Status
What is up, measured, and what we promise about it.
The health check below is live and hits the one dependency that matters, the key store. The availability commitments underneath are definitions, not decoration: they say what counts, what does not, and what you get back when we miss.
unknownHosted endpointthe health check itself did not answer
— msKey store round tripunknown
—Uptime historyNone yet. This service is new; a history we do not have is not one we will invent
Availability commitments
The SLA, defined.
| Tier | Commitment | What counts as down | Credit |
|---|---|---|---|
| Sandbox | None. Best effort, no credit. | — | — |
| Operator | None. Best effort, no credit. | — | — |
| Operator Pro | Priority support. No availability target yet: /status publishes live health, and we will not publish a target until we have twelve months of it | Not an availability SLA. Priority support means a human reply within one working day. | — |
| Enterprise | Availability commitment and service credits are set in the contract, against the live record on /status; no target is published until we have twelve months of it | The intent endpoint returning 5xx or not answering within 10 seconds, for 5 consecutive minutes, measured from two regions, excluding scheduled maintenance announced 48 hours ahead and excluding refusals we make on purpose (rate limits, private addresses, robots.txt). | 10% of the monthly fee per 0.1% below 99.9%, capped at 50%, claimed within 30 days. |
The health endpoint is /api/health. Incidents will be posted on this page. There have been none, because there has not been time for any; that is a statement about age, not about quality.