Skip to main content
A notification destination is where Ankra sends alerts and notifications: a Slack or Microsoft Teams channel, or any address that takes a JSON POST - Discord, PagerDuty, Opsgenie, Datadog or your own receiver. Alert rules tell the destinations ticked on them, and notification routes send any notification kind to a destination. Each destination keeps its own payload, a JSON template that Ankra fills in for every notification it sends there.

Before you start

  • A role that manages alerts. Without it the page reads Your role can look at destinations but not change them. Someone who manages alerts can.
  • For a Slack or Teams channel without a webhook address: the Ankra Slack app or the Ankra bot for Teams, installed under Ankra AI → Gateway connections. See AI Connections.

Home channel

Set an organisation home channel and Ankra posts there when something important happens - failures, agents going offline, security findings - and no notification route matches it. It is the default destination, so your team hears about important events before anyone wires up a single rule.
1

Connect Slack

Install the Ankra Slack app under Ankra AI → Gateway connections, and invite the Ankra bot to any private channel you want to use. Until Slack is connected, the home channel section says Connect a Slack workspace under Integrations to choose a home channel.
2

Pick the home channel

Open Ankra AI → AI workspaces (/next/organisation/ai-workspaces). Under Ankra’s home channel in Slack, pick the channel and click Save home channel.You should see Saved. Ankra posts in followed by the channel’s name, and the section reads The home channel is that channel.
Only critical and warning notifications reach the home channel; informational ones stay under the bell. The home is a fallback - when a routing rule already matches a notification it goes there instead, and the home is not notified as well. Alerts that already post to their own destinations are left to those destinations.

Add a destination

1

Open the form

Open Integrations & alerts → Notification destinations (/next/organisation/notification-destinations) and click Add destination. The dialog is titled Add a destination.
2

Choose the kind

In Kind, choose what you send to:Pasting an address from Slack, Teams, Discord or PagerDuty switches Kind to match.
3

Choose where it posts

For Slack and Microsoft Teams, once the Ankra Slack app is installed or the Ankra bot is in a Teams team, How Ankra posts offers To a channel (Ankra’s app posts to the channel you pick) or To a webhook address. With To a channel, pick it in Channel; Teams channels are listed across every team the bot is in. When Teams lists none, the dialog says Microsoft Teams lists no channels Ankra can post to yet. Add the Ankra bot to a team, or mention it once in a channel of a team it is already in, then reload. See Connecting Microsoft Teams.Otherwise paste the address into Webhook address. Ankra keeps it encrypted and never shows it again, only its host.
4

Name it and set the payload

  • Name - up to 100 characters, for example PagerDuty production.
  • Description - optional, up to 500 characters.
  • Payload - the JSON Ankra posts. For Slack and Another webhook, an empty payload sends Ankra’s own card; write JSON to send your own. Microsoft Teams, Discord and PagerDuty start with a payload filled in. Use the kind’s payload (or Use Ankra’s card) puts the starting payload back. Variables you can use lists every template variable.
  • Send to it - on by default. Off keeps the destination and sends nothing to it.
5

Save and test

Click Add destination. Then open the new row’s options menu (…) and choose Send a test.You should see Test sent to the destination’s name, with The receiver took it and the HTTP status, and the sample in the channel or receiver.
From the CLI, ankra alerts destinations create and ankra alerts destinations test do the same, see Managing destinations and routes from the CLI.

Template Variables

Use these variables in your custom templates. They are replaced with actual values when the notification is delivered. One destination stores one template, and that template renders every notification it receives - alert firings, alert resolutions, and the operational notifications a routing rule sends it. The tables below mark which of those a variable carries a value for.

Notification

Alert Information

Alert firings and resolutions carry all of these. Operational notifications carry {{severity}} and its two presentation values; the rest are empty for them, except a firing routed by a routing rule, which also carries {{condition_type}} and {{trigger_value}}.

Resource Information

Every notification kind fills what it has. Where a producer records no finer resource kind, {{resource_type}} reads as the thing that went wrong - “operation”, “resource”, “agent”, “insight”, “gitops repository”, “security report”. Agent notifications are about the agent on a cluster, so their {{resource_name}} is that cluster’s name.

