Platform status
Each card says what it actually measured — a live query, or only that a component is wired up. The API caches the report for about thirty seconds, so the stamp can be up to half a minute old. Nothing here is a green light we didn't earn.
API
Answering requests with a complete deployment.
That the API Worker answered this request, and that every binding it cannot serve without resolved in the running deployment.
Database
Answered a read.
A live read against the D1 database that stores agents, runs, and approvals.
File storage
Answered a lookup.
A live metadata lookup against the R2 bucket holding run artifacts and uploads.
Integration catalog
Answered a health call.
A live call to the tools gateway, the Worker that serves the integration catalog to the app.
Background runs
Bindings resolved.
Only that the background Worker's bindings resolved. It takes no public route, so this row is not a live check of whether runs are executing.
What these checks are
This page asks the API for its status report. The checks behind it are real: a read against the database, a metadata lookup against object storage, and a call to the Worker that serves the integration catalog. The app caches the result for about thirty seconds, so a second load within that window reuses the same stamp rather than re-running every probe.
One card is honest about being weaker than the rest. Background runs execute on a Worker with no public route, so we can confirm it is attached to this deployment but cannot ask whether runs are flowing. It reads Not measured rather than borrowing a green badge, and is excluded from the headline above. A degraded integration catalog means the dedicated catalog Worker is unavailable and the app is reading the vendor catalog directly — slower, but working, which is why it never shows as down.
Coverage
This page reports the present moment only. We do not publish uptime percentages or a ninety-day history, because nothing records one yet — we would rather show nothing than a chart we haven't been keeping. There is no incident log here either.
If an incident touches your data or your runs, we notify you directly by email. See the Trust Center for how we handle that, and Security to report a vulnerability.