AgentCircuits.Providers.OpenAI 0.7.134

AgentCircuits.Providers.OpenAI

OpenAI Responses API provider for AgentCircuits.

👁 NuGet
👁 License: MIT

Installation

dotnet add package AgentCircuits.Providers.OpenAI

Quick Start

using AgentCircuits.Providers;
using AgentCircuits.Providers.OpenAI;

var model = OpenAI.LanguageModel("gpt-5-mini");

var response = await model.GenerateAsync(new LlmRequest
{
 Messages = [Message.User("Summarize the latest architecture trends in one paragraph.")],
 MaxTokens = 300
});

OpenAI.LanguageModel(...) uses OPENAI_API_KEY from the environment.

Custom Provider Options

using AgentCircuits.Providers.OpenAI;

var provider = OpenAI.Create(new OpenAIProviderOptions
{
 ApiKey = "your-api-key",
 BaseUrl = "https://api.openai.com/v1",
 MaxRetries = 2
});

var model = provider.CreateLanguageModel("gpt-5-mini");

Response Lifecycle Helpers

For background responses and retrieval flows, use the lifecycle client:

using AgentCircuits.Providers.OpenAI;
using OpenAI.Responses;

var lifecycle = OpenAI.ResponseLifecycle("gpt-5-mini");

var response = await lifecycle.GetAsync("resp_123");
var cancelled = await lifecycle.CancelAsync("resp_123");
var deleted = await lifecycle.DeleteAsync("resp_123");

await foreach (ResponseItem input in lifecycle.GetInputItemsAsync("resp_123"))
{
 // Inspect stored input items
}

await foreach (StreamingResponseUpdate update in lifecycle.GetStreamingUpdatesAsync("resp_123"))
{
 // Resume streaming from an existing response
}

You can also create the lifecycle client from a configured provider:

var lifecycle = provider.CreateResponseLifecycleClient("gpt-5-mini");

Notes

• Supports OpenAI model families used by this provider (gpt-*, chatgpt-*, codex*, and o* reasoning families).
• Uses the OpenAI .NET SDK Responses API surface.
• Streaming, tool calls, citations, and reasoning metadata are mapped into AgentCircuits LlmResponse/LlmStreamChunk formats.

See the main AgentCircuits repository for end-to-end integration examples.