# The alert center

Gather in a single feed, sorted by severity then by date, the clone alerts, the products declared stolen and the failing webhooks.

Source: https://docs.sealtrust.io/en/console/centre-alertes/

---

The signals that call for a reaction are scattered across several screens. The
alert center reads them in one go and presents them in a single feed, ordered
from the most severe to the most recent. You open it to know what asks for your
attention today.

![Console alert center, with its counters and its alert feed](/console/centre-alertes.en.webp)

## What the screen shows

Four counters at the top: "Critical alerts", "Stolen products", "System health"
and "Failed webhooks". The first adds up the critical and high clone alerts.
The second adds up the plain declarations and those whose token has been
frozen, freezing being an answer to theft without being its closure.

Two menus filter the feed: the type, with "All types", "Security", "System" and
"Webhooks", and the severity, with "All severities", "Critical", "High",
"Medium" and "Low".

Every card in the feed carries a colour bar on its left according to severity, a
family icon, a type pill, a severity pill, a title, a description and its age in
plain words: "just now" under a minute, then in minutes, in hours, in days and
finally in months. When the feed is empty, the page shows "No active alerts" and
"All systems are operating normally".

## What you can do here

Filter by family. "Security" gathers two sources, the clone alerts and the
products declared stolen. "Webhooks" only keeps the subscriptions with at least
one consecutive failure.

Filter by severity. The order retained is critical, high, medium, low, info.
The feed is sorted first by severity, then, at equal severity, from the most
recent to the oldest.

"Refresh" restarts the five reads of the page.

## What conditions access

The "Alerts center" entry lives in the "Analysis & Audit" section, with no plan
lock and no profile hiding, and the content waits until the account is
recognized as an administrator. Each source is bounded server-side to the
brands of your account.

On a brand console, the "System health" card shows "Operational". The health
check is not queried from this screen, so the "System" filter stays empty there.

## What the screen refuses

The five reads leave together and each one is isolated. The one that fails
simply leaves its family absent from the feed. The error banner only appears if
every read has failed.

An account with no accessible brand gets neither security statistics, nor clone
alerts, nor stolen products: those three reads fail and their families stay
absent from the feed. The webhook list, for its part, comes back empty without
an error.