Class AzureContentSafetyService

Namespace

FoundationaLLM.Gatekeeper.Core.Services

Assembly

FoundationaLLM.Gatekeeper.Core.dll

Implements the IContentSafetyService interface.

public class AzureContentSafetyService : IContentSafetyService

Inheritance

object

AzureContentSafetyService

Implements

IContentSafetyService

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Extension Methods

ObjectExtensions.GetPropertyValues(object, List<string>)

Constructors

AzureContentSafetyService(IOrchestrationContext, IHttpClientFactoryService, IOptions<AzureContentSafetySettings>, ILogger<AzureContentSafetyService>)

Constructor for the Azure Content Safety service.

public AzureContentSafetyService(IOrchestrationContext callContext, IHttpClientFactoryService httpClientFactoryService, IOptions<AzureContentSafetySettings> options, ILogger<AzureContentSafetyService> logger)

Parameters

callContext IOrchestrationContext

Stores context information extracted from the current HTTP request. This information is primarily used to inject HTTP headers into downstream HTTP calls.

httpClientFactoryService IHttpClientFactoryService

The HTTP client factory service.

options IOptions<AzureContentSafetySettings>

The configuration options for the Azure Content Safety service.

logger ILogger<AzureContentSafetyService>

The logger for the Azure Content Safety service.

Methods

AnalyzeText(string)

Checks if a text is safe or not based on pre-configured content filters.

public Task<AnalyzeTextFilterResult> AnalyzeText(string content)

Parameters

content string

The text content that needs to be analyzed.

Returns

Task<AnalyzeTextFilterResult>

The text analysis restult, which includes a boolean flag that represents if the content is considered safe. In case the content is unsafe, also returns the reason.

DetectPromptInjection(string)

Detects attempted prompt injections and jailbreaks in user prompts.

public Task<string?> DetectPromptInjection(string content)

Parameters

content string

The text content that needs to be analyzed.

Returns

Task<string>

The text analysis restult, which includes flags that represents if the content contains prompt injections or jailbreaks.