Multi Transport Server

A comprehensive Model Context Protocol (MCP) implementation featuring an expense tracking system with both local and remote server configurations, along with multiple client interfaces.

🎯 Overview

This project demonstrates a production-ready MCP server implementation for expense tracking, showcasing:

• Dual Server Architecture: Local (STDIO) and Remote (HTTP) implementations
• Multiple Client Interfaces: CLI and Web-based clients
• Modern Tech Stack: FastMCP, LangChain, Streamlit, and async Python
• Real-world Application: Complete expense management with categories, summaries, and date-range queries

🏗️ Architecture

mcp-expense-tracker/
├── servers/
│   ├── local-server/      # STDIO-based MCP server
│   └── remote-server/     # HTTP-based MCP server with async support
├── clients/
│   ├── cli-client/         # Command-line interface
│   └── web-client/        # Streamlit web interface
├── docs/                   # Documentation and learning resources
└── README.md

✨ Features

MCP Tools

• add_expense: Add new expense entries with category, subcategory, and notes
• list_expenses: Query expenses within date ranges
• summarize: Generate category-wise expense summaries
• categories: Resource endpoint for expense categories

Server Implementations

• Local Server: Synchronous SQLite-based server using STDIO transport
• Remote Server: Async aiosqlite-based server with HTTP transport support
• Proxy Server: FastMCP Cloud proxy for remote deployments

Client Implementations

• CLI Client: Async command-line client with LangChain integration
• Web Client: Interactive Streamlit chat interface with tool calling

🚀 Quick Start

Prerequisites

• Python 3.11+
• OpenAI API key (for clients)
• FastMCP framework

Local Server Setup

cd servers/local-server
pip install -e .
python main.py

Remote Server Setup

cd servers/remote-server
pip install -e .
python main.py  # Runs on http://0.0.0.0:8000

CLI Client

cd clients/cli-client
pip install -r requirements.txt
# Set OPENAI_API_KEY in .env
python client.py

Web Client

cd clients/web-client
pip install -r requirements.txt
# Set OPENAI_API_KEY in .env
streamlit run app.py

📊 Database Schema

CREATE TABLE expenses(
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    date TEXT NOT NULL,
    amount REAL NOT NULL,
    category TEXT NOT NULL,
    subcategory TEXT DEFAULT '',
    note TEXT DEFAULT ''
)

🔧 Technology Stack

• FastMCP: MCP server framework
• LangChain: LLM orchestration and tool integration
• Streamlit: Web interface
• SQLite/aiosqlite: Database storage
• OpenAI API: LLM backend

📁 Project Structure

• servers/local-server/: Synchronous MCP server with STDIO transport
• servers/remote-server/: Async MCP server with HTTP transport
• clients/cli-client/: Command-line client implementation
• clients/web-client/: Streamlit web application
• docs/: Comprehensive documentation and learning resources (PDFs)

📚 Documentation

Comprehensive documentation is available in the docs/ folder:

• How to Build Local MCP Servers.pdf - Local server implementation guide
• MCP Client — Quick Revision Notes.pdf - Client development reference
• MCP Trilogy.pdf - Complete MCP protocol overview
• Remote MCP Servers.pdf - Remote deployment guide

These documents provide the theoretical foundation and implementation patterns used throughout this project.

🎓 Learning Resources

This project demonstrates:

• MCP protocol implementation
• Multiple transport types (STDIO, HTTP, SSE)
• Async/await patterns in Python
• LangChain tool integration
• FastMCP framework usage
• Production-ready code structure

📝 License

This project is part of a learning portfolio demonstrating MCP protocol expertise.

🤝 Contributing

This is a portfolio project showcasing MCP implementation skills.

---

Built with ❤️ using FastMCP and LangChain