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

# Guardrail Rules

> Configure guardrail rules in the AI Gateway to validate and control LLM requests and responses with evaluators triggered by CEL conditions.

Guardrail Rules define conditions under which [**Evaluators**](/ai-studio/optimize/evaluators) (automated checks that inspect a request or response) run against requests passing through the [**AI Gateway**](/ai-gateway/get-started/introduction). Conditions are written as CEL expressions, built visually with the Rule Builder covered below. A guardrail is only triggered when its rule conditions are matched, not on every request.

Guardrail Rules run on LLM requests and responses only. They do not run on tool calls served through an [**MCP Gateway**](/ai-gateway/mcp-portal/mcp-gateways); the [plugins attached to an MCP Gateway](/ai-gateway/mcp-portal/mcp-gateways#plugins) are the control for MCP tool traffic.

## Use cases

Guardrail rules are most useful when the same safety or compliance check needs to apply consistently across many requests.

<AccordionGroup>
  <Accordion title="Add jailbreak protection to all customer-facing banking traffic" icon="shield">
    Runs a jailbreak detection [**Evaluator**](/ai-studio/optimize/evaluators) on all customer-facing requests at the gateway level, adding an extra security layer across **AI Gateway** traffic.
  </Accordion>

  <Accordion title="Enforce GDPR PII checks across the entire workspace" icon="lock">
    Enforces GDPR compliance by running PII detection on all matching requests workspace-wide from a single rule.
  </Accordion>

  <Accordion title="Validate customer detail access for the sales team" icon="users">
    Validates customer detail access for the sales team by calling an external [**Evaluator**](/ai-studio/optimize/evaluators) on every matching request before it reaches the model.
  </Accordion>

  <Accordion title="Monitor tone of voice across all company traffic" icon="waveform">
    Applies a tone of voice [**Evaluator**](/ai-studio/optimize/evaluators) at the gateway level so every response is checked against the company's tone guidelines.
  </Accordion>

  <Accordion title="Apply EU AI Act compliance checks to EU-routed requests" icon="scale-balanced">
    Runs a compliance [**Evaluator**](/ai-studio/optimize/evaluators) on EU-routed requests only, scoped using the Rule Builder so the guardrail applies exactly where it is needed without affecting other traffic.
  </Accordion>

  <Accordion title="Run safety and quality checks on targeted traffic using metadata" icon="robot">
    Runs jailbreak detection and response relevance [**Evaluators**](/ai-studio/optimize/evaluators) at 50% sample rate each, scoped to specific traffic using a metadata condition in the Rule Builder.
  </Accordion>
</AccordionGroup>

## Visibility

* Visible to workspace administrators only.

## Creating a guardrail rule

From the **Guardrail Rules** list, click <kbd className="key"><Icon icon="circle-plus" color="#fff" /> Add New Rule</kbd>. A panel opens on the right with the following fields.

<Frame caption="Create Guardrail Rule panel in the AI Gateway.">
  <img src="https://mintcdn.com/orqai/dycyWau_l9E63fTw/images/guardrail-rule-create.png?fit=max&auto=format&n=dycyWau_l9E63fTw&q=85&s=ab8de239478b0add2f9c33961145aa77" alt="Create Guardrail Rule" width="1525" height="1217" data-path="images/guardrail-rule-create.png" />
</Frame>

### General

| Field           | Description                                                                                                   |
| --------------- | ------------------------------------------------------------------------------------------------------------- |
| **Rule Name**   | A display name for the guardrail rule.                                                                        |
| **Description** | Optional context for administrators.                                                                          |
| **Enable Rule** | Toggle to activate or deactivate the rule. When enabled, the rule is active and applied to matching requests. |

### Evaluators

Select the [**Evaluators**](/ai-studio/optimize/evaluators) to run when this rule is triggered. Click <kbd className="key"><Icon icon="circle-plus" color="#fff" /> Add</kbd> to attach one or more evaluators from the scoped project. The Add menu groups options into **System** and **Workspace**: Workspace evaluators are the project's custom evaluators; System guardrails are covered below.

#### System Guardrails

System Guardrails are the pre-built checks **Orq.ai** maintains in the Evaluators list's System group. Attach one and it runs immediately as a pass/fail check that can block a request; it never rewrites content.

| System Guardrail     | What it checks                                                                                                                   | Configurable                                                                                                                                                                        |
| -------------------- | -------------------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **PII Detection**    | Detects personally identifiable information in requests or responses and blocks the request by default, instead of redacting it. | Yes. Once added, a <Icon icon="sliders" /> icon appears next to it in the Evaluators list. Click it to set `language`, `regions`, `entities`, `threshold`, and `entity_thresholds`. |
| **Secret Detection** | Detects API keys, tokens, credentials, and other secrets in requests or responses and blocks the request by default.             | No configurable options.                                                                                                                                                            |

Toggle the <Icon icon="shield" /> icon next to a System Guardrail in the Evaluators list to switch it to monitoring-only mode. The same check still runs and its result is still recorded, but a match no longer blocks the request.

The detectable entity catalog for **PII Detection** is shared with the [**PII Redaction**](/ai-gateway/features/plugins/pii-redaction#supported-entity-types) plugin, and is region-scoped rather than language-scoped. `GET /v2/pii/capabilities` is the live source of truth for the supported regions, base and regional entity types, and the `region_entities` mapping. A few [entity type names changed](/ai-gateway/features/plugins/pii-redaction#entity-type-names-that-changed) when the catalog moved to regions; the old keys are rejected at write time.

Each guardrail runs on the request **Input**, the model **Output**, or both. Click the icon next to a guardrail in the Evaluators list to cycle through <kbd><Icon icon="arrow-up-right" color="#22c55e" /></kbd> **Input**, <kbd><Icon icon="arrow-down-left" color="#ef4444" /></kbd> **Output**, and <kbd><Icon icon="up-right-and-down-left-from-center" /></kbd> **Both**, set independently per guardrail.

A rule cannot set a per-guardrail `timeout`. Guardrails injected by a rule always run with the 60 second default. To bound a Guardrail explicitly, attach it inline on the request instead; see [Guardrail timeout](/ai-gateway/configuration/guardrails#timeout).

<Warning>
  System Guardrails always fail closed: if the underlying check errors (for example the detection service is unavailable or the call itself fails), the guardrail blocks the request instead of letting it through unchecked.
</Warning>

### Rule Builder

The Rule Builder constructs the match conditions that determine when the guardrail is triggered. Clicking <kbd className="key"><Icon icon="circle-plus" color="#fff" /> Add</kbd> opens a dropdown with the following condition types:

| Condition    | Description                                |
| ------------ | ------------------------------------------ |
| **Header**   | Match on a request header name and value.  |
| **Model**    | Match on the model being called.           |
| **Identity** | Match on the identity making the request.  |
| **Metadata** | Match on metadata attached to the request. |
| **Project**  | Match on the project scope of the request. |

You can also click **Add group** to nest conditions into a logical group. Multiple conditions within a group are joined with `and`. Groups themselves can be joined with either `and` or `or`. Select the operator between groups to control how they combine. Each condition can be removed with <kbd><Icon icon="xmark" /></kbd>.

The builder generates a CEL (Common Expression Language) expression shown read-only in the **CEL Expression Preview** below. The guardrail is only triggered when the expression evaluates to true.