, the allowance, the cost of extra texts and the option to use your own Twilio account are on Text message alerts.
Microsoft Teams
In Teams, open Workflows and create a workflow that starts with "When a Teams webhook request is received" and posts the card to a channel. Save it, copy the web address it shows, and paste it into the channel. The address is the credential: anyone who holds it can post to that channel.
Slack
Either add an Incoming Webhook to your workspace and paste its address, or build a workflow that starts from a webhook and define these text variables: title, severity, organization, summary and link. Choose which of the two you pasted in the channel's settings.
A ticket in your PSA
Pick the PSA: HaloPSA, ConnectWise PSA, Autotask, Syncro, Atera, Gorelo or SuperOps. Fill in the ticket fields it needs, such as the client and ticket type. The PSA must already be connected to StackJack. The ticket is opened through that connection.
For HaloPSA, ConnectWise PSA and Autotask, save the channel first. Then Look up options asks the PSA for its boards, types, priorities and statuses so you can pick them from a list. You can set a priority for each severity. If the ticket cannot be created, the alert is emailed instead to the people you name. A rule that asks for a ticket sends to every PSA channel marked Use for rule-requested tickets.
Signed webhook
StackJack posts each alert as JSON to the address you give and signs it the Standard Webhooks way. StackJack creates the signing secret and shows it to you once, when you add the channel. Copy it then. If you lose it, rotate the secret.
To check a request:
- Read the three headers
webhook-id,webhook-timestampandwebhook-signature. - Build the text
{webhook-id}.{webhook-timestamp}.{request body}, with the body exactly as received. - Compute HMAC-SHA256 of it with the key you get by base64-decoding the part of your signing secret after
whsec_. - Base64-encode the result. The signature header holds
v1,followed by that value. It is a space-separated list when there is more than one. - Reject a request whose timestamp is too old.
The webhook-id is the delivery's id. Use it to ignore a repeat, because a delivery can be sent twice if a network fault hides the first answer.
Splunk HTTP Event Collector
Paste your HEC address and its token. The index is optional. StackJack adds the standard HEC path when your address has none.
Microsoft Sentinel
- Create an app registration in your Entra tenant.
- Give it the Monitoring Metrics Publisher role on your data collection rule.
- Create a custom table named
StackJackAiGovernance_CLwith the columns shown on the channel's page, and a data collection rule with the streamCustom-StackJackAiGovernance_CL. - Give StackJack the data collection endpoint address, the rule's immutable id, the stream name, your Entra tenant id, and the app registration's client id and client secret.
Sentinel in the public Azure cloud is supported.
Syslog over TLS
Give the host and TLS port of your collector. The usual port is 6514. StackJack sends RFC 5424 messages over TLS (RFC 5425). Leave the certificate fingerprint blank to trust a certificate your system trust accepts. Or paste the SHA-256 fingerprint of your collector's certificate to trust exactly that one. This is common for a self-signed collector.
OpenTelemetry export
Give your collector's logs endpoint address. Add headers, such as an authorization header, one per line as Name: value. You can add up to ten.
Which alerts go where: policies
A policy says which alerts go to which channels. Its filter can name kinds of alert, a lowest severity, sources, people, groups, rules and rule packs. An empty filter matches everything. A policy sends immediately, or as a daily or weekly digest. An alert that matches no policy is still in the inbox.
- Joining repeats. Within a policy's window (60 minutes by default, 0 to 1,440) the same kind of alert about the same rule, person and source is joined into one message. The message says how many similar alerts came after.
- An hourly limit. Each email, Teams, Slack and PSA channel sends at most 30 immediate messages an hour (10 for a PSA ticket). You can set the limit from 1 to 500. Anything more in that hour is joined into one message at the end of the hour that says how many more there were.
- A first scan of unapproved tools is one message. When a new source shows five or more unapproved AI tools at once, an email, Teams or Slack channel gets one message that lists each tool with its own link. It lists the first 25, then a link to the inventory for the rest. Every tool is still its own alert in the inbox. A digest policy or quiet hours keep their own way of grouping. A tool that turns up later on its own is a message of its own.
- Machine channels get everything at once. Webhook, Sentinel, Splunk, syslog and OpenTelemetry channels always receive every alert as it happens. Digests, quiet hours and the hourly limit do not hold them back, and they are not given the one-message list above.
Quiet hours and digests
Open the Schedule tab. Set the organization's Time zone, Quiet from and Quiet until, the hour Digests go out at and the day Weekly digests go out on.
- During quiet hours, email, Teams and Slack alerts below Critical are held and sent as one summary when quiet hours end. Critical alerts always go at once.
- A digest policy joins its alerts into one message per period.
- A ticket channel still gets each alert at once, because a ticket is a queue item and not a page.
A digest or summary groups alerts by kind and rule with a count and the highest severity. It never counts, sorts or lists by person.
Names in alerts
By default an alert names a person by their pseudonym. If your name policy lets governance administrators see real names, you can mark a channel Audience is governance administrators when everyone who can read that channel is one. That channel's alerts then carry real names, and every alert that names someone is recorded in the audit trail. Under the pseudonymous policy the option does not exist, and a name is only ever shown through an approved reveal.
If a channel fails
Each channel shows its health: working, having trouble, failing, needs a higher plan, or not configured. A failed send is retried with growing waits up to eight times. After that it is marked failed. A channel that fails three times in a row raises its own "Alert channel failing" alert, which is never sent to the failing channel. A failed PSA ticket falls back to email. Use Send test on a channel to see what it receives and the code of any failure.
StackJack support can pause a channel if it is causing harm. The reason is recorded in your audit trail. Only support can resume a channel it paused.
Troubleshooting
Next
Was this helpful?
More in Control and respond
Rule packs, modes and custom rulesFindings and triageData-loss rules, findings and scanning StackJack tool trafficFilter personal data out of tool results (the PII filter)Still need help? Ask the team