AIHelperLibrary 1.1.0

AIHelperLibrary

AIHelperLibrary is a modular, multi-provider .NET library enabling seamless integration with OpenAI (GPT models) and Anthropic (Claude models). Designed for extensibility, robust configuration, multi-turn conversation support, and advanced prompt management.

Installation

Install via .NET CLI:

dotnet add package AIHelperLibrary --version 1.1.0

Or manually in your .csproj file:

<PackageReference Include="AIHelperLibrary" Version="1.1.0" />

Quick Start

using AIHelperLibrary.Configurations;
using AIHelperLibrary.Models;
using AIHelperLibrary.Services;

var config = new OpenAIConfiguration
{
    DefaultModel = OpenAIModel.GPT_4o,
    MaxTokens = 500,
    Temperature = 0.7,
    EnableLogging = true
};

var client = new OpenAIClient("your-api-key", config);
var result = await client.GenerateTextAsync("Explain quantum computing.");
Console.WriteLine(result);

Full Configuration Reference

OpenAIConfiguration

AnthropicConfiguration

Usage Examples

Generate a Basic Response

var response = await client.GenerateTextAsync("What is the capital of Japan?");
Console.WriteLine(response);

Persistent Chatbot Session

var chatResponse = await client.GenerateChatResponseAsync(
    instanceKey: "Session1",
    userMessage: "Can you help me book a hotel?",
    initialPrompt: "You are a polite and helpful assistant."
);
Console.WriteLine(chatResponse);

Dynamic Prompt Manager

var promptManager = new DynamicPromptManager();
promptManager.AddPrompt("FriendlyGreet", "Please greet the user warmly.");

var dynamicResponse = await client.GenerateTextWithDynamicPromptAsync(promptManager, "FriendlyGreet", "Hello AI!");
Console.WriteLine(dynamicResponse);

Supported Models

OpenAI

Anthropic (Claude)

Advanced Topics

Retry Logic

If a request fails due to a network or transient server issue, the library will automatically retry according to MaxRetryCount and RetryDelayMs settings.

Proxy Support

You can route API calls through a proxy server using ProxyUrl and ProxyPort configuration.

Custom Headers

Any additional HTTP headers (e.g., metadata) can be attached via the CustomHeaders dictionary.

Special Handling for o-Series Models

Models like o1, o3-mini, and o4-mini ignore temperature and top_p parameters and instead use max_completion_tokens.

Contributing

Pull requests, bug reports, and feature suggestions are welcome. Please fork the repository and submit a pull request.

License

Distributed under the MIT License.

Changelog

See CHANGELOG.md for version history and release notes.