MCP Servers for VS Code (Crash Course)
Supercharge GitHub Copilot in VS Code with the Model Context Protocol (MCP): install and configure MCP servers (including Firecrawl), wire up custom tools over STDIO, and build a Python MCP server with FastMCP to retrieve Confluence docs—while staying secure with MCP tools. --- 🔗 *Links* - 📦 Code from the video: https://github.com/alejandro-ao/vs-code-mcp-demo - 🚀 Complete AI Engineer Bootcamp: https://aibootcamp.dev - ❤️ Buy me a coffee... or a beer (thanks!): https://link.alejandro-ao.com/l83gNq - 💬 Join the Discord Help Server: https://link.alejandro-ao.com/HrFKZn - ✉️ Newsletter (AI Engineering updates): https://link.alejandro-ao.com/AIIguB --- 🤓 *Topics Covered* - Model Context Protocol (MCP) fundamentals: client–server model and tool invocation - Using MCP in VS Code with GitHub Copilot (Agent Mode) - Installing MCP servers from the VS Code Marketplace - Manual MCP configuration (STDIO transport, command/args/env, absolute paths) - Firecrawl MCP server setup (npx, API key) and testing the scrape tool - Workspace vs Global MCP server scopes in VS Code - Building a custom Python MCP server with fastmcp - Defining tools with @mcp.tool, type hints, and precise docstrings - Exposing Confluence-like documentation: list pages and fetch by ID - Running servers with uv or Python (main.py) over STDIO - Security best practices: only tool descriptions go to the LLM, risks of malicious/misleading servers, and trust guidelines --- ⏰ *Timestamps* 0:00:00 - Intro 0:01:28 - What is MCP (Model Context Protocol)? 0:04:58 - Installing MCP servers from the VS Code Marketplace 0:08:05 - Installing custom MCP servers 0:11:05 - Building a Python MCP server in Python 0:17:52 - Connecting Copilot to your MCP server 0:21:18 - MCP security
