OpenAIModelProvider

Namespace AgentPrism · Assembly AgentPrism.OpenAI.dll

The IModelProvider implementation for OpenAI.

public sealed class OpenAIModelProvider : IModelProvider, IModelProviderHealthCheck, IModelProviderConfigurationDiagnostics

Inheritance

object ← OpenAIModelProvider

Implements

IModelProvider, IModelProviderHealthCheck, IModelProviderConfigurationDiagnostics

Inherited Members

object.GetType(), object.ToString(), object.Equals(object?), object.Equals(object?, object?), object.ReferenceEquals(object?, object?), object.GetHashCode()

Remarks

UseOpenAI registers two instances of this type: OpenAIProviderNames.ChatCompletions and OpenAIProviderNames.Responses. Both share the same OpenAIChatClientFactory instance, and therefore the same HTTP connection pool; only their OpenAIModelProvider.ApiSurface values differ.

The provider does not reject a model name that is absent from the catalog. When OpenAI publishes a new model, no new AgentPrism release is needed; if the catalog is not empty, only an informational log entry is written. Because AgentPrism carries no built-in model list, an empty catalog is normal; in that case nothing is logged.

Constructors

OpenAIModelProvider(string, OpenAIApiSurface, OpenAIChatClientFactory, IReadOnlyList<ModelDescriptor>, ILogger<OpenAIModelProvider>?, OpenAIProviderOptions?, string?, EgressSocketGuard?)

Initializes a new provider.

public OpenAIModelProvider(string name, OpenAIApiSurface apiSurface, OpenAIChatClientFactory chatClientFactory, IReadOnlyList<ModelDescriptor> models, ILogger<OpenAIModelProvider>? logger = null, OpenAIProviderOptions? healthCheckOptions = null, string? configurationSectionKey = "AgentPrism:Providers:OpenAI", EgressSocketGuard? egressGuard = null)

Parameters

name string

The provider name. The ModelBinding.Provider of agent definitions matches this value.

apiSurface OpenAIApiSurface

The OpenAI API surface to use.

chatClientFactory OpenAIChatClientFactory

The chat client factory.

models IReadOnlyList<ModelDescriptor>

The models this provider offers.

logger ILogger<OpenAIModelProvider>?

The logger.

healthCheckOptions OpenAIProviderOptions?

When given, OpenAIModelProvider.CheckHealthAsync calls the GET {endpoint}/models endpoint with the address and key from these options. When null, the health status is always ModelProviderHealthStatus.Unknown.

configurationSectionKey string?

The fixed configuration section that OpenAIModelProvider.GetConfigurationDiagnostic reports. The default is OpenAIProviderOptions.SectionName — it is meant for UseOpenAI. UseOpenAICompatible passes null because it takes the key freely in code (there is no fixed section path); in that case no ConfigurationDiagnostic is reported.

egressGuard EgressSocketGuard?

The outbound network guard. Attached only to a client built from a tenant-supplied endpoint override; when null, such an override is not guarded.

Exceptions

ArgumentNullException

A required dependency is null.

ArgumentException

name is empty.

Properties

ApiSurface

Gets the OpenAI API surface this provider uses.

public OpenAIApiSurface ApiSurface { get; }

Property Value

OpenAIApiSurface

Models

The models this provider offers.

public IReadOnlyList<ModelDescriptor> Models { get; }

Property Value

IReadOnlyList<ModelDescriptor>

Name

The provider name. ModelBinding.Provider matches this value. Comparison is case-insensitive.

public string Name { get; }

Property Value

string

Methods

CheckHealthAsync(CancellationToken)

Checks the provider’s reachability.

public ValueTask<ModelProviderHealth> CheckHealthAsync(CancellationToken cancellationToken = default)

Parameters

cancellationToken CancellationToken

The cancellation token.

Returns

ValueTask<ModelProviderHealth>

The check result.

Remarks

The health check does not use OpenAIModelProvider._chatClientFactory — it makes a separate, lightweight HTTP GET. See AgentPrism.OpenAIProviderHealthCheck.

CreateChatClient(ModelBinding, ModelProviderCredential?)

Produces a raw chat client for the given binding.

public IChatClient CreateChatClient(ModelBinding binding, ModelProviderCredential? credential = null)

Parameters

binding ModelBinding

The model binding.

credential ModelProviderCredential?

A resolved per-tenant credential (BYOK). When null, the provider’s own setup-time credential is used and behavior is identical to a setup without per-tenant credentials. When given, the provider builds (or reuses a cached) client using ModelProviderCredential.ApiKey and, if present, ModelProviderCredential.Endpoint, instead of its setup-time credential.

Returns

IChatClient

The provider-specific client. Decorators specific to the provider (example: Anthropic’s settings decorator) may be added here.

Remarks

Do not build the common pipeline here. UseFunctionInvocation, UseOpenTelemetry, the content guard, the circuit breaker, and extra resolution are added by ModelProviderRegistry.CreateChatClient. Before that shared pipeline, every provider package built the tool-call loop inside itself; the result was that no ring the registry wraps around could see the loop’s turns — a tool result entered the model uninspected.

If the loop is also built here, two nested FunctionInvokingChatClient instances form: the inner one resolves tools, the outer one never sees any call. The damage is not functional but measurable (double wrapping, a misleading span tree).

GetConfigurationDiagnostic()

Returns the resolution status of the configuration key required by this provider.

public ConfigurationDiagnostic? GetConfigurationDiagnostic()

Returns

ConfigurationDiagnostic?

null when the provider was not configured with an options object.