MCP Dev Environment Setup for Python

Contenido del curso

Introducción a MCP

Conceptos basicos de MCP

MCP avanzado

Integrando tu MCP con un agente

Tomar examen

Cursos Empresas Blog Live Conf Eventos Precios

0:007:59

MCP Dev Environment Setup for Python

Curso de MCP con Microsoft Azure

Clase 3 de 22

Clase 3 / 22

Recursos

Setting up your MCP development environment is the first practical step before building anything with the Model Context Protocol. You will learn which tools to install, why each one matters, and how to verify your setup works on Windows, macOS or Linux, so you can focus on coding instead of fixing configuration issues.

What do you need before starting an MCP course?

MCP sits inside the AI Engineer specialization, so a bit of groundwork goes a long way. Before touching the protocol itself, make sure you already feel comfortable with two areas: the fundamentals of LLMs and the basics of artificial intelligence. Platzi has dedicated courses for both, and going through them first will make every concept here click much faster [01:20].

The second piece is programming fluency. You can absorb every AI concept in the world, but if your chosen language still feels foreign, you will get stuck on syntax instead of ideas. Pick your language, get comfortable, and then come back.

What prerequisites do I need for an MCP course? Solid programming skills in your chosen language, plus the LLM fundamentals and AI fundamentals courses. Without these, the learning curve gets steep fast.

Which tools should you install for MCP with Python?

The stack depends on the language you choose. With .NET, the build system handles the small server for you. With Python, which is the path used here, you need a few specific pieces working together [00:30].

Here is the checklist to run in your terminal:

python3 --version to confirm Python is installed. The reference version used is 3.13.7 [03:10].
pip3 --version to verify pip, the package manager. Version 25 works perfectly [03:35].
curl -LsSf with the official UV install script to set up UV, the virtual environment manager [04:00].
pip3 install uvicorn to add Uvicorn, the lightweight web server commonly paired with FastAPI [04:30].
npx --version to confirm npx, the npm package runner [05:00].

Each command in this list will be available in the class resources, so you can copy and paste instead of retyping.

Why do you need npx if you are coding in Python?

This is the part that surprises most people. MCP ships with an inspector, a graphical tool that lets you see your server in action. That inspector runs as an npm package executed through npx, which is why Node ends up on a Python developer's machine [05:15].

You will not write JavaScript during the course, but you will lean on npx every time you want to debug your MCP server visually.

What is UV in Python? UV is a fast virtual environment and package manager for Python. It replaces the typical venv plus pip workflow with a single, faster tool, and it is the recommended way to isolate dependencies in this course.

How do you handle Python virtual environments on Linux or WSL?

If you work on Linux or inside WSL (Windows Subsystem for Linux) with Ubuntu, you will hit a wall the first time you try to install a package globally. Python now blocks global pip installs by default, and that is intentional, for security reasons [06:30].

The fix is straightforward: create a virtual environment for your project and install everything inside it. UV makes this almost invisible, and one of the upcoming classes walks through the exact commands.

A quick note on operating systems. The rest of the course runs on Windows with WSL Ubuntu, but the workflow is nearly identical on macOS or native Linux. Small variations exist in install paths, nothing that breaks the experience.

Do you need an Azure subscription for this MCP course?

An Azure subscription is not mandatory, but it is highly recommended. Some classes deploy the MCP server to the cloud so it can be installed on any instance you want, and that step is much smoother with Azure ready to go [07:30].

The good news on cost:

A free Azure subscription gives you around $200 in credits.
The full course is expected to consume only three or four dollars if resources are managed carefully.
Forgetting to shut down a resource is the main risk, so keep an eye on your dashboard.

With those credits, you have more than enough room to follow every cloud related class without worrying about your wallet.

Got your environment ready? Drop a comment with the operating system and Python version you are using, and let me know if any of these tools gave you trouble during installation.

Comentarios (10)10 comentarios

Pedro Jesus Hincapie Garcia

Hace un año

Les comparto, los enlaces hacia los cursos en mención:

https://platzi.com/cursos/llms/

https://platzi.com/cursos/ai-ml/

Javier Medel

Gracias por el recurso.

Zaidibeth Ramos

Les dejo los enlaces de los cursos relacionados

https://platzi.com/cursos/machine-learning/

https://platzi.com/cursos/ai-datos/

https://platzi.com/cursos/introduccion-ai/

https://platzi.com/cursos/aprender-data-ia/

Nery Alberto Cano Ortigoza

Hace 6 meses

¿Qué pasa si instalo paquetes globalmente?

