Search and process in-app notifications

Before you begin

Sign in with the role and team memberships you normally use. Search and inbox results respect the current user's authorization, so missing results may be an access boundary rather than a data problem.

Open the feature

Use the desktop sidebar to move between Dashboard, Incidents, Services, Teams, on-call, analytics, and settings surfaces available to your role. Press Cmd+K on macOS or Ctrl+K elsewhere to open product-wide sidebar search. Press / only in a page that implements a local search shortcut, such as the Incidents list. Select the bell in the top bar to open the personal in-app notification drawer.

Configure and use search

Enter at least two meaningful characters. Queries are bounded and normalized.
Review the result type before opening a match. Search can return authorized incidents, services, teams, users, escalation policies, and postmortems.
Add a more specific title, identifier, service, or person fragment if results are broad.
Open the result and verify its record identity before taking an action.

Record results are permission-aware. Incident and service queries apply authorization filters, and privileged-only result types are not a bypass around normal access. Generic navigation suggestions are a static convenience list and can include a destination the current role cannot use; the destination still enforces authorization. The endpoint is rate limited, so rapid automation or repeated queries can return a temporary rate-limit response.

Process the notification inbox

Open the top-bar bell and inspect the unread badge.
Use All, Unread, Incidents, or Shifts to narrow the drawer. Service notifications remain identifiable by their service type in the list.
Select a notification with an incident target to open that incident.
Use the inline control to mark one item read, or Mark all as read after reviewing the outstanding items.
Interpret Connecting as initial setup, Live as a connected event stream, Reconnecting as an SSE retry/backoff period, Offline as a lost network, Paused as a background/hidden page, and Polling as the fallback used only when the browser lacks EventSource. Verify critical work on the Incidents page rather than relying only on the badge.

How it works

In-app inbox: a personal, signed-in activity drawer with read/unread state.
Paging providers: email, SMS, voice, push, WhatsApp, Slack, or Teams delivery used to reach responders outside or alongside the web application.
Delivery history/operations: administrator evidence about provider attempts, outcomes, retries, and failures; it is not a user's unread inbox.

Reading an in-app item does not acknowledge or resolve its incident, and it does not prove an external page was delivered.

Verify the result

Create or use a controlled event that produces a visible notification. Confirm the unread count changes, the appropriate filter contains the item, the target opens correctly, and marking it read persists after closing and reopening the drawer. Confirm a user without access cannot discover the protected record in search.

Change or undo it

Read state can be changed only through the available inbox actions; marking an item read does not mutate the underlying incident. Restore or change access in the owning team/role workflow, not through search. Adjust external channel preferences under Settings → Profile → Notifications.

Troubleshooting

Search returns nothing: use two or more characters, simplify punctuation, check spelling, and verify access to the record.
Search is temporarily unavailable: wait for the displayed/retry interval if rate limited, then try a narrower query.
The unread count looks stale: reopen the drawer; check whether it reports Live or Polling, then reload and verify the source event.
A drawer item does not navigate: service or shift items may not carry an incident target. Use the relevant service or schedule area.
An external page is missing: inspect provider configuration, routing, user preferences, and delivery history; inbox state is separate.