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

# Security

> Guardrail activity, severity breakdowns, and MCP traffic across every attached gateway.

## What this is for

**Dashboard → Security** summarizes guardrail evaluations and MCP traffic from every attached gateway. Use it to spot bursts of blocks, find the most-triggered policies, and confirm MCP servers are getting the volume you expect.

For request and token volume, see [Usage](/platform/dashboard/usage). For cost, see [Spend](/platform/dashboard/spend). To inspect individual events, open [Logs](/platform/logs) on the relevant gateway.

## Options

### Time range

`Last 24 Hours`, `Last 7 Days`, `Last 30 Days`. Comparisons against the previous window appear next to each stat.

### Filters

* **Gateway** — multi-select; leave empty for all attached gateways.

### Stat blocks

| Block               | Metric                                                    |
| ------------------- | --------------------------------------------------------- |
| **Total Blocks**    | Requests blocked by a guardrail in the window.            |
| **Total Events**    | Every guardrail evaluation (block + warn + redact + log). |
| **MCP Requests**    | Total MCP tool calls across all attached servers.         |
| **MCP Avg Latency** | Mean MCP tool-call latency in milliseconds.               |

### Charts and breakdowns

* **Blocks over time** — bucketed by the selected range.
* **MCP requests over time** — bucketed by the selected range.
* **By severity** — `Critical`, `High`, `Medium`, `Low` event counts.
* **Triggered on** — top categories that fired (e.g. PII detection, prompt injection, content filter).
* **Event type** — top event types observed.
* **Top MCP servers** — top 5 servers by request volume.
* **Top MCP tools** — top 5 tools by request volume.

If no guardrails fired and no MCP calls flowed in the window, the page shows an empty state instead of charts.

## How to configure

This is a read-only analytics view — guardrail policies and MCP servers are configured on [Configuration → Security](/platform/configuration/security).

<Frame caption="Dashboard → Security">
  <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/dashboard/security.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=41d3d9a2d9c44572d84eafa697ae0d55" alt="Security analytics dashboard" width="3438" height="1892" data-path="images/screenshots/platform/dashboard/security.png" />
</Frame>

<Steps>
  <Step title="Open Dashboard → Security">
    Open **Dashboard → Overview** from the dashboard sidebar and click the **Security** tab.
  </Step>

  <Step title="Pick a window and gateways">
    Set **Time range** and optionally narrow to specific gateways with the **Gateway** filter.
  </Step>

  <Step title="Drill into a triggered category">
    Use the **Triggered on** breakdown to identify which guardrail fired most. Open [Logs → Guardrails](/platform/logs) on the affected gateway to inspect individual evaluations.
  </Step>

  <Step title="Tune the policy">
    If a guardrail is over-firing or under-firing, adjust thresholds and actions on [Configuration → Security](/platform/configuration/security).
  </Step>
</Steps>

## Related

* [Configuration → Security](/platform/configuration/security) — guardrail policies, SLM Guardrails, MCP servers.
* [Logs](/platform/logs) — guardrail evaluation and MCP request logs on a specific gateway.
* [Notifications](/platform/settings/notifications) — fire on `guardrail.violation`.