Status
Service health
Checked live when you load this page — not read from a dashboard someone remembered to update.
| Service | Endpoint | Status | Response |
|---|---|---|---|
| Probing services… | |||
Machine‑readable: /status.json. Each service also exposes its own
/health. Response time is measured from our edge, so it includes the tunnel but not your
network.
Registry health
The index, and how much of it is alive.
Every indexed service is re‑probed on a 30‑minute cycle. A sweep age above one hour means the verification worker needs attention — it does not mean the services listed are down.
How we run
Independent hardware, stated plainly.
Boris Inc runs on machines we own, in one location, behind Cloudflare tunnels. Not a cloud region, not a Kubernetes cluster, not somebody else's managed platform.
That is a deliberate trade. It is why inference costs what it costs and why there is no per‑account minimum. It also means a power cut is a power cut. If your agent needs five nines, route through Aegis with failover rather than pointing at a single endpoint — ours included.
If something is down
- Paid calls use auth‑then‑capture: an authorisation that never delivers is never captured, so a failed call does not cost you.
- Routing through
/routeon Aegis fails over to the next‑best trusted upstream automatically. - Free
/sample/*and/trial/*endpoints are the fastest way to check whether a service is answering.