Skip to main content
Glama
GitHub Copilot Custom MCP Server

GitHub Copilot Custom MCP Server

by owainow

GitHub Copilot Custom MCP Server with Azure Functions Workshop

Open in GitHub Codespaces Deploy to Azure

πŸš€ Overview

Welcome to this comprehensive workshop where you'll learn to extend GitHub Copilot's capabilities by building and deploying a custom Model Context Protocol (MCP) server on Azure Functions. This workshop demonstrates the complete journey from local development to production AI integration.

What You'll Build

  • Custom MCP Server: A serverless MCP server with three types of tools

  • Educational Tools: Markdown review and dependency checking with local algorithms

  • Production AI Tool: Azure AI-powered code review demonstrating true MCP architecture

  • Azure Functions Deployment: Scalable, serverless hosting for your MCP server

  • GitHub Copilot Integration: Seamless connection between Copilot and your custom tools

Learning Outcomes

By the end of this workshop, you'll understand:

  • βœ… The difference between educational and production MCP tools

  • βœ… How to build and deploy serverless MCP servers on Azure Functions

  • βœ… True MCP architecture: tools provide context, AI provides intelligence

  • βœ… GitHub Copilot integration patterns and best practices

  • βœ… Azure AI Foundry integration with graceful fallback patterns

Related MCP server: Remote MCP with Azure Functions (Python)

⚑ How to start! Quick Start Options:

🌟 Option 1: GitHub Codespaces (Recommended - Zero Setup)

Click the "Open in GitHub Codespaces" badge above for instant setup!

  • βœ… No local installation required

  • βœ… Pre-configured Linux environment with all tools

  • βœ… Works on any device with a browser

  • βœ… Ready in 2-3 minutes

  • πŸ“– Follow the Linux/Bash Documentation Path

πŸ’» Option 2: Local Development

Choose your platform for local development:

🐧 Linux/macOS (Bash)

  • πŸ“– Linux Workshop Documentation - Complete Linux setup with Bash commands

  • πŸ› οΈ Requirements: Node.js, Azure CLI, Azure Functions Core Tools, VS Code

  • πŸ’‘ Best for: Linux/macOS developers, Bash users, script automation


πŸ“š Choose Your Workshop Path

Platform

Quick Start

Documentation

Best For

🌟 Codespaces

Open in GitHub Codespaces

πŸ“– Linux/Bash Docs

Zero setup, any device

🐧 Linux/macOS

Setup Guide

πŸ“– Linux/Bash Docs

Linux/macOS developers

πŸ› οΈ Prerequisites

Required Software

Azure Account

  • Azure subscription (free tier works!)

  • Contributor access to create resources

Knowledge Level

  • Intermediate JavaScript/TypeScript

  • Basic Azure Functions knowledge

  • Familiarity with GitHub Copilot

πŸ“‹ Workshop Flow (3 Hours Total)

Choose your platform path above, then follow these sequential steps:

[Part 1: Setup and Understanding] (30 minutes)

  1. Understanding MCP and Architecture Patterns

  2. Environment Setup and Prerequisites

  3. Project Structure and Dependencies

[Part 2: Local Development] (45 minutes)

  1. Building the MCP Server Core

  2. Creating Educational Tools (markdown review, dependency check)

  3. Implementing Production AI Tool Architecture

  4. Testing Locally with Azure Functions

[Part 3: Azure Deployment] (30 minutes)

  1. Infrastructure as Code with Bicep

  2. Deploying to Azure Functions

  3. Monitoring and Troubleshooting

[Part 4: GitHub Copilot Integration] (30 minutes)

  1. Configuring MCP in VS Code

  2. Testing Tool Discovery and Usage

  3. Advanced Integration Patterns

[Part 5: AI Integration](45 minutes) πŸ€–

  1. Setting Up Azure AI Foundry

  2. Implementing Real AI Analysis

  3. Comparing Educational vs Production Tools

  4. Understanding True MCP Architecture

πŸ—οΈ Architecture Overview

graph LR
    A[GitHub Copilot] -->|MCP Protocol| B[Azure Functions]
    B --> C[Markdown Tool - Educational]
    B --> D[Dependency Tool - Educational]
    B --> E[AI Code Review - Production]
    
