Monitoring that doesn't cry wolf.
HTTP, keyword and heartbeat checks with multi-region consensus and flap-proof incidents. You hear about it when something is really wrong, not when one region blinks.
Built to be believed
A false alarm at 3am costs you more than a missed one. Every default here is tuned so an alert means something.
HTTP and keyword
Check any endpoint for status codes, and for a keyword that should or shouldn't be present in the response body.
Heartbeat checks
For cron jobs and workers, get a URL to ping on each run. Miss the window and Sentivel opens an incident.
Multi-region consensus
A monitor is checked from several regions and the results collapse to one verdict, so a single region's network blip never opens an incident.
Flap-proof incidents
Configurable open and resolve thresholds mean a brief recovery stays one incident, and a real outage isn't lost to a lucky retry.
Latency and degraded
Mark a component degraded when a response is slow rather than failed, with a threshold you set per monitor.
Response-time history
See how each endpoint's latency has moved over time, not just whether it was up.
No premature green
A brand-new monitor reads as warming up until it has proven itself, so a page never claims all-clear before it knows.
Bulk setup
Paste a list of URLs to import monitors at once, then pause, resume or retune them in bulk.
Alerts on change
Every open and resolve can notify your integrations or page a person, deduplicated so one incident is one alert.
Your page follows your monitors
Component status is derived from the flap-confirmed monitor state, not the raw check, so a single blip can't flip a component to red on your public page. What visitors see is what your team has actually confirmed.
Daily uptime rollups feed the 90-day history on the status page, and scheduled maintenance windows are excluded so planned work never dents your numbers. Monitoring and the page are one system, not two tools you keep in sync by hand.