Website and infrastructure monitoring
See what's happening from the outside.
Sentinel monitors websites, APIs and the infrastructure behind them from 15 regions on six continents. Every check runs from every region, so when something breaks you can see where, and whether it's your service or the network in between.
Monitoring
More than an uptime check.
A site can answer 200 OK with a broken page, a certificate a week from expiring and a backup job that quietly stopped on Tuesday. Sentinel watches each layer on its own, so every service gets the checks it needs.
Websites and APIs
HTTP
Uptime monitoring
Websites and endpoints checked as often as every 15 seconds, depending on your plan, with every check run from all 15 regions.
Content
Keyword monitoring
Require text that should always be on the page, or alert on text that shouldn't be, like an error message after a bad deploy.
Errors
Server error detection
PHP warnings, WordPress critical errors, SQL errors and stack traces caught on pages that still return 200. On by default.
JSON
API monitoring
Custom requests with your own headers and body, and assertions on the JSON response by path and value.
Infrastructure
TLS
SSL certificates
A warning before a certificate expires, 14 days out by default. Broken chains and hostname mismatches fail the uptime check itself.
DNS
DNS monitoring
A, AAAA, CNAME, MX, NS, TXT and SOA records watched for changes and resolution failures.
WHOIS
Domain expiration
An alert before the registration lapses and another if it does. Subdomains roll up to the domain you actually renew.
ICMP
Ping monitoring
Reachability for servers and network devices that don't speak HTTP.
TCP
Port monitoring
Databases, mail servers, SSH and anything else listening on a port, with the connect time and the failure each region saw.
Jobs and agents
Push
Heartbeat monitoring
Your job pings a unique URL when it finishes. When a ping doesn't arrive, Sentinel opens an incident.
Cron
Cron job monitoring
Each job checked against its own cron expression and timezone, so a missed run is flagged when it was due, not when someone notices.
402
Agent payment checks
For x402 and MPP endpoints that sell to AI agents. A valid 402 is healthy. A changed price, network or pay-to address opens an incident.
Global monitoring
Every check.
Every region.
Plenty of monitors rotate checks between locations, so each result is one place's opinion. Sentinel doesn't rotate. All 15 regions check every monitor in every interval, spread out so they never arrive as one burst.
A failure becomes a pattern you can read. Global or confined to one part of the world, and how response times compare from Seoul to São Paulo.
How multi-region monitoring works →Sentinel's network
15 regions · 6 continents
Outage confirmation
One failed check isn't an outage.
Routes flap, and a single data center loses its connection now and then. Sentinel opens an incident only when at least half the regions fail, with three or more of them on at least two continents. It closes once failures drop below that line.
When a service is known to blip, give its monitor a confirmation period and a recovery period of its own.
How Sentinel confirms an outage →Alert decision
- Regions checked
- 15
- Failing
- 1 (Frankfurt)
- Needed for an outage
- 8, on two continents
- Verdict
- Regional, no alert
Frankfurt's timeouts stay on the monitor for you to see. Nobody gets paged.
Incident evidence
See what happened, not just that it failed.
A red dot starts an investigation. Sentinel keeps what every region observed around it, so you start from evidence.
Incident management →Results by region
Each region's latest status and response time sit side by side on the monitor, instead of one blended number.
Request phases
DNS, TCP, TLS, TTFB, downloadResponse time split into its phases, so a slow check shows where the time went.
Observed failure reasons
Connection refused, DNS failures, TLS errors, timeouts and HTTP errors, recorded the way each region saw them.
Incident timeline
The failing checks, the confirmation and the recovery, kept with the incident so the history is there when someone asks what happened.
Traceroute
Captured automaticallyWhen an incident opens, Sentinel records the network path to the target.
Alerts and status pages
Get the alert to whoever can fix it.
Direct
Email and SMS
Incident and recovery alerts sent straight to the people on call.
Chat
Slack, Teams, Discord and Telegram
Alerts in the channels your team already reads.
On call
PagerDuty
Incidents open and resolve in your PagerDuty service, so escalation follows your existing rota.
HMAC
Webhooks
Signed JSON events for your own systems and automations.
Public
Status pages
Live monitor status on your own domain, public or password protected.
Scheduled
Uptime reports
Availability and incident history delivered on a schedule.
Developer tooling
The dashboard is optional.
Monitors, incidents and status are reachable from the same places you manage the rest of your infrastructure, including the terminal, your Terraform plans and your AI assistant.
REST
API
Monitors, checks, incidents and status pages over a versioned REST API with scoped tokens.
npm
CLI
Create monitors, read status and run one-off checks from the terminal or a CI job.
HCL
Terraform
A provider on the Terraform Registry, so monitors live in code next to the infrastructure they watch.
MCP
AI assistants
27 tools for Claude, ChatGPT and other MCP clients, from status questions to incident updates.
Deploys
GitHub
Deploy markers on response time charts and incidents, and checks paused during deploy windows.
macOS
Menu bar app
Monitor status and open incidents one click away.
How it works
Monitoring in 30 seconds.
01
Add what you want to watch
A URL, a hostname or a port, and the checks that matter for it.
02
Sentinel starts checking
From all 15 regions, as often as every 15 seconds depending on your plan.
03
A failure comes with evidence
Regional results, request phases and what each region saw, before anyone starts guessing.
04
The right people hear about it
Email, SMS, chat, PagerDuty or webhooks, and a status page when customers need to know too.
What can Sentinel monitor?
HTTP and HTTPS uptime, page content and keywords, server errors on pages that return 200, JSON APIs, SSL certificates, DNS records, domain expiration, ICMP ping, TCP ports, heartbeats, cron jobs, and x402 and MPP payment endpoints.
Who is Sentinel built for?
Developers, DevOps and platform teams who need an outside view of the sites and services they run. SaaS and ecommerce teams use it too, and so do agencies and freelancers looking after client sites.
Does every check run from all 15 regions?
Yes, on every plan, the free plan included. Every region checks once in each interval, so no region sits an interval out. The checks are spread out rather than sent together, so your server never gets fifteen requests in the same instant.
How does Sentinel decide a site is down?
An incident opens when at least half the regions fail and the failures include three or more regions on at least two continents. It closes once failures drop below that. Each monitor can also wait out short blips with its own confirmation and recovery periods.
How often does Sentinel check?
Every 15 seconds on Business, 30 seconds on Pro, one minute on Starter and five minutes on Free.
Can I manage Sentinel without the dashboard?
Yes. There's a REST API, a CLI, a Terraform provider and an MCP server for AI assistants, and a GitHub app that marks deploys and pauses checks during deploy windows.
How will I be alerted?
By email, SMS, Slack, Microsoft Teams, Discord, Telegram, PagerDuty or signed webhooks. You choose which channels each monitor uses.
Is there a free plan?
Yes. The free plan has 10 monitors and needs no card. Paid plans include a 14-day trial, and they need a card, trials included.
See your stack from the outside.
Your first monitor, checked from 15 regions, in 30 seconds.
No card for the free plan · 14-day trial on paid plans · Cancel anytime