Webhook payload
A webhook alert channel receives one HTTP request per alert. Set it up under Alert channels.
Request
- Method:
POSTto the URL you saved. - Headers:
Content-Type: application/json; charset=utf-8andIdempotency-Key: <alert_id>. A retry of the same alert sends the same key, so you can drop duplicates. - Authentication: none. Requests are not signed, so keep the URL secret and validate what you receive.
- Timeout: 10 seconds. Any 2xx response counts as delivered; anything else, or no answer, is retried. See Delivery and retries.
Fields
Top-level field names are snake_case. Fields with no value are sent as null, not omitted. Timestamps ending in _at or _since are ISO 8601 strings in UTC.
| Field | Type | Meaning |
|---|---|---|
alert_id | string (UUID) | The alert's ID; equals the Idempotency-Key header |
test | boolean | true for test alerts |
event_type | string | monitor.down, monitor.still_down, monitor.never_pinged, monitor.recovered, monitor.vanished, or automation.outcome_failed |
alert_type | string | InitialFailure, ContinuedFailure, Recovered, NeverExecuted, OutcomeFailure, or Vanished |
message | string | One-line summary |
why | string | Why CronRadar alerted, such as the missed deadline and grace period |
owner | string | The job's owner; defaults to the team name |
severity | string | warning, error, or critical |
environment | string | The job's environment, or Environment not set |
recent_events | array | Up to three latest reports, newest first: type (heartbeat, start, complete, or fail), occurredAt, summary |
investigation | object | runbookUrl, repositoryUrl, logsUrl (each a string or null) |
outcome_summary | string or null | Which outcome expectation failed; set only for OutcomeFailure |
vanished_count | number or null | For Vanished, how many jobs left the registry in the same sync, this job included; null otherwise |
triggered_at | string | When the alert was raised |
application | object | id, name, slug, timezone |
monitor | object | id, key, name, schedule, grace_period (seconds), last_ping_at, next_expected_at, is_down, down_since |
downtime | object | down_since, duration_minutes (number or null), duration_formatted |
escalation | object | level, description (Initial alert or Daily reminder), next_escalation_at |
dashboard_url | string | The job's page in the console |
The objects inside recent_events and investigation use camelCase names, as shown.
Sample
A missed nightly run. Values are examples.
{
"alert_id": "3f2b8c1e-9a4d-4e7b-8c21-5d6e7f8a9b0c",
"test": false,
"event_type": "monitor.down",
"alert_type": "InitialFailure",
"message": "Monitor nightly-backup missed its expected ping",
"why": "No report arrived by 2026-11-16 02:00:00 UTC plus the 300-second grace period.",
"owner": "Platform team",
"severity": "error",
"environment": "production",
"recent_events": [
{
"type": "complete",
"occurredAt": "2026-11-15T02:00:41.512345Z",
"summary": "Complete received"
},
{
"type": "start",
"occurredAt": "2026-11-15T02:00:02.104882Z",
"summary": "Start received"
}
],
"investigation": {
"runbookUrl": "https://wiki.example.com/runbooks/backup",
"repositoryUrl": null,
"logsUrl": null
},
"outcome_summary": null,
"vanished_count": null,
"triggered_at": "2026-11-16T02:05:30.4127730Z",
"application": {
"id": "app_example",
"name": "Billing API",
"slug": "billing-api",
"timezone": "UTC"
},
"monitor": {
"id": "b7e4d2a1-0c5f-4a8e-9d3b-1f2e3a4b5c6d",
"key": "nightly-backup",
"name": "Nightly backup",
"schedule": "0 2 * * *",
"grace_period": 300,
"last_ping_at": "2026-11-15T02:00:41.5123450Z",
"next_expected_at": "2026-11-16T02:00:00.0000000Z",
"is_down": true,
"down_since": "2026-11-16T02:05:30.4100000Z"
},
"downtime": {
"down_since": "2026-11-16T02:05:30.4100000Z",
"duration_minutes": null,
"duration_formatted": null
},
"escalation": {
"level": 0,
"description": "Initial alert",
"next_escalation_at": "2026-11-17T02:05:30.4100000Z"
},
"dashboard_url": "https://app.cronradar.com/applications/app_example/monitors/nightly-backup"
}downtime.duration_minutes and duration_formatted are filled in for ContinuedFailure and Recovered alerts.
A Vanished alert is sent once when jobs leave their framework registry in a sync, and the jobs are paused. When several jobs leave together, one request covers them all: monitor is the first of them by key and vanished_count counts every job, so message reads like sync-crm and 11 other jobs left the hangfire registry in one sync; monitoring is paused. See Jobs that vanish from your scheduler.
Handle it safely
- Respond with a 2xx quickly and do slow work afterwards; a response after 10 seconds counts as a failure.
- Use
alert_idto ignore repeats. - Route on
event_typeoralert_type, and ignore unknown fields so new ones don't break your receiver. - Check
testbefore paging anyone.