Si intentas instalar paquetes globalmente usando pip install en sistemas operativos modernos (como las versiones recientes de Ubuntu o macOS), el sistema operativo bloqueará la acción por seguridad. Esto ocurre porque Python ahora protege sus librerías internas; modificar el entorno global puede romper herramientas críticas del propio sistema operativo. Al forzar una instalación global, corres el riesgo de corromper tu instalación de Python. La solución definitiva y la mejor práctica en la industria es crear siempre un entorno virtual (usando herramientas como uv o venv). Esto crea una "caja de arena" aislada donde puedes instalar cualquier versión de cualquier librería sin afectar al resto de tu computadora.

Yudy Lorena Olarte Cortes

Hace 23 días

Excelente, muchas gracias

Yonatan Alesander Ordoñez Diego

Hace un año•editado

$ curl -LsSf | sh

Para activarlo de forma permanete (CHTGPT)

$ echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc

$ source ~/.bashrc

Sí, puedes usar Node.js para el servidor. Aunque el profesor usará Python, mencionó que el inspector de MCP usa un paquete de Node (NPX), que será necesario instalar sin importar el lenguaje que elijas.

Carlos Torres

Hace 4 meses•editado

La instalación de UV en windows me funciono asi, desde la PowerShell

powershell -ExecutionPolicy ByPass -c "irm | iex"

¿Cuándo debo usar Azure en MCP?

Debes integrar Azure cuando estés listo para sacar tu servidor MCP de tu entorno local (localhost) y exponerlo para que otros agentes de IA o aplicaciones externas puedan consumirlo de forma remota. Durante la fase de desarrollo y pruebas, tu propia computadora es suficiente. Sin embargo, el verdadero poder del protocolo MCP brilla cuando centralizas tus herramientas en la nube. Utilizar Azure te permitirá alojar el servidor de manera escalable, gestionar la seguridad de los endpoints y conectar tus herramientas con modelos de lenguaje alojados en la nube sin preocuparte por mantener tu computadora encendida 24/7.

¿Por qué necesito NPX con Python?

Aunque el núcleo de la lógica y el servidor MCP estén construidos enteramente en Python, el ecosistema de desarrollo moderno es híbrido. Necesitas npx porque la herramienta oficial para inspeccionar visualmente los servidores MCP (el Inspector MCP) está desarrollada en el ecosistema de Node.js. Piensa en el inspector como un panel de control gráfico; en lugar de leer logs crudos en la terminal para ver cómo se comunican tus agentes de IA con el servidor, npx te permite ejecutar este panel directamente desde la web sin tener que instalar dependencias globales pesadas. Es el puente perfecto para depurar tus herramientas de forma visual y cómoda.

Omar Cuarenta

Hace 9 meses

¿deberia preocuparme si la pc que estoy usando solo la uso yo y ya tengo varias de estas cosas instaladas? (No soy desarrollador)

Mateo Aquino

Hace 8 meses

No, algunos sistemas operativos ya lo traen preinstalados

Ivan Darío Lobatón Vivas

Hola Soy solo yo o este link esta caido error. 404? https://platzi.com/cursos/ai-ml/

Jose David Angarita Pertuz

si, tambien me sale caido

seguramente van a actualizar el curso pronto

Introducción a MCP

Introducción a MCP

MCP Core Architecture: Server, Client, and Host

MCP Dev Environment Setup for Python

MCP Server Setup With Python and Inspector

Conceptos basicos de MCP

Four Core Blocks of Every MCP Server

Building an MCP Client That Talks to Your Server

Adding an LLM to Your MCP Client

STDIO vs SSE in MCP Servers

LLM Connected to a Local MCP Server

Testing MCP Servers Without a Browser

Deploying an MCP Server to Azure Container Apps

Using MCP Servers in VS Code Agent Mode

MCP avanzado

Query Azure Resources Directly From VS Code

Herramientas avanzadas de MCP para optimizar servidores y seguridad

GPT-4 Reading Local Files via MCP Server

Image Brightness Analysis with MCP and NumPy

How MCP Agents Remember Conversations

Enrutamiento de herramientas con MCP Server

Integrando tu MCP con un agente

Building a Web Search MCP Server with SERP API

Adding News, Products and Q&A Tools to MCP

Why MCP Docs Already Feel Outdated

MCP as the New Standard for AI Systems

MCP Dev Environment Setup for Python

Progreso del curso

Introducción a MCP

Conceptos basicos de MCP

MCP avanzado

Integrando tu MCP con un agente

Recursos

Recursos