Heartbeat monitoring for scheduled agents

Your agents already run on schedules.
Know the moment one stops.

WatchCron is a dead-man's switch for scheduled AI agent jobs and cron workflows. Each run pings a heartbeat — the moment a run goes silent or misses its window, the right channel gets alerted.

One heartbeat URL per job. Alerts only when something breaks.

Everything you need to trust your schedules

Purpose-built liveness monitoring for AI agents, cron jobs, and any workflow that must keep beating on a schedule.

Heartbeat pings, three ways

Ping a unique URL via plain HTTP after every run, call the SDK one-liner inside your agent, or fire the MCP tool. One line of setup per job.

Dead-man's-switch alerts

If a scheduled run is late, silent, or failing, WatchCron notices the missing heartbeat — and alerts before the silence becomes an incident.

Multi-channel routing

Deliver alerts to email, Slack, Discord, or any webhook. Route each job to the channel that can actually act on it.

Cron-aware grace windows

WatchCron understands cron expressions and jittery schedules. Define a grace window so a slow run never pages you, but a dead one always does.

Failure + latency timeline

Every check is recorded: last seen, duration, streak, and failure history. See which jobs flake and which ones quietly died weeks ago.

Status API & public page

Query job health programmatically or publish a status page for your fleet. Your agents’ uptime, visible at a glance.

Live in five minutes

No agents to install, no dashboards to configure. One URL per job and you're covered.

01

Add a heartbeat

Create a job in WatchCron and ping its unique heartbeat URL after each run — from an agent hook, a cron line, GitHub Actions, or any HTTP client. Works with anything that can make a web request.

02

Define cadence + grace

Tell WatchCron the expected schedule (cron expression, interval, or custom) and a grace window. WatchCron learns what "on time" means for each job.

03

Get alerted on silence

The moment a run misses its window, WatchCron alerts email, Slack, Discord, or your webhook — with the job, the last seen beat, and the failure timeline.

Simple pricing

Start free for 14 days. Pay only when your fleet grows.

Starter

$19/mo

For indie builders with a handful of scheduled agents.

  • 10 monitored jobs
  • Heartbeat pings via HTTP, SDK, or MCP
  • Email + Discord alerts
  • 30-day check history
Most popular

Team

$49/mo

For teams running agents across many cron schedules.

  • 50 monitored jobs
  • Cron-expression aware grace windows
  • Slack + webhook routing
  • Failure + latency timeline
  • 90-day history, weekly digest

Business

$99/mo

For agent fleets where silence is a production incident.

  • Unlimited monitored jobs
  • Multi-channel escalation policies
  • Status API + public status page
  • 1-year history
  • Priority support

Loved by agent fleets

Teams running scheduled agents use WatchCron to know their fleet is alive.

WatchCron caught a dead Airflow agent three days before anyone noticed. The silence alert paid for itself in one incident.

Sarah Chen
CTO, PromptLayer

We run 40+ scheduled agents for customer content. One heartbeat line each — and no more discovering failures in customer tickets.

Marcus Webb
Head of Engineering, SynthAPI

Cron-aware grace windows mean we stopped paging on slow runs but still catch the truly dead ones. Exactly the signal we needed.

Elena Ruiz
Co-founder, VectorScale

Questions & answers

Everything else you might want to know before you start monitoring.

WatchCron is a dead-man’s switch for scheduled work. Each of your jobs pings a unique heartbeat URL after it runs; if a run doesn’t arrive within its expected window, WatchCron alerts your team. It detects silent failures — the ones that never throw an error you can see.
Every job has a configurable grace window. A run that checks in inside the window is counted as on time — you only get alerted when the heartbeat is genuinely missing past the grace period.
No. WatchCron only sees the heartbeats you send. No API keys to your stack, no code to install, no logs to ship. A single HTTP ping per run is enough.
Yes. WatchCron works with anything on a schedule — AI agent workflows, cron lines, GitHub Actions, Airflow DAGs, schedulers, or a script on a Raspberry Pi. If it can curl a URL, it can have a heartbeat.
One heartbeat URL = one monitored job. Monitor nightly content runs, per-customer agents, sync pipelines — anything with its own cadence.

Ready to trust your schedules?

Join teams using WatchCron to keep their agent fleets, cron jobs, and pipelines on a heartbeat — and sleep through the nights nothing breaks.