    E -->|API Calls| F[Azure AI Foundry]
    B --> G[Azure Monitor]
    B --> H[Azure Key Vault]
    B --> I[Azure Storage]

Tool Categories

This workshop demonstrates three distinct tool patterns:

πŸŽ“ Educational Tools

  • markdown_review: Local analysis algorithms, quality scoring

  • dependency_check: Static package analysis, security checks

  • Purpose: Learn MCP concepts, no external dependencies

πŸš€ Production Tools

  • ai_code_review: Azure AI integration with intelligent analysis

  • Purpose: Demonstrate true MCP architecture (tools provide context, AI provides intelligence)

  • Features: Real LLM analysis, graceful fallback to mock analysis

πŸ”„ Hybrid Approach

All tools work without Azure costs (minor GPT3.5 cost($.10) if desired) through intelligent fallback patterns, making the workshop accessible to everyone while demonstrating production capabilities.

🎯 Key Learning Outcomes

Technical Skills

  • βœ… MCP protocol implementation with JSON-RPC 2.0

  • βœ… Azure Functions serverless development

  • βœ… TypeScript development with Azure tooling

  • βœ… Infrastructure as Code with Bicep

  • βœ… Azure AI service integration

Architectural Understanding

  • βœ… Educational vs Production MCP tool patterns

  • βœ… Graceful degradation and fallback strategies

  • βœ… True MCP architecture: tools provide context, AI provides intelligence

  • βœ… Serverless cost optimization strategies

  • βœ… Security considerations for production MCP servers

GitHub Copilot Integration

  • βœ… MCP server configuration in VS Code

  • βœ… Tool discovery and usage patterns

  • βœ… Advanced integration scenarios

  • βœ… Troubleshooting and monitoring

🏁 Workshop Navigation

🌟 Codespaces Users (Recommended)

Start Here: Part 1: Setup and Understanding

🐧 Linux/macOS Users

Start Here: Part 1: Setup and Understanding


🌟 What Makes This Workshop Special

  1. Progressive Complexity: From simple local tools to production AI integration

  2. Cost Conscious: Carefully curated services for lowest possible cost.

  3. Real-World Ready: Production patterns with security considerations

  4. Complete Coverage: Local development β†’ Azure deployment β†’ Copilot integration β†’ AI enhancement

  5. Hands-On Testing: Comprehensive test scripts for every stage

🀝 Support and Contributing

πŸ“„ License

MIT License - see LICENSE file for details.


Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      B
      quality
      D
      maintenance
      Integrates GitHub Copilot with MCP-compatible tools to provide AI-powered code assistance, including chat, code explanation, and reviews. It leverages existing GitHub CLI authentication to support multiple models like GPT-4o and Claude 3.5 Sonnet.
      4
      2
      MIT
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables developers to build, run locally with debugging, and deploy to the cloud custom remote MCP endpoints hosted on Azure Functions, with built-in Microsoft Entra authentication. Sample projects demonstrate snippet CRUD and QR code tools, resource templates, prompt workflows, and an interactive weather demo UI.
      -
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables developers to run and deploy a custom remote MCP server on Azure Functions that exposes MCP tools, resources, and promptsβ€”such as snippet CRUD, QR code generation, code review checklists, and interactive app UIsβ€”with built-in Microsoft Entra authentication. Supports local debugging and one-command cloud deployment, with optional API Management and VNET network isolation.
      -
    • F
      license
      Not graded
      quality
      A
      maintenance
      Enables AI assistants to save and retrieve code snippets, generate language-specific style guides and comprehensive wiki documentation, and orchestrate multi-agent documentation workflows through MCP tools backed by vector search and Azure Functions.
      -