> ## Knowledge Base Index
> Fetch the complete knowledge base index at: https://doc.ecomx.co/sitemap.xml
> Use this file to discover available pages before exploring further.
> Pure-Markdown content can be obtained by appending a '.md' suffix to the content URLs listed in the sitemap (without the trailing slash).

# Alerts and Settings

## Configure email alerts

1. Enter a valid address in **Alert email**.
2. Select **Enable email alerts**.
3. Select one or more categories:

* **AI channel traffic anomalies**
* **Global Catalog exposure anomalies**
* **Repeated health-check failures**
4. Select **Save settings**.

The master **Enable email alerts** setting must be selected in addition to each category you want.

AgentWatch sends one deduplicated notification when an incident:

* Opens.
* Escalates from Warning to Critical.
* Resolves automatically after recovery.

Manually resolving an incident does not send an automatic-recovery email. AgentWatch also does not send a message for every monitoring run or every evidence update.

If Settings says the production email provider is not configured, contact AgentWatch Support. That configuration cannot be fixed from the merchant UI.

## Monitoring operations

### Sync 30 completed days

Reads aggregate analytics for the latest 30 completed days and safely updates stored metrics. Existing good data is retained when an upstream source is temporarily unavailable.

### Run health checks

Runs the seven read-only health checks and records a new health run. You can also run the same operation from the Health page.

Each type of manual operation has a one-minute cooldown. Wait for the current operation to finish before starting another one.

## Recent runs

The table shows the 10 most recent manual and scheduled monitoring jobs:

* **Succeeded:** the operation completed successfully.
* **Partial:** at least one source succeeded and at least one source was unavailable; previously good data was retained.
* **Failed:** the operation did not complete.
* **Running:** the operation is still in progress.

The Records column shows how many records or checks the job processed. Trigger identifies whether the operation was manual, scheduled, or installation-based.

If a scheduled run fails, wait for the next run or retry the equivalent manual operation. Persistent failures should be reported with the relevant incident diagnostic packet or exact run timestamp.

## Delete AgentWatch data

The **Delete AgentWatch data** action permanently deletes:

* Aggregate store and channel metrics.
* Product exposure history.
* Health-check history.
* Incidents, evidence, and alert-delivery records.
* Monitoring job history.

You must select the confirmation checkbox before deletion. This action does not uninstall AgentWatch. Monitoring history can begin accumulating again after a future sync or health run.

Deletion cannot be undone. Export or copy any incident details you need before confirming.

Next: [Pricing and billing](https://help.ecomx.co/en/articles/17607674-pricing-and-billing)