microsoft-extensions-ai

name	microsoft-extensions-ai
description	Use when building provider-agnostic AI applications with Microsoft.Extensions.AI. Covers IChatClient, IEmbeddingGenerator, middleware pipelines, caching, telemetry, and DI integration for chat completions and embeddings. USE FOR: provider-agnostic AI abstractions, dependency-injected chat clients, embedding generation, middleware pipelines (caching, logging, rate limiting), switching between OpenAI/Azure/Ollama without code changes DO NOT USE FOR: Azure-specific model catalog features (use azure-ai-inference), building agent workflows with tools and orchestration (use agent-framework), MCP tool serving (use mcp), ML model training (use mlnet)
license	MIT
metadata	{"displayName":"Microsoft.Extensions.AI","author":"Tyler-R-Kendrick","version":"1.0.0"}
compatibility	claude, copilot, cursor
references	[{"title":".NET AI Documentation","url":"https://learn.microsoft.com/dotnet/ai"},{"title":"Microsoft.Extensions.AI NuGet Package","url":"https://www.nuget.org/packages/Microsoft.Extensions.AI"},{"title":".NET Extensions GitHub Repository","url":"https://github.com/dotnet/extensions"}]

Microsoft.Extensions.AI

Overview

Microsoft.Extensions.AI provides a set of provider-agnostic abstractions for AI services in .NET, centered on IChatClient for chat completions and IEmbeddingGenerator<,> for vector embeddings. It follows the same patterns as ILogger and HttpClient in .NET -- services program against interfaces, providers are configured at the composition root, and middleware can be composed in pipelines for caching, telemetry, rate limiting, and function calling.

NuGet Packages

dotnet add package Microsoft.Extensions.AI
dotnet add package Microsoft.Extensions.AI.OpenAI        # OpenAI / Azure OpenAI provider
dotnet add package Microsoft.Extensions.AI.Ollama        # Ollama local models
dotnet add package Microsoft.Extensions.AI.AzureAIInference  # Azure AI model catalog

Basic Chat Completion

using Microsoft.Extensions.AI;

IChatClient chatClient = new OllamaChatClient(new Uri("http://localhost:11434"), "llama3.1");

var response = await chatClient.GetResponseAsync("What is dependency injection?");
Console.WriteLine(response.Messages.Last().Text);

Streaming Chat Completion

using Microsoft.Extensions.AI;

IChatClient chatClient = new OpenAIChatClient("gpt-4o", apiKey);

await foreach (var update in chatClient.GetStreamingResponseAsync("Explain SOLID principles"))
{
    Console.Write(update.Text);
}
Console.WriteLine();

Multi-Turn Conversations

var messages = new List<ChatMessage>
{
    new(ChatRole.System, "You are a C# tutor. Give concise explanations with code examples."),
    new(ChatRole.User, "What are records in C#?")
};

 response =  chatClient.GetResponseAsync(messages);
Console.WriteLine(response.Messages.Last().Text);


messages.AddRange(response.Messages);
messages.Add( ChatMessage(ChatRole.User, ));

 followUp =  chatClient.GetResponseAsync(messages);
Console.WriteLine(followUp.Messages.Last().Text);