MCP Model Context Protocol Server Development in TypeScript

Build MCP servers callable from AI assistants and editors. 50 auto-graded labs: remote MCP, OAuth, npm. No API keys
What you'll learn

---

Understand what happened to AutoGen and what MAF inherited from it, and explain the three sticking points when migrating AutoGen code to MAF.
Understand ChatClient / Agent / Session, shape a personality with instructions, keep conversations going with Session, and separate history per user.
Build custom tools with `@tool`, understand how type hints and docstrings double as the AI's instruction manual, and build an agent that chooses between tools.
Explain the five coordination patterns - Sequential, Concurrent, Handoff, GroupChat, and Magentic - with diagrams, and decide which one fits your own problem.
Cut processing time with parallel execution, and implement a Handoff that routes requests to the right specialist - adding a new desk costs almost nothing.
6. Build a directed graph with WorkflowBuilder, design branching with switch-case and fan-out / fan-in, and catch bugs before execution via type checking.
Pause a Workflow for human approval with `request_info`, and resume a crashed run with `FileCheckpointStorage` - features AutoGen never had.
Requirements

---

Basic Python syntax - variables, functions, classes, enough to read imports and decorators. No async experience needed; Section 0 covers MAF's conventions.
A PC with internet and a modern browser (Chrome recommended) is all you need. The course uses CodeSandbox, so no Python install or setup is required.
Ability to get one Gemini API key from Google AI Studio (free tier, no credit card needed - Section 0 walks through the steps).
No Azure subscription required. Although this is a Microsoft framework, the course is designed so the free Gemini tier alone gets you through.
Description

---

This course contains the use of artificial intelligence.
"I get MCP as a user. But what about building one?" This course answers that question in code.
MCP (Model Context Protocol) is a shared standard connecting AI assistants to outside tools and data. Most material available today leans toward the using side: plugging an existing MCP server into an AI tool. But the real value appears when you move to the other side and expose your own business data and internal systems in a form an AI can call. This course covers only that building side, as 50 exercises in TypeScript.
This is not a course you read, or a course you watch. All 50 exercises run to the same rhythm: task, acceptance criteria, implement it yourself, grade it with npm run check, then check your work in the review video. There are two videos per exercise. The problem video hands you the knowledge and the expected result, you pause there and write it yourself, and the review video covers the model answer, common errors and the design decisions. Nothing about this course lets you watch it through with your hands still.
The curriculum is six chapters. Chapter 1 covers Tools in 10 exercises: registerTool, Zod schemas, structured output, error design, progress notifications and logging. Chapter 2 covers Resources and Prompts in 8 exercises: URI templates, binary data, subscription notifications, prompts with arguments, and resource_link. Unlike Tools-heavy material, all three primitives are practised evenly, by hand.
From Chapter 3 the servers get practical. In the file exercises, the test actually attacks your server with path traversal. Read-only SQLite queries, REST API wrappers, rate limiting, CSV analysis, pagination. It also gets into a design question specific to MCP: how to return results without eating up the context window. Chapter 4 turns to the client side. In exercise 29 you write the test harness that has been grading you all along, which pulls back the curtain on the grading system. You then implement sampling, elicitation, roots and completion, and finish by assembling a mini agent that runs a tool use loop, so you feel in your hands that MCP is a component inside an agent.
Chapter 5 is going remote and authentication. You move from stdio to Streamable HTTP, design it stateless (and prove with a test that it still works when requests are round-robined across two independent servers), validate Origin, implement Protected Resource Metadata (RFC 9728), and hand-write JWT verification with no external library so that tools are exposed by audience and scope. Chapter 6 is quality and extension. Exercise 44 is the one with the structure reversed: the server is finished, and you write the tests. Prompt injection defence, tracing with OpenTelemetry, the Tasks extension, the groundwork for MCP Apps, and publishing to npm.
The last exercise, 50, is a capstone. You turn a problem from your own work into an MCP server, write ten realistic questions an LLM ought to be able to answer against it, and verify with a real client. Your tests turning green and an LLM genuinely using your server well are two different problems, and experiencing that gap is how the course closes.
No environment setup is required. Open the repository in CodeSandbox or GitHub Codespaces and a Node 24 environment comes up with dependencies already installed. You can finish all 50 exercises in a browser. And you need no API keys at all. The external API exercises stub fetch and the GitHub exercise lets you swap the endpoint, so all 50 gradings pass offline. Neither the network nor a credit card gets in the way of learning.
Who this course is for

---

Anyone who set out to learn AutoGen and was thrown to find it's in maintenance mode, wanting to know what to relearn and if prior knowledge still counts.
Developers actively writing AutoGen or Semantic Kernel code who are considering a move to Microsoft Agent Framework.
Engineers who can build a single AI agent but aren't sure how to design coordination between multiple agents or fold them into a business workflow.
Anyone who wants an AI agent in an internal system but is stuck on whether it's safe to hand it everything, and wants to design approval and audit trails.
Anyone who has learned LangChain or CrewAI and wants to see how Microsoft's official stack solves the same problems, to broaden their framework options.
Independent developers and students who want to try multi-agent development on the free tier alone, without paying for Azure.
Homepage

---