OpenStatus

Promptfree, for everyone, and the only version there is

Open the prompt in Claude CodeOpen the prompt in CodexOpen the prompt in Cursor

Copy

Build monitoring that runs somewhere other than the thing it watches.

THE ONE RULE
- The checker and the status page must not live on the infrastructure they report on. If they do, an outage takes both

STACK
- A tiny worker on a different provider from your application, SQLite, a static status page on a third

CHECKS
- HTTP with a timeout, an expected status and an optional body match
- Every check writes a row whether or not anything changed, so "stable" and "we stopped looking" stay different questions
- Two consecutive failures before declaring down; one failure is usually a network blip

INCIDENTS
- Opened automatically on a confirmed failure, closed by a human
- Updates appended, never edited, each with a timestamp

THE PUBLIC PAGE
- Static, rebuilt on every state change, served from a different provider again
- Ninety days of uptime per service, computed from the check rows

SUBSCRIBERS
- Email and webhook on incident open, update and close, with double opt-in

Include the checker, the schema and the deploy topology.