See it work.
One outage, from the first failed check to the client's status page — and then the view an agency gets across every client. Screen by screen.
Every picture is the real app, captured from a running copy. The companies and people in them are invented.
All three regions agree. Now it's worth waking someone.
Alderbank's online banking starts answering 503. Amsterdam, Virginia and Singapore all see it, so this is an incident — and the dashboard says so in one line, with who is on call and how long it's been.
What each probe saw, and where to look first.
The incident keeps what the probes captured when it opened — status, headers, the first 2 KB of the body, and the last healthy check to compare against. Mercury reads that evidence and writes what it points to. It's labelled as a hypothesis, with a confidence, because that's what it is.
- Acknowledge from the page, from the email, or from the Slack message — one signed link, no login at 3am.
- If nobody acknowledges, the next person in the rotation is paged. Held alerts never escalate.
- Every decision — this page, and every time the engine chose not to page — is kept with its reason.
One region timing out isn't an outage. It's written down instead.
Brightfell's dashboard is slow to reach from Singapore a few times a week. Amsterdam and Virginia are fine in the same minute, so nobody is paged — and every one of those decisions is in the noise ledger, so you can check our working.
Everything you watch, in words a person would use.
HTTP with assertions, TCP ports, DNS records, certificates, heartbeats for cron jobs. Each row says what kind of check it is, how it's doing over 30 days, how often it runs and how many regions have to agree.
Sign in, search, get a price. If any step fails, that's the alert.
Up to eight requests in a row from each region. A step saves a value from its answer — a token, a room id — and later steps use it. Saved values are masked in incidents, a variable can never choose the host, and redirects aren't followed.
{{{{token}}}}; steps 2 and 3 send it as a header; step 2 saves the room that step 3 prices. Team plan and up.Your customers get a page. Not your dashboard.
Every workspace has a public status page, on your own hostname if you like, with your logo and colour. It shows only the monitors you choose, ninety days of history, and the updates you publish — Mercury drafts them, a person presses publish.
The right person, at the hour their shift actually starts.
Rotations hand off at 09:00 in your timezone, and stay there across daylight saving. Holidays are covered without editing the rotation. Alerts go to Slack, Discord, Microsoft Teams, email, your own webhook, and push on your phone.
Forty sites that aren't yours, and one place to watch them.
Northwind Digital looks after six clients. Each has its own workspace, status page and team — all under Northwind's one Agency subscription, sharing its monitors as a pool. How agencies use Repose →
Check a site of your own from three continents right now, or start free — 25 monitors, commercial use included, no card.