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

# Advisories

> Proactive issue detection and prioritized alerts across your cluster fleet.

Advisories surface potential problems before they become incidents. Pastures continuously evaluates cluster health signals and presents findings as severity-ranked cards with one-click AI diagnosis.

## How it works

The Advisories page polls the Engine for known issue patterns, version EOL timelines, and configuration anti-patterns. Each advisory includes:

* **Severity** — color-coded `CRITICAL`, `HIGH`, or `MEDIUM`
* **Affected component** — the specific resource or subsystem involved
* **Recommended action** — a brief remediation summary
* **Time detected** — when the issue was first observed

Every card includes a **Diagnose →** link that opens the [Pastures AI drawer](/features/pastures-ai) pre-filled with the advisory title, so you can get a grounded fix immediately.

<Info>
  In demo mode, advisories are generated from synthetic data. Connect a Pastures Engine to receive live advisories from your clusters.
</Info>

## Engine API

<ParamField path="GET /api/advisories" type="endpoint">
  Returns the current list of active advisories for the selected cluster.
</ParamField>

## Demo data

When no Engine is connected, the following advisories are shown:

| Severity | Advisory                                              | Component |
| -------- | ----------------------------------------------------- | --------- |
| CRITICAL | etcd cluster has only 1 member — no quorum redundancy | etcd      |
| HIGH     | K3s version reaches EOL in 45 days                    | K3s       |
| MEDIUM   | Longhorn replica count below recommended minimum      | Longhorn  |

## Diagnose → integration

Clicking **Diagnose →** on any advisory card opens the Pastures AI drawer with the advisory title as the initial prompt. The AI response includes source citations, confidence scoring, and expandable fix commands tailored to the specific issue.