CoreTemplate.AI 1.1.0

🧠 CoreTemplate.AI

Modular & Configurable AI Service Layer for .NET (OpenRouter + Ollama)

CoreTemplate.AI is a pluggable, strategy-based AI integration library for .NET Core projects.
It lets you easily switch between cloud-based and local LLM providers — like OpenRouter and Ollama — at runtime via configuration.

Designed with Clean Architecture, Strategy Pattern, MediatR, and strong extensibility in mind.

✨ Features

• ✅ Plug-and-play AI integration
• ⚙️ Provider switch via appsettings.json (OpenRouter, Ollama)
• 🔄 Dynamic model override (per-request)
• ✅ Model support validation (checks if model exists via real API)
• 🧱 Fully testable & decoupled structure
• 💡 Built-in CQRS Command: PromptTextCommand (MediatR)

📦 Installation

dotnet add package CoreTemplate.AI

Or add manually in .csproj:

<PackageReference Include="CoreTemplate.AI" Version="1.1.0" />

🛠️ Program.cs Setup

// --- AI Services ---
builder.Services.AddOptions<AISettings>()
    .Bind(builder.Configuration.GetSection("AiSettings"))
    .Validate(settings => Enum.IsDefined(typeof(AIProvider), settings.Provider),
        "Invalid AI provider configured in AiSettings.Provider");

builder.Services.AddSingleton(sp =>
    sp.GetRequiredService<IOptions<AISettings>>().Value);

// Providers
builder.Services.AddScoped<OpenRouterAiService>();
builder.Services.AddScoped<OllamaAiService>();
builder.Services.AddScoped<IAIService, AIServiceResolver>();

// Model Providers
builder.Services.AddScoped<OpenRouterModelProvider>();
builder.Services.AddScoped<OllamaModelProvider>();
builder.Services.AddScoped<AIModelProviderResolver>();

// MediatR
builder.Services.AddMediatR(cfg =>
{
    cfg.RegisterServicesFromAssembly(typeof(PromptTextCommandHandler).Assembly);
});

⚙️ Configuration

appsettings.json

"AiSettings": {
  "Provider": "OpenRouter", // or "Ollama"
  "Model": "anthropic/claude-3-haiku" // any model that supported
},
"OpenAI": {
  "ApiKey": "sk-your-openrouter-api-key" // need for OpenRouter
}

🚀 Example Usage

🧩 1. Using MediatR (Preferred) Send prompt through built-in PromptTextCommand

var result = await _mediator.Send(new PromptTextCommand
{
    Prompt = "Write a poem about code.",
    Options = new AIRequestOptions
    {
        Context = "You are a helpful assistant",
        Temperature = 0.7,
        Model = "mistralai/mistral-7b-instruct" // overrides config default
    }
});

🧩 2. Using IAIService or with API Controller Directly

public class MyService
{
    private readonly IAIService _ai;

    public MyService(IAIService ai) => _ai = ai;

    public async Task<string> AskAsync(string prompt)
    {
        return await _ai.PromptAsync(prompt, new AIRequestOptions
        {
            Model = "llama2",
            Temperature = 0.8
        });
    }
}

or

[Route("api/ai")]
[ApiController]
public class AiController : ControllerBase
{
    private readonly IMediator _mediator;
    private readonly IAIService _aiService;

    public AiController(IMediator mediator, IAIService aiService)
    {
        _mediator = mediator;
        _aiService = aiService;
    }

    // POST: /api/ai/prompt
    [HttpPost("prompt")]
    public async Task<IActionResult> Prompt([FromBody] PromptTextCommand command)
    {
        var result = await _mediator.Send(command);

        return Ok(new { result });
    }

    // GET: /api/ai/model-supported?model=anthropic/claude-3-haiku
    [HttpGet("model-supported")]
    public async Task<IActionResult> IsModelSupported([FromQuery] string model)
    {
        var isSupported = await _aiService.IsModelSupportedAsync(model);

        return Ok(new { model, isSupported });
    }
}

📘 Supported Providers

🛡️ Model Validation

Before sending the prompt, the system checks if the model exists using:

https://openrouter.ai/api/v1/models for OpenRouter

http://localhost:11434/tags for Ollama

🧪 Example cURL Test

curl -X POST https://localhost:5001/api/ai/prompt
-H "Content-Type: application/json"
-d '{ "prompt": "Explain quantum computing", "options": { "model": "anthropic/claude-3-haiku" } }'

🧰 Technologies

• .NET 9
• Clean Architecture
• Strategy Pattern
• MediatR (CQRS support)
• FluentValidation
• JSON-based configuration
• OpenAI-compatible APIs

👤 Author

Developed by @CerenSusuz GitHub: [https://github.com/CerenSusuz/CoreTemplateApp/tree/master/Core.AI]

📄 License

This project is licensed under the MIT License.