Action URLs

Empty values and fallbacks

No notification kind carries a value for every variable - a cluster-level alert has no job, an agent update has no condition. Two things stop that showing up as a column of blank rows in your card:
  • Empty rows are dropped for you. After substitution Ankra removes the rows whose whole purpose was to show one value: Adaptive Card FactSet facts whose value came out empty (and a FactSet left with no facts), empty TextBlock entries, Slack section fields whose *Label* has no value under it, and the {"name", "value"} rows a Discord embed or a legacy Teams MessageCard carries. All four built-in presets therefore render correctly for every kind with no editing. Containers, columns and action sets are never removed, and a card is never pruned down to nothing.
  • You can supply your own fallback with {{variable|text}}, for example {{resource_name|not applicable}}. The fallback is used both when the variable is unknown and when it resolves empty, so it wins over the row being dropped.
A variable name Ankra does not recognise renders as an empty string rather than as literal {{...}} text. The JSON-valued variables ({{data}}, {{route_hint}}, {{trigger_details}}, {{resource_snapshot}}) follow where you put them: written unquoted they splice in as a nested object, and written inside a quoted string they are escaped for you so the payload stays valid JSON.
Every other value is escaped for the string it sits in, so a resource name containing a quote or a newline cannot break your template.

Integration Examples

Slack

Slack webhooks use the Block Kit format for rich messages:
Create a Slack webhook at Your Workspace → Apps → Incoming Webhooks.

Microsoft Teams

Microsoft retired Office 365 Connectors, so a Teams channel no longer offers an “Incoming Webhook” - the supported route is a Power Automate (Workflows) flow using the “Send webhook alerts to a channel” template, which hands you an HTTP POST URL to paste into Ankra. Payloads on that route must be Adaptive Cards wrapped in a message attachment. The legacy MessageCard format is silently degraded: the flow reports success, but Teams drops every potentialAction button. Use {{severity_style}} for the card colour - Adaptive Cards cannot consume the {{severity_color}} hex value:
Choosing Microsoft Teams as the Kind fills Payload with this Adaptive Card format - adjust it from there. For interactive alerts inside Teams without Power Automate, connect the Ankra bot instead: see AI Connections.

Discord

Discord uses the embed format:

PagerDuty

PagerDuty Events API v2 format:
Use PagerDuty’s Events API v2 endpoint: https://events.pagerduty.com/v2/enqueue. Choosing PagerDuty as the Kind fills Payload with a payload of this shape; replace YOUR_PAGERDUTY_INTEGRATION_KEY with your integration key.

Opsgenie

Opsgenie Alert API format:

Custom/Generic

For any other service, create a JSON payload that matches their expected format:

Testing a destination

Send a test, in a destination’s options menu, sends a sample notification to it and reports what the receiver answered: Test payloads populate every template variable, so a template that renders correctly under test renders correctly on delivery - the test and the delivery pipeline run the same renderer. A real notification only fills the variables its kind has a value for, and the rows for the rest are dropped as described in Empty values and fallbacks. Send a test on a routing rule, in its options menu on Integrations & alerts → Notification routes, sends the same sample through the rule’s own kind, so you can check a rule and its destination’s payload together.

Managing destinations

  • Switch off - the switch on the row. The row then shows Off and nothing is sent to it while it is off; it keeps its settings.
  • Change - Change… in the row’s menu. A destination’s kind stays; to change it, add a new destination. Leave Webhook address empty to keep the address it has.
  • Delete - Delete… in the row’s menu, then Delete destination. Alert rules and routing rules that send to it stop reaching anyone through it.

Managing destinations and routes from the CLI

Everything above can be scripted with the ankra CLI, using the same personal access token as the rest of the CLI, so alert routing can live in Git and be applied from CI alongside your clusters. -o json and -o yaml keep the output parseable.
Updates send only the flags you pass, so the rest of a destination or route stays untouched. The same endpoints are available directly for other tooling under /api/v1/org/alerts/integrations and /api/v1/org/notifications/routes with a bearer token. Before you rely on a set of rules, run ankra alerts routes preview to see which destinations a notification would actually reach. Notification routing covers the full matching, ordering, and de-duplication behaviour.

Still have questions? Join our Slack community and we’ll help out.