Automate a cron job monitor that alerts when a scheduled task goes quiet

A web request that fails shows up in logs and error trackers. A job that never starts shows up nowhere, because there is nothing to log. Heartbeat monitoring turns that around: each job pings a URL when it runs, and the monitor raises an alarm when a ping fails to arrive. Healthchecks.io, used here as the example, describes its check states as new, up, late, down and paused.

Each check has an expected period or cron schedule, plus a grace time. In the Healthchecks.io docs example, a check with a one-hour period and five minutes of grace, last pinged at 12:00, turns late at 13:00 and down at 13:05. Late means the signal has not arrived yet. Down means the grace time ran out and alerts go out. Separating the two stops a slow run from looking like a dead one.

You add the job and grant it read access to the monitor; the team requests the access when you add it. Healthchecks.io offers read-access API keys that can list checks and get a single check. Each day a scheduled agent lists your checks, reads each status and last ping time, keeps those that are in grace or down, sorts by how long each has been quiet, and sets paused checks aside so deliberate pauses do not look like failures. Jobs can also append /start or /fail to their ping URL; where a job uses them the report shows a run that began and never finished, without guessing the cause.

The CTO agent, as department head, reviews the report before it goes out: is each job really late rather than paused, are durations right for its schedule, is any suggested starting point reasonable. Then you approve, edit or skip it. Your monitor keeps sending its own alerts, so this is a recap, not your sole alarm.