1. Create a webhook under Account settings > Integrations
Name it after the tool, pick the status page, and choose the default component and region used when an alert names none. Optionally mint a signing secret. The URL (which contains the token) and the secret are shown once; rotate or revoke them from the same list.
2. Alertmanager or Grafana: paste the URL as a webhook receiver
The standard webhook body is understood unchanged: each alert's fingerprint is the key, its summary the title, its description the body. Add the labels realuptime_component (a check name or id on that page) and realuptime_region (iad, sjc, fra, nrt) to an alert rule to steer where it lands. Repeated notifications of the same firing are no-ops; resolved notifications close the incident.
3. Datadog, New Relic, anything else: send the generic shape
A JSON object with key and status (firing or resolved), optionally title, body, component, region and severity, or events: [...] for a batch. The API docs show a Datadog payload template and a New Relic custom payload that emit exactly this.
4. Check the audit log
Every accepted event writes an audit row saying what it did (opened, updated, resolved, ignored) and which incident; every refusal writes one with the reason, so a misconfigured sender or a rotated-but-not-updated URL is visible without guessing.
Go deeper
The full reference lives in the docs: Inbound incident webhook reference. Error codes named above are each explained in the error-code reference.