Simple Code Review Assistant

Focus: Model Context Protocol (MCP) Basics

🎯 Learning Objectives

By completing this assignment, you will:

• Build a basic MCP server using Model Context Protocol
• Implement simple tool interfaces for AI models
• Practice GitHub API integration
• Understand how AI models connect to external data sources

📋 Prerequisites

• Python 3.11+ experience
• Basic understanding of REST APIs
• Familiarity with GitHub API
• Basic knowledge of AI/LLM concepts

🚀 Quick Start

1. Environment Setup

# Clone and navigate to project
cd MCPAssignment

# Create virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

2. Configuration

Create .env file in the project root:

GITHUB_TOKEN=your_github_token_here
GITHUB_REPO_OWNER=your_github_username
GITHUB_REPO_NAME=your_repository_name

3. Run the MCP Server

# Start the MCP Server
python server.py

# Test the server (in another terminal)
python client.py

📁 Project Structure

mcp-code-review/
├── README.md                 # Your implementation documentation
├── requirements.txt          # Dependencies
├── .env.example             # Environment template
├── server.py                # Single MCP Server file
├── client.py                # Simple client to test server
└── docs/
    ├── api_guide.md         # Sample documentation
    └── setup_guide.md       # Sample documentation

🏗️ System Architecture

MCP Tools (Keep it Simple!)

Tool	Purpose	Implementation
get_repository	Get repo info from GitHub	GitHub API call
search_docs	Search local documentation	Simple file search
get_file_content	Read file from repo	GitHub API call

Simple Flow

Client → MCP Server → GitHub API / Local Files → Response

📋 Core Requirements (Simplified)

Must-Have Features (6-8 hours scope)

1. Basic MCP Server

   • Implement ONE MCP server file (server.py)
   • Support 3 simple tools (listed above)
   • Follow basic MCP protocol
   • Handle errors gracefully

2. GitHub Integration

   • Connect to GitHub API using token
   • Implement get_repository tool
   • Implement get_file_content tool
   • Add basic rate limiting

3. Documentation Search

   • Implement search_docs tool for local files
   • Search through markdown files in /docs folder
   • Return relevant file content
   • Support simple keyword matching

4. Simple Client

   • Create client.py to test your MCP server
   • Demonstrate all 3 tools working
   • Show real GitHub data retrieval
   • Display search results

🔧 Implementation Steps

Step 1: Setup

pip install mcp requests
# Create basic file structure
# Setup GitHub token

Step 2: Basic MCP Server

• Implement MCP protocol basics
• Add the 3 required tools
• Test with simple responses

Step 3: GitHub Integration

• Connect to GitHub API
• Implement repository and file tools
• Add error handling

Step 4: Documentation Search

• Create simple file search
• Add sample documentation files
• Test search functionality

Step 5: Client & Testing

• Build simple client
• Test all tools
• Create demo

📋 How You Will Be Evaluated

• Total marks: 100. Passing marks: 70/100.
• Evaluation focuses on functionality: MCP server runs, three tools (get_repository, get_file_content, search_docs) work, GitHub integration and docs search are demonstrated, and the client can exercise the tools. Basic implementations and reasonable variations in response structure are accepted. Code quality is secondary to a working, understandable solution.

📝 Submission Requirements (Minimal)

Required Files

• ✅ server.py - Working MCP server
• ✅ client.py - Simple test client
• ✅ requirements.txt - Dependencies
• ✅ .env.example - Environment template
• ✅ Sample docs in /docs folder

Demo Requirements

• ✅ Show MCP server starting up
• ✅ Demonstrate GitHub repository access
• ✅ Show documentation search working
• ✅ Explain your implementation approach

🔗 Resources

• MCP Python SDK
• GitHub REST API Documentation
• MCP Examples

🔧 Development Commands

# Start MCP Server
python server.py

# Test with client (in another terminal)
python client.py --test-all

# Test individual tools
python client.py --test-github
python client.py --test-docs