Overview
The Overview is the landing page of the admin console. It is built around one question — is anything wrong, and what do I do about it — rather than around a wall of charts. Everything on it is scoped to a time window you choose at the top right: last hour, last 24 hours, or last 7 days.
Needs your attention
The first section, and the only one that disappears when there is nothing to say. It lists things with a deadline attached:
- Budgets approaching refusal — a budget that will start blocking traffic
- Providers pulled from rotation — a provider the circuit breaker has opened
Entries are ordered by how soon they bite, not by percentage. A budget at 92% with a week left is calmer than one at 71% burning through today, and sorting by the number would put them the wrong way round. Each entry carries a direct link to the page where you would fix it.
If this section is absent, nothing is pressing.
The stat strip
Four figures for the selected window:
| Figure | Description |
|---|---|
| Requests | Total requests through the gateway in the window |
| Spend | Estimated cost accrued in the window |
| Failed | Requests that did not return a usable response |
| Blocked by filters | Requests a guard refused |
Failure and block counts are shown as counts rather than rates, because a rate hides how much is happening — 2% of ten requests and 2% of ten thousand are not the same morning.
What’s failing
Recent errors and refusals, with the failover trail for each: which provider was tried, in what order, and what each one said. This is the section that answers “the request failed, but which provider actually refused it and why” — the question a single error message on its own cannot.
Safety filters
Guard activity split into blocks and watches: what was refused, and what was flagged at a monitoring level without being stopped. Keeping the two apart matters, because a guard in monitor mode producing a lot of activity is a configuration decision waiting to be made, not an incident.
Spend by routing rule
Cost attributed to the routing rule that chose the provider, not to the provider itself. That is deliberate: the provider tells you where the money went, and the routing rule tells you which decision sent it there — which is the one you can change.
Budgets
Current budget consumption across the scopes that have one, so the numbers behind the attention list are visible even when nothing is close enough to be flagged.
What it does not show
There is no provider health map, request-volume chart, or recent-requests table on this page. Provider health lives on the Providers page beside the controls for acting on it, and full request history — including failed and abandoned requests — is on the Request Log.