Quickstart

New account to a page on your phone. About five minutes, most of it waiting for an app to install.

1. Create your organization

Sign in at app.acked.dev with your email. You will be asked to name an organization — this is the boundary everything else lives inside: services, schedules, people.

2. Install the app and allow notifications

Do this before sending anything. A page with nowhere to go is the most common reason a first test looks broken.

Install Acked on iOS or Android, sign in with the same address, and allow notifications when asked. iOS shows that prompt exactly once per install; if you dismiss it, turn notifications back on in Settings rather than reinstalling.

If the app offers Critical Alerts, allow those too — that is what lets a high-urgency page through the silent switch and Focus.

3. Create a service

A service is the thing being monitored — an application, a database, a job. Alerts arrive at a service, and the service decides who hears about them.

Create one in the dashboard. It will ask which schedule handles it; a new organization has one with you on it, which is all you need to test.

Every new service comes with an ingest URL already minted. You do not have to configure anything else to receive alerts.

4. Send a real alert

Copy the service's ingest URL from its page, then:

curl -X POST "<YOUR-INGEST-URL>" \
  -H "content-type: application/json" \
  -d '{"title":"Hello from curl","event_key":"quickstart","priority":"high"}'

Your phone should ring within seconds.

If it does not, open the alert in the dashboard — it will exist. The alert page shows every delivery attempt and why each one succeeded or failed, which is almost always enough to see what happened. Notifications not being allowed is the usual answer.

5. Acknowledge it

Acknowledge from the push notification, the app, or the dashboard. Acking stops the reminders and stops the escalation clock — it says I have this, not this is fixed.

When the underlying condition clears, resolve it. If your monitor can send a recovery, have it resolve the alert automatically using the same event_key.

What to set up next

• A real schedule. One person is fine for a test and not for a rotation. Add the rest of your team and set who is on call when.
• Escalation. Decide how long to wait before the next person is woken. The default is that an unacknowledged alert climbs the schedule until somebody answers.
• Your monitors. Point them at the ingest URL. See Sending alerts for the payload, and for deduplication — the thing that stops a monitor firing every 30 seconds from paging you 120 times an hour.
• Outbound webhooks, if you want alert events in Slack or your own systems. See Outbound webhooks.