Guides · Reviewed October 2026
AI agents and MCP in .NET
The .NET AI stack has moved quickly. This page puts the most useful articles we have featured into an order that makes sense to read them in: the building blocks first, then MCP servers, then agents, then putting agents in front of users and shipping them responsibly. Every link goes to the original article.
Start with the building blocks
The abstractions and learning material the rest of the stack builds on.
Build an MCP server
From a first server in C# to the official SDK releases and hosting on Azure Functions.
Build a Model Context Protocol (MCP) server in C#
MCP C# SDK v1.0 Released: Build AI Agents and Context-Aware Apps in .NET
Announcing v2.0 of the official MCP C# SDK
Build MCP Remote Servers with Azure Functions
How to build long-running MCP tools on Azure Functions
MCP as Easy as 1-2-3: Introducing the Fluent API for MCP Apps
Build agents with Microsoft Agent Framework
The framework itself, durable workflows, agent skills, and a complete example that combines Agent Framework, Foundry, MCP and Aspire.
Introducing Microsoft Agent Framework: The Open-Source Engine for Agentic AI Apps
Microsoft Agent Framework: Building Blocks for AI Part 3
Durable Workflows in the Microsoft Agent Framework
Agent Skills for .NET Is Now Released
Discover Agent Skills from MCP servers in .NET
Build a real-world example with Microsoft Agent Framework, Microsoft Foundry, MCP and Aspire
Put agents in front of users
Streaming agent output into user interfaces, from the AG-UI protocol to experimental Blazor components.
AG-UI Protocol now has a first-class .NET SDK
Microsoft and CopilotKit released an MIT-licensed .NET SDK for AG-UI, the open protocol for streaming agent events to user-facing apps. Any .NET service can now expose or consume agent endpoints that work with TypeScript and Python clients.
Build Agentic UI with the new Blazor AI components
The experimental Microsoft.AspNetCore.Components.AI package renders streamed agent output, tool calls, approval prompts and shared state as Blazor components, on top of any IChatClient. Try it with the .NET 11 RC1 SDK and the AgenticUI sample.
Test, evaluate and govern
Checking what agents and models actually do before you ship them.
Use AI on your own codebase
Getting better results from coding agents in .NET projects.
How to Make AI Write High-Quality Code in .NET
Rules written in prompts fade as an agent's context fills up, so the article moves them into the build with TreatWarningsAsErrors, analyzers, .editorconfig and Central Package Management.
Extend your coding agent with .NET Skills
Ten Months with Copilot Coding Agent in dotnet/runtime
Section introductions are dotnetdevs.io commentary. Each article belongs to its author; see the full link library for everything we have featured.