---
schemaVersion: 1
title: "Model Context Protocol (MCP)"
canonicalUrl: "https://techtown.fr/en/formations/mcp/"
language: "en"
---

## Objectifs pédagogiques

- Understand the foundations and protocol of MCP
- Discover the three key components: Tools, Resources and Prompts
- Understand the client-server architecture and build MCP servers

## Programme

### MCP Fundamentals and Architecture

- Presentation of the day's objectives and the MCP protocol
- Why MCP is a game changer in AI architecture
- Discover the 3 key server-side components: Tools (Functionality), Resources (Data), Prompts (Instructions)
- Understanding the Client-Server architecture redesigned for AI (LLM Client vs. MCP Server)

### MCP Servers in Action

- Review of available SDKs (Python/TypeScript)
- Common server types (Filesystem, GitHub, PostgreSQL)
- Explanation of the MCP request lifecycle: from LLM request to Tool execution

### Hands-on Workshops

#### Workshop 1: Building a Simple Python Server

Create and start a simple MCP server. Set up testing with FastMCP.

#### Workshop 2: Client Interaction - Java

Create and start a simple MCP server exposing a utility function.

#### Workshop 3: MCP & Agents

Build an AI agent and configure an MCP server with Cloud Run.

## Pré-requis

- Have software development experience
- Know Python, TypeScript or Java
- Be comfortable with a terminal