DotnetAi.Sdk 1.0.1

DotnetAi.Sdk

Unified and extensible .NET SDK for interacting with LLMs like OpenAI, Claude (Anthropic), and DeepSeek – using a single interface.

Quick Start

using DotentAi.Sdk;
using DotentAi.Sdk.Models;

AiClient client = new("openai", "sk-your-api-key");

ChatRequest request = new()
{
    Model = "gpt-4.1",
    Prompt = "What is the capital of Romania?"
};

string response = await client.SendAsync(request);
Console.WriteLine(response);

ChatRequest Parameters

The ChatRequest object defines how a chat interaction is structured. It supports both simple prompts and multi-turn messages.

Temperature Parameter

The temperature parameter controls how random or creative the AI's response will be. Lower values make the output more focused and deterministic. Higher values increase variability and originality, but can sometimes reduce coherence.

Supported Models

• All Anthropic models: https://docs.anthropic.com/en/docs/about-claude/models/overview
• All Deepseek models: https://api-docs.deepseek.com/quick_start/pricing
• All OpenAi models: https://platform.openai.com/docs/models

Advanced Usage

With ChatMessage history

The ChatMessage object defines each message within a conversation.

Each message includes:

List<ChatMessage> messages =
[
    new() { Role = "user", Content = "What is 5 + 3?" },
    new() { Role = "assistant", Content = "5 + 3 = 8" },
    new() { Role = "user", Content = "Add 10" }
];

string response = await client.SendAsync(new ChatRequest
{
    Model = "gpt-4.1",
    Messages = messages,
    Temperature = 0.0f,
    MaxTokens = 20
});

Console.WriteLine(response);

Advanced Prompting: Putting Words in the AI's Mouth

You can guide some AI models to produce tightly constrained output, for example, forcing them to pick a single answer like A, B, or C, by using a crafted assistant message and limiting MaxTokens.

This technique is inspired by Anthropic's example and works well with some openai and claude models. May not work with deepseek in current implementation.

Example

List<ChatMessage> messages =
[
    new() { Role = "user", Content = "What is latin for Ant? (A) Apoidea, (B) Rhopalocera, (C) Formicidae" },
    new() { Role = "assistant", Content = "The answer is (" },
];

ChatRequest request = new()
{
    Model = "gpt-4",
    Messages = messages,
    MaxTokens = 1,
};

string response = await client.SendAsync(request);
Console.WriteLine(response); // Output: C

Using custom HttpClient

var http = new HttpClient();
var client = new AiClient("openai", "your-api-key", http);