> For the complete documentation index, see [llms.txt](https://docs.umbraco.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.umbraco.com/ai-in-umbraco/19.latest/reference/models/ai-guardrail.md).

# AIGuardrail

Model representing an AI guardrail with evaluation rules.

Represents a collection of rules that evaluate AI inputs and responses for safety, compliance, and quality enforcement.

## Namespace

```csharp
using Umbraco.AI.Core.Guardrails;
```

## Definition

{% code title="AIGuardrail" %}

```csharp
public sealed class AIGuardrail : IAIVersionableEntity
{
    public Guid Id { get; internal set; }
    public required string Alias { get; set; }
    public required string Name { get; set; }
    public IList<AIGuardrailRule> Rules { get; set; } = [];

    // Audit properties
    public DateTime DateCreated { get; init; } = DateTime.UtcNow;
    public DateTime DateModified { get; set; } = DateTime.UtcNow;
    public Guid? CreatedByUserId { get; set; }
    public Guid? ModifiedByUserId { get; set; }

    // Versioning
    public int Version { get; internal set; } = 1;
}
```

{% endcode %}

## Properties

| Property           | Type                     | Description                                     |
| ------------------ | ------------------------ | ----------------------------------------------- |
| `Id`               | `Guid`                   | Unique identifier (auto-generated)              |
| `Alias`            | `string`                 | Unique alias for programmatic lookup (required) |
| `Name`             | `string`                 | Display name (required)                         |
| `Rules`            | `IList<AIGuardrailRule>` | Ordered collection of evaluation rules          |
| `DateCreated`      | `DateTime`               | When created (UTC)                              |
| `DateModified`     | `DateTime`               | When last modified (UTC)                        |
| `CreatedByUserId`  | `Guid?`                  | User who created                                |
| `ModifiedByUserId` | `Guid?`                  | User who last modified                          |
| `Version`          | `int`                    | Current version number                          |

## Example

{% code title="Example" %}

```csharp
var guardrail = new AIGuardrail
{
    Alias = "content-safety",
    Name = "Content Safety Policy",
    Rules =
    [
        new AIGuardrailRule
        {
            EvaluatorId = "contains",
            Name = "Block competitor brand mentions",
            Phase = AIGuardrailPhase.PostGenerate,
            Action = AIGuardrailAction.Block,
            Config = JsonSerializer.SerializeToElement(new { searchPattern = "CompetitorBrand", ignoreCase = true }),
            SortOrder = 0
        },
        new AIGuardrailRule
        {
            EvaluatorId = "regex",
            Name = "Redact email addresses",
            Phase = AIGuardrailPhase.PreGenerate,
            Action = AIGuardrailAction.Redact,
            Config = JsonSerializer.SerializeToElement(new { pattern = @"[a-zA-Z0-9._%+-]+@[a-zA-Z0-9.-]+\.[a-zA-Z]{2,}", ignoreCase = true }),
            SortOrder = 1
        },
        new AIGuardrailRule
        {
            EvaluatorId = "llm-judge",
            Name = "Quality check",
            Phase = AIGuardrailPhase.PostGenerate,
            Action = AIGuardrailAction.Warn,
            Config = JsonSerializer.SerializeToElement(new { evaluationCriteria = "Check brand compliance", safetyThreshold = 0.7 }),
            SortOrder = 2
        }
    ]
};
```

{% endcode %}

## Related

* [AIGuardrailRule](#aiguardrailrule) - Rule model
* [IAIGuardrailService](/ai-in-umbraco/19.latest/reference/services/ai-guardrail-service.md) - Guardrail service

***

## AIGuardrailRule

Represents a single evaluation rule within a guardrail.

## Definition

{% code title="AIGuardrailRule" %}

```csharp
public sealed class AIGuardrailRule
{
    public Guid Id { get; internal set; }
    public required string EvaluatorId { get; init; }
    public required string Name { get; set; }
    public AIGuardrailPhase Phase { get; set; } = AIGuardrailPhase.PostGenerate;
    public AIGuardrailAction Action { get; set; } = AIGuardrailAction.Block;
    public JsonElement? Config { get; set; }
    public string? GuardrailName { get; set; }
    public int SortOrder { get; set; }
}
```

{% endcode %}

## Properties

| Property        | Type                | Description                                                      |
| --------------- | ------------------- | ---------------------------------------------------------------- |
| `Id`            | `Guid`              | Unique identifier (auto-generated)                               |
| `EvaluatorId`   | `string`            | Registered evaluator ID (required, immutable)                    |
| `Name`          | `string`            | Display name (required)                                          |
| `Phase`         | `AIGuardrailPhase`  | When to evaluate (default: PostGenerate)                         |
| `Action`        | `AIGuardrailAction` | What to do when flagged: Block, Warn, or Redact (default: Block) |
| `Config`        | `JsonElement?`      | Evaluator-specific configuration                                 |
| `GuardrailName` | `string?`           | Name of the parent guardrail (set during resolution)             |
| `SortOrder`     | `int`               | Controls evaluation order within the guardrail                   |

## Enums

### AIGuardrailPhase

{% code title="AIGuardrailPhase" %}

```csharp
public enum AIGuardrailPhase
{
    PreGenerate = 0,   // Before sending to AI provider
    PostGenerate = 1   // After receiving AI response
}
```

{% endcode %}

### AIGuardrailAction

{% code title="AIGuardrailAction" %}

```csharp
public enum AIGuardrailAction
{
    Block = 0,   // Block content and throw AIGuardrailBlockedException
    Warn = 1,    // Allow content through unchanged, log a warning
    Redact = 2   // Replace flagged content with [REDACTED]
}
```

{% endcode %}


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.umbraco.com/ai-in-umbraco/19.latest/reference/models/ai-guardrail.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.