RepletoryLib.Ai.OpenAi 1.0.0

RepletoryLib.Ai.OpenAi

OpenAI provider implementation for RepletoryLib AI services (GPT, DALL-E, Whisper, TTS, Moderation).

Part of the RepletoryLib ecosystem -- standalone, reusable .NET 10 libraries with zero business logic.

Overview

RepletoryLib.Ai.OpenAi implements all five AI service interfaces from RepletoryLib.Ai.Abstractions using direct HTTP calls to the OpenAI API. Each service uses a named HttpClient with Polly resilience policies (retry with Retry-After support, circuit breaker) and returns Result<T> for consistent error handling.

Key Features

• Chat Completions -- GPT-4o, GPT-4, GPT-3.5 with streaming, tool calling, and structured output
• Embeddings -- text-embedding-3-small/large with batch support and configurable dimensions
• Image Generation -- DALL-E 3/2 with size, quality, and count options
• Audio Transcription -- Whisper speech-to-text with language detection
• Text-to-Speech -- TTS-1/TTS-1-HD with multiple voices and audio formats
• Content Moderation -- omni-moderation-latest with category flag mapping
• Polly Resilience -- Retry with Retry-After header support, circuit breaker per service
• Result<T> Wrapping -- All responses wrapped in Result<T> with HTTP status code mapping

Installation

dotnet add package RepletoryLib.Ai.OpenAi

Dependencies

Quick Start

using RepletoryLib.Ai.OpenAi;

var builder = WebApplication.CreateBuilder(args);
builder.Services.AddRepletoryOpenAi(builder.Configuration);

{
  "Ai": {
    "OpenAi": {
      "ApiKey": "sk-...",
      "DefaultChatModel": "gpt-4o",
      "DefaultEmbeddingModel": "text-embedding-3-small",
      "TimeoutSeconds": 120,
      "RetryCount": 3
    }
  }
}

Configuration

OpenAiOptions

Usage Examples

Chat Completion

using RepletoryLib.Ai.Abstractions.Interfaces;
using RepletoryLib.Ai.Abstractions.Models.Chat;

public class ChatService
{
    private readonly IChatCompletionService _chat;

    public ChatService(IChatCompletionService chat) => _chat = chat;

    public async Task<string?> AskAsync(string question)
    {
        var request = new ChatRequest
        {
            Messages = [ChatMessage.User(question)]
        };

        var result = await _chat.CompleteAsync(request);
        return result.IsSuccess ? result.Data?.Content : null;
    }
}

Streaming

await foreach (var chunk in _chat.StreamAsync(request))
{
    if (chunk.ContentDelta is not null)
        Console.Write(chunk.ContentDelta);

    if (chunk.IsFinal)
        Console.WriteLine($"\nTokens: {chunk.FinalUsage?.TotalTokens}");
}

Embeddings

var result = await _embedding.EmbedSingleAsync("Hello, world!");
if (result.IsSuccess)
    Console.WriteLine($"Dimensions: {result.Data!.Length}");

Image Generation

var result = await _image.GenerateAsync(new ImageGenerationRequest
{
    Prompt = "A cat wearing a top hat",
    Size = ImageSize.Square1024,
    Quality = ImageQuality.Hd
});

Registered Services

All services are registered with TryAddScoped, allowing them to be overridden by registering your own implementation first.

Integration with Other RepletoryLib Packages

Troubleshooting

License

This project is licensed under the MIT License.

Copyright (c) 2024-2026 Repletory.

For complete documentation, infrastructure setup, and configuration reference, see the RepletoryLib main repository.