KindaCan Grok Bot build Cronitor?
It gets most of the way, but you stay in the loop · and here's exactly where.
- Price
- —
- You'd save
- —
- Who runs it
- The bot
- Category
- Monitoring & uptime
Cron monitoring is a dead-man's switch: the job pings a URL when it finishes, and if the ping doesn't arrive on time, something's wrong. A bot can build the receiving end in an hour - it's a table of expected check-ins and a loop looking for overdue ones. The honest caveat is the same one as uptime monitoring, just softer: if your bot's machine is also where your cron jobs run, then one crash silences both the jobs and the thing watching them. Cheap fix, and the prompt does it: have the watcher ping an external dead-man's switch of its own.
The prompt
Written for a bot that already has a browser, a terminal and your logins · paste it in whole.
You have a computer. Build me a dead-man's switch for my scheduled jobs, and then guard the guard. The service: - A config file listing jobs: name, expected schedule as a cron expression, and a grace period. - Each job gets a secret ping URL. It calls /start when it begins and /done when it finishes cleanly. - Every minute, check for jobs that should have checked in and haven't. Alert once per incident, not once per minute - alert fatigue is how this stops working. - Alert to my email, and to a Slack webhook if one is in the env. Include the job name, when it last succeeded, and how overdue it is. Also track: - Duration per run. Alert if a job takes more than twice its 30-day median - that catches the slow death before the hard one. - A status page on the host, behind a password from the env: every job, last check-in, current state, duration sparkline. The important bit, don't skip it: - The watcher lives on a small host under my account (walk me through the signup · wait for me at the payment step) - the ping URLs have to be reachable from my servers anyway. If that host dies, the watcher goes quiet and I learn nothing. - So: have the watcher itself ping a free external dead-man's switch every five minutes, and set that to alert me if it stops hearing from you. - Test it. Kill the watcher on purpose and show me the external alert arriving.
The bot will need to sign into: resend
What you lose
- a watcher that fails independently of the jobs it watches
- phone and SMS escalation when a nightly job dies at 3am
- job duration trends that catch the slow degradation
- the free external tiers, which already do this well
The numbers
Pay as you go: $2 per monitor plus $5 per user, monthly. The free Hacker plan covers 5 monitors with email and Slack alerts. Enterprise starts at $6,000/yr.
cronitor.io/pricingChecked against the live page on 2026-08-23