> ## 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

> Configure MCP servers, guardrail rules, and SLM-based guardrails for the gateway.

## What this is for

**Configuration → Security** is where you control what the gateway is allowed to do beyond plain LLM inference: which MCP servers it exposes to clients, which content/intent rules it enforces on every request, and which on-gateway SLM (small language model) classifiers run as additional guards.

The page has three tabs: **MCP**, **Guardrails**, and **SLM Guardrails**.

## MCP

Register the Model Context Protocol servers your gateway should make available to clients. Once registered, MCP servers are scoped per API key on the [API Keys](/platform/configuration/api-keys) MCP tab.

<Frame caption="Security → MCP">
  <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-mcp.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=7d52a6d8ad52b785589d224298b6f760" alt="Registered MCP servers" width="2628" height="1146" data-path="images/screenshots/platform/configuration/security-mcp.png" />
</Frame>

### Options

The **Register MCP Server** dialog has two tabs: **Template** for ready-made configurations and **Manual** for free-form setup.

| Field                  | Notes                                                                                                               |
| ---------------------- | ------------------------------------------------------------------------------------------------------------------- |
| **Server Name**        | Display name (e.g. `My MCP Server`).                                                                                |
| **Server Type**        | `STDIO`, `HTTP`, or `SSE`.                                                                                          |
| **Command**            | STDIO only (e.g. `npx`).                                                                                            |
| **Arguments**          | STDIO only, comma-separated (e.g. `-y, @modelcontextprotocol/server-filesystem, /tmp`).                             |
| **URL**                | HTTP/SSE only — the server endpoint.                                                                                |
| **Authentication**     | `None`, `API Key`, `Bearer Token`, `Basic Auth`, or `OAuth2`. Reveals secret/credential fields based on the choice. |
| **Additional Headers** | Optional JSON for advanced HTTP/SSE setups.                                                                         |

### How to configure

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

  <Step title="Register a server">
    Click **Register MCP Server**. Pick a **Template** to autofill a known integration (filesystem, github, etc.) or switch to **Manual**.

    <Frame caption="Register MCP Server">
      <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-mcp-register.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=ff2c2cc37a9965726093fb53b4a06a64" alt="Register MCP Server dialog" width="1066" height="1500" data-path="images/screenshots/platform/configuration/security-mcp-register.png" />
    </Frame>
  </Step>

  <Step title="Pick the transport">
    Choose **STDIO** for processes the gateway spawns, or **HTTP** / **SSE** for remote MCP servers. Fill the matching fields.
  </Step>

  <Step title="Set authentication">
    Select an auth scheme and provide the secret. Use **None** only for local STDIO servers on the gateway host.
  </Step>

  <Step title="Save and scope to keys">
    The server appears in the list with status `connected` once the gateway can reach it. Open [API Keys](/platform/configuration/api-keys), edit a key, and use the **MCP** tab to scope the key to specific servers and tools.
  </Step>
</Steps>

## Guardrails

Rule-based content controls applied to every request that flows through the gateway. Use these to redact PII, block prompt injection patterns, enforce keyword filters, or just log policy hits.

<Frame caption="Security → Guardrails">
  <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-guardrails.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=858907c7d0531fe3f44eee3304b06709" alt="Guardrail rules list" width="2622" height="1164" data-path="images/screenshots/platform/configuration/security-guardrails.png" />
</Frame>

### Options

**Create Guardrail Rule** dialog:

| Field                         | Notes                                                                                                                        |
| ----------------------------- | ---------------------------------------------------------------------------------------------------------------------------- |
| **Rule Name**                 | e.g. `PII Redaction`.                                                                                                        |
| **Apply To**                  | `Input Only`, `Output Only`, or `Input & Output`.                                                                            |
| **Category**                  | `Content Filter`, `PII Detection`, `Prompt Injection`, `Topic Block`, `Regex Filter`, `Token Limit`, `Cost Limit`, `Custom`. |
| **Action**                    | `Block — reject the request`, `Warn — Allow but flag`, `Redact — replace matched content`, `Log — Record only`.              |
| **Description**               | Free-form notes.                                                                                                             |
| **Apply to MCP**              | Toggle. When on, the rule also runs against MCP traffic on this gateway.                                                     |
| **Keywords to block**         | One per line — for keyword-style categories.                                                                                 |
| **Regex patterns (optional)** | One per line — e.g. `\b\d{3}-\d{2}-\d{4}\b`.                                                                                 |
| **Ignore case**               | Toggle for keyword/regex matches.                                                                                            |

Rules are listed with **Name**, **Category**, **Type** (input / output / both), **Action**, **Priority**, **Triggers**, and **Enabled**. A built-in **Test** dialog lets you paste sample text and see which rules fire.

