MCP Training: Learn the Model Context Protocol
A free, structured curriculum that takes you from "what is MCP?" to shipping production MCP servers. Hands-on tutorials, real code, no filler — and no payment wall. Start with module one or jump to what you need.
How this training works
The curriculum is organized into five modules. Each module is a set of written, code-first guides you can work through at your own pace. Everything runs on your machine with free tooling — you'll need Node.js (or Python) and an MCP client like Claude Desktop, Claude Code, or VS Code.
New guides are published regularly. If you want them in order as they come out, grab the free email curriculum at the bottom of this page.
The curriculum
What Is MCP? Plain-English Explanation
The protocol from the ground up: clients, servers, transports, and why a standard for AI tooling matters.
Start here →MCP vs Function Calling
How MCP relates to the tool/function calling you already know — and when each approach wins.
Read guide →Build Your First MCP Server in TypeScript
Step-by-step: scaffold a server, expose a tool, and connect it to Claude. ~30 minutes, working code.
Start tutorial →MCP Security Checklist
12 checks before you ship: input validation, OAuth 2.1, rate limiting, prompt injection defenses, and more.
Run the checklist →Full module outline
| Module | What you'll learn | Status |
|---|---|---|
| 01 · MCP Fundamentals | Protocol architecture, transports (stdio, Streamable HTTP), message lifecycle, capability negotiation | ✅ Live |
| 02 · Building MCP Servers | Tools, resources, and prompts with the TypeScript and Python SDKs; testing with the MCP Inspector | ✅ First tutorial live |
| 03 · Agent Integration | Connecting servers to Claude, ChatGPT, and custom agents; sessions; multi-tool orchestration | 📝 In progress |
| 04 · Skills & Slash Commands | Building and distributing reusable Skills for Claude Code; packaging expertise into shareable modules | 📝 In progress |
| 05 · Security & Production | Input validation, OAuth 2.1 flows, rate limiting, deployment patterns for production MCP servers | ✅ Checklist live |
Who this training is for
- Backend developers who want to expose existing APIs and services to AI assistants.
- AI engineers giving agents real-world capabilities through standardized tooling.
- DevTool builders making their products available to every MCP-compatible client.
- Platform teams rolling out MCP across an organization with security and governance in mind.
You should be comfortable with basic TypeScript or Python. No prior MCP or AI-agent experience needed.
Get the MCP curriculum by email
New modules and tutorials as they're published, in order — from fundamentals to production. Free.
First lesson is on its way. Meanwhile, start with module one.
No spam. Unsubscribe anytime.