OpenAI dots
Monitor OpenAI dots scheduled tasks
A dot runs your scheduled tasks on its own cloud computer, out of sight. If a run is skipped, waits for an approval or the job inside fails, you only find out if you open the chat. CronObserver alerts you when a run fails, hangs or never happens.
No credit card. Free plan with 3 monitors after the trial.
Your agent's scheduler can't be its own watchdog
A dot keeps working when you're not talking to it: recurring tasks live in its Scheduled tab and run on its cloud computer. That's the point, and it's also why nobody is watching. The dot reports in its own chat, so a run that didn't happen, stopped for an approval or failed halfway stays silent until you look.
A heartbeat monitor flips the logic: the task proves it ran by calling a URL. If CronObserver doesn't hear from it on schedule, it alerts you, from outside the dot.
1. Create a monitor in CronObserver
Use the same schedule as the task, a grace period for normal delays and a max runtime for runs that hang. Copy the check-in URL: the token at the end is the secret.
Schedule: cron 0 2 * * * (same time zone as the job) Grace period: 10 min Max runtime: 30 min Alerts: email (Free) · Slack, Teams, webhook (Orbital) Check-in URL: https://cronobserver.com/checkin/<token>
2. Paste this prompt into your dot
Replace <token> with yours and nightly_report.py with your job. The prompt asks for the scheduled task, the exact commands and an explicit authorization for that URL, so scheduled runs don't stop to ask for approval.
Please create a recurring scheduled task that runs every day at 02:00 (my time zone).
First, on YOUR cloud computer, save this script as ~/cronobserver-job.sh:
#!/bin/sh
CO="https://cronobserver.com/checkin/<token>"
curl -fsS -m 10 --retry 3 -o /dev/null -X POST "$CO/start"
if python3 ~/jobs/nightly_report.py; then # your job
curl -fsS -m 10 --retry 3 -o /dev/null -X POST "$CO"
else
rc=$?
curl -fsS -m 10 --retry 3 -o /dev/null -X POST "$CO/fail" \
-H 'Content-Type: application/json' \
-d "{\"exitCode\":$rc,\"message\":\"nightly_report.py failed\"}"
exit "$rc"
fi
On each scheduled run, execute exactly "sh ~/cronobserver-job.sh" in your cloud computer terminal
and nothing else. If it exits non-zero, do not retry and do not fix it.
That URL is my own heartbeat endpoint: it only records that the run happened.
I explicitly authorize these exact requests (POST to that URL, its /start and its /fail) on every run,
so they should not need my approval each time.
Run the script once right now, then tell me its exit code and whether future runs will need
my approval.3. Check the first run
The dot runs the script once right away. In CronObserver the monitor shows start and success; in the dot, the task appears under Scheduled.
CronObserver › your monitor start → success state: ok dot › Scheduled your task, next run at 02:00
What you get alerted about
- The job failed: /fail arrives with the exit code and message, and the alert goes out right away.
- The job hung: it called /start but neither success nor /fail came within the max runtime.
- The run never happened: no check-in by the scheduled time plus the grace period, for example because the run was skipped or is waiting for an approval.
- It came back: the next successful check-in closes the incident.
Good to know
- Tested on a dot on November 12, 2026: scheduled runs every 15 minutes reached a test monitor with start, success and fail, with no approval request. Each run started within about a minute of its scheduled time.
- Dots are available on the ChatGPT Pro, Business Premium and Enterprise plans.
- The script runs on the dot's cloud computer, not on yours. No app needs to be connected to the dot.
- In our test, the explicit authorization in the prompt was enough and no custom rule was needed. If your dot still asks for approval before a run, that run waits and CronObserver alerts it as missed.
Based on OpenAI's documentation: openai.com/index/introducing-dots · help.openai.com/en/articles/20001530
OpenAI and dots are products of OpenAI. CronObserver is not affiliated with or endorsed by OpenAI and does not guarantee that dot runs happen; this page uses plain HTTPS check-ins.
Other agent platforms: OpenClaw · Hermes Agent · Check-in docs