### How to configure

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

  <Step title="Create a rule">
    Click **Create Guardrail Rule**. Set **Rule Name**, pick **Apply To**, **Category**, and **Action**. Add **Keywords** and/or **Regex patterns** for keyword-style categories.

    <Frame caption="Create Guardrail Rule">
      <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-guardrails-create.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=6b1305983deff54688d05fb62412cf26" alt="Create Guardrail Rule dialog" width="1028" height="1642" data-path="images/screenshots/platform/configuration/security-guardrails-create.png" />
    </Frame>
  </Step>

  <Step title="Test before enabling">
    Use the **Test** action on the rule row, paste representative sample text, and confirm the rule matches what you expect.
  </Step>

  <Step title="Toggle Enabled">
    Flip **Enabled** to turn the rule on. Trigger a matching request from [Playground](/platform/playground) and verify the action in [Logs](/platform/logs).
  </Step>
</Steps>

## SLM Guardrails

On-gateway small language models that classify, score, detect, or rewrite content. Use these when keyword/regex rules aren't expressive enough — toxicity classifiers, jailbreak detectors, intent scorers.

<Frame caption="Security → SLM Guardrails">
  <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-slm.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=1313d9dd191b47f0f5729fffd78e1730" alt="SLM guardrails list" width="2628" height="976" data-path="images/screenshots/platform/configuration/security-slm.png" />
</Frame>

### Options

**Deploy SLM Guardrail** dialog:

| Field            | Notes                                                                      |
| ---------------- | -------------------------------------------------------------------------- |
| **Name**         | e.g. `Toxicity Classifier`.                                                |
| **Model**        | Pick a curated model or enter a model ref (e.g. `guardway/model-name-v1`). |
| **Type**         | `classifier`, `scorer`, `detector`, or `rewriter`.                         |
| **Apply To**     | `Input Only`, `Output Only`, or `Input & Output`.                          |
| **Action**       | `block`, `warn`, `rewrite`, or `log`.                                      |
| **Threshold**    | Decision threshold (slider).                                               |
| **Description**  | What this model detects.                                                   |
| **Labels**       | Comma-separated — e.g. `safe, toxic, harmful`.                             |
| **Apply to MCP** | Toggle.                                                                    |

Each row exposes **Avg latency**, **Accuracy**, and **Evaluations** so you can see real performance before promoting it to a blocking action.

### How to configure

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

  <Step title="Deploy a model">
    Click **Deploy SLM Guardrail**, pick a **Model** and **Type**, set **Apply To**, **Action**, and **Threshold**.

    <Frame caption="Deploy SLM Guardrail">
      <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/configuration/security-slm-deploy.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=fb040723ac80cb9a3271be5578fb0f36" alt="Deploy SLM Guardrail dialog" width="886" height="1320" data-path="images/screenshots/platform/configuration/security-slm-deploy.png" />
    </Frame>
  </Step>

  <Step title="Start in log mode">
    Set **Action** to `log` first. Run real traffic for a window, watch **Accuracy** and **Evaluations** on the row.
  </Step>

  <Step title="Promote to enforcement">
    Once you trust the model, change **Action** to `warn`, `block`, or `rewrite`. Verify in [Logs](/platform/logs).
  </Step>
</Steps>

<Tip>
  SLM Guardrails run on the gateway, not in the cloud. Inference latency depends on the gateway host's available GPU/CPU.
</Tip>

## Related

* [API Keys](/platform/configuration/api-keys) — scope MCP servers and tools per key.
* [Logs](/platform/logs) — every guardrail trigger is logged with the rule that matched.
* [Notifications](/platform/settings/notifications) — subscribe to `guardrail.violation` events.