FastMCP Beginner MCP Servers

Overview

This repository is a beginner-friendly exploration of MCP (Model Context Protocol) servers built using FastMCP. The goal of this project is to understand how to create, run, and validate MCP servers that can later be connected to AI agents for practical tasks such as calculations, API access, and data retrieval.

This project focuses on fundamentals:

• Creating MCP-compliant servers

• Exposing tools and resources to AI agents

• Verifying MCP servers using the official MCP Inspector

• Understanding different transport methods (stdio, HTTP, SSE)

Each scenario is intentionally simple and educational, making this project ideal for a GitHub portfolio that demonstrates early MCP and AI-agent infrastructure knowledge.

Project Structure

This repository contains four MCP scenarios, each showcasing a different way MCP servers can be implemented and tested.

Additionally, supporting notes are included to explain:

• How to run each MCP server

• How to validate functionality using the MCP Inspector

• Common troubleshooting tips

Tools & Technologies Used

• Python

• FastMCP

• Model Context Protocol (MCP)

• MCP Inspector

• HTTP / SSE transports

• RSS (XML feeds)

MCP Inspector (Required for Testing)

All MCP servers in this project are validated using the official MCP Inspector:

🔗 https://modelcontextprotocol.io/docs/tools/inspector

General Command Format

Depending on the scenario, the command may be a Python script or a URL endpoint.

Scenario 1: Basic FastMCP Script (Stdio Transport)

Description

This scenario demonstrates a basic FastMCP server running directly from a Python script using standard input/output (stdio). It exposes simple tools (such as a calculator) that an AI agent can call.

How to Run

Key Concepts

• Stdio-based MCP servers

• Tool exposure via FastMCP

• Direct script execution

Validation

• Launch the MCP Inspector

• Ensure tools appear correctly

• Call exposed tools to verify functionality

Scenario 2: FastMCP Web API (HTTP + SSE Transport)

Description

This scenario runs a FastMCP server as a web API, allowing MCP connections over HTTP using Server-Sent Events (SSE).

Running the API

Start the server locally (example port shown):

If the browser shows "Not Found", navigate to:

This opens the interactive API documentation.

MCP Inspector Command

Important Inspector Settings

• Transport Type: SSE

• URL: Your selected MCP endpoint (e.g., http://localhost:8001/mcp)

Key Concepts

• HTTP-based MCP servers

• SSE transport

• API-style MCP services

Scenario 3: RSS Feed MCP Server

Description

This scenario uses RSS (Really Simple Syndication) feeds to provide structured XML-based updates to an AI agent.

Data Sources

• Website RSS feeds

• YouTube RSS feed for freeCodeCamp.org

  • Channel ID: UC8butISFwT-Wl7EV0hUK0BQ

Purpose

• Demonstrates MCP servers as information providers

• Shows how AI agents can consume external XML data

Key Concepts

• RSS and XML parsing

• Content syndication

• MCP as a data ingestion layer

Scenario 4: Project Aggregation & MCP Registration

Description

After completing all scenarios, project metadata is collected into a JSON configuration file. This file is used to register and manage MCP servers inside an AI development environment.

VS Code Agent Mode Setup

1. Open Agent Mode in VS Code

   • Click the talkbox icon near the top center

2. Select an AI model at the bottom of the agent panel

3. Navigate to:

   Extensions → MCP Servers → Installed

4. Right-click your MCP server

5. Select Start Service

This allows AI agents to discover and interact with your MCP servers.

Architecture Overview

What This Project Demonstrates

• Practical understanding of MCP fundamentals

• Ability to build MCP servers from scratch

• Knowledge of multiple transport types (stdio, HTTP, SSE)

• Experience validating MCP servers with official tools

• Early-stage AI agent infrastructure development

Why This Matters

MCP is becoming a core building block for agentic AI systems. This project shows:

• You understand how AI tools are exposed

• You can wire services into agent workflows

• You are building toward scalable, modular AI systems

For a beginner project, this is exactly the right direction.

Future Improvements

• Add authentication and security layers

• Expand tool complexity

• Integrate databases or vector stores

• Deploy MCP servers remotely

• Connect multiple MCP servers to a single agent

Status

✅ Beginner project completed

This repository represents a learning-first implementation of MCP servers and lays the foundation for more advanced AI-agent systems.