> ## Documentation Index
> Fetch the complete documentation index at: https://evalgate.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Read an alert

> What a fired alert tells you and where to go next.

# Read an alert

When an active monitor's threshold is crossed, EvalGate creates an alert and
delivers it to the channels you configured. Here is how to read one and what to
do next.

## What an alert contains

An alert records:

* the **monitor** and its **exact version** that fired,
* the **sample** and **monitor run** that crossed the threshold,
* the **traces** that contributed to the failure,
* the **failure topic** the alert was grouped into (if any).

## Step 1 — Triage

Open the alert. Look at the contributing traces — are they the same failure
repeated, or several different ones? Group them by failure category to find out.

## Step 2 — Inspect a representative trace

Open one contributing trace in the [Logs explorer](../traces/inspect-trace).
Confirm the failure is real and not a scorer false positive.

## Step 3 — Route the failure

From the alert or the trace, choose a next step:

* **Promote to case** — turn the failure into a test case.
* **Create review task** — hand it to a teammate.
* **Start an eval-loop workflow** — carry the failure end to end to a signed
  report.

## Step 4 — Resolve or snooze

When the failure is fixed or acknowledged, resolve the alert. If it is a known
issue you are tracking elsewhere, snooze it with a reason. Resolved and snoozed
states are recorded with the actor and rationale.

## Up next

Let the [Copilot](../copilot/ask-copilot) draft a fix for the failure.