Twelve regions, one timeline
A check fails in Frankfurt but passes in São Paulo. Kestrel shows you both on the same timeline before it pages anyone.
Know before
your users do.
Kestrel watches your endpoints from twelve regions and tells you what broke, not just that something did.
Used at
Everything a two-person on-call rotation needs.
Alerts that route themselves
Rules read your on-call calendar. Nobody gets woken up for a staging deploy.
Stack traces, deduplicated
Forty thousand identical exceptions collapse into one issue with a count next to it.
Status pages your customers trust
Hosted on a domain you control, updated from the same incident you are already working on. No second dashboard to remember.
Running in about four minutes.
-
01
Install
One package, one environment variable. Node, Python, Go, Ruby, Elixir and PHP are supported today.
npm i @kestrel/agent -
02
Connect
Point Kestrel at the URLs that matter and pick who gets called at three in the morning.
kestrel watch ./routes.yml -
03
Sleep
You hear from Kestrel when something is genuinely wrong, and not one notification before that.
all systems nominal
“We cut our alert volume by two thirds in the first week. The pages we still get are the ones worth getting up for.”
Head of Platform, Cadence Labs
Two plans. No sales call.
Solo
€14/month
For one developer and up to 20 checks.
- 20 endpoint checks, 60 second interval
- Error tracking, 50k events a month
- Email and Telegram alerts
- 7 day history
Team
€47/month
For a rotation of up to eight people.
- 200 endpoint checks, 20 second interval
- Error tracking, 1M events a month
- On-call schedules and escalation rules
- Hosted status page on your domain
- 90 day history and audit log
Both plans bill monthly and cancel from the dashboard. No card needed for the 14 day trial.
Questions worth answering.
What happens when Kestrel itself goes down?
Checks run from twelve independent regions with no shared control plane. If a region cannot report in, the others still page you, and the gap is marked on your timeline rather than hidden.
Can I self-host it?
Yes. The agent and the collector are both available as containers, and the Team plan licence covers a self-hosted collector on your own infrastructure at no extra cost.
Do you store request bodies?
Not by default. Headers and bodies are stripped at the agent before anything leaves your network. You can opt specific routes back in if you need them for debugging.
How is this different from what my cloud provider gives me?
Provider dashboards tell you a region is healthy. Kestrel tells you whether a customer in that region can actually complete a checkout, which is a different question and usually the one being asked at 3am.