Advanced
Build MCP Servers: Extend Claude with Custom Tools
Go beyond using AI tools — build them. Learn the Model Context Protocol from the ground up: define custom tools, expose resources, handle authentication, and publish a server that connects Claude to any API or data source you choose.
01
What Is MCP and Why It Exists
MCP (Model Context Protocol) is an open standard that lets AI assistants connect directly to your tools, databases, and APIs — eliminating the copy-paste workflow that slows every developer down. This lesson explains the problem it solves and why it matters.
›
02
Architecture: Hosts, Clients, and Servers
MCP defines three distinct roles — host, client, and server — and a precise lifecycle for how they interact. Understanding this architecture is essential before writing a single line of server code.
›
03
The Three Primitives: Tools, Resources, and Prompts
MCP servers expose three types of capabilities: tools, resources, and prompts. Each serves a different purpose and is controlled by a different actor. Knowing when to use each one is the foundation of good MCP server design.
›
04
Transport Types: stdio and Streamable HTTP
MCP servers communicate with clients over one of two transports: stdio for local processes, or Streamable HTTP for remote servers. Choosing the right transport — and understanding the critical stdio output rule — is essential before writing your first server.
›
Prefer to learn offline?
Download the complete workbook — yours to keep, print, or use in teaching.
Download Workbook →
Online lessons always free · PDF includes personal & educational use license