Expense Tracker MCP

A remote Model Context Protocol (MCP) server built with FastMCP that allows AI assistants such as Claude to interact with an expense-tracking system through structured tools.

🚀 Overview

This project demonstrates how an AI assistant can interact with external data and application functionality through MCP.

The Expense Tracker MCP server uses FastMCP to expose expense-management tools and SQLite to store expense data.

The server is deployed remotely using FastMCP Cloud, allowing an MCP-compatible client such as Claude to connect to it through a remote MCP endpoint.

Related MCP server: Expense Tracker MCP

🏗️ Architecture

Claude
   │
   │ MCP
   ▼
Remote MCP Endpoint
   │
   ▼
FastMCP Cloud
   │
   ▼
FastMCP Expense Tracker Server
   │
   ▼
SQLite Database

✨ Features

• Add and manage expenses

• Retrieve expense records

• Categorize expenses

• Store expense data using SQLite

• Expose expense functionality through MCP tools

• Deploy the MCP server remotely

• Connect the remote MCP server to Claude

• Allow AI assistants to interact with structured expense data

🛠️ Tech Stack

• Python

• FastMCP

• Model Context Protocol (MCP)

• SQLite

• JSON

• FastMCP Cloud

• Claude

🔌 Example Interactions

Once connected to Claude, users can interact with the expense tracker using natural language.

"Add an expense of ₹500 for groceries."

"Show me my recent expenses."

"How much did I spend on food?"

"List my expenses by category."

Claude interprets the user's request and invokes the appropriate MCP tool exposed by the server.

☁️ Remote Deployment

The MCP server is deployed on FastMCP Cloud and exposed through a remote MCP endpoint.

This allows Claude and other MCP-compatible clients to access the server without running it locally.

Local Development
       ↓
FastMCP Server
       ↓
FastMCP Cloud
       ↓
Remote MCP Endpoint
       ↓
Claude

⚙️ Local Setup

Clone the repository:

git clone https://github.com/khushisonwane23/expense-tracker-mcp.git
cd expense-tracker-mcp

Create a virtual environment:

python -m venv .venv

Activate the virtual environment on Windows:

.venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

If you are using uv:

uv sync

▶️ Run Locally

Run the FastMCP server:

fastmcp run server.py

The exact command may vary depending on the project configuration.

🔗 Connecting to Claude

After deploying the server to FastMCP Cloud, the application provides a remote MCP endpoint.

This endpoint can be configured in an MCP-compatible client such as Claude.

Claude
   ↓
Remote MCP Endpoint
   ↓
FastMCP Cloud
   ↓
Expense Tracker MCP Server
   ↓
SQLite Database

Once connected, Claude can discover and use the tools exposed by the MCP server.

🔐 Security

Sensitive information should never be committed to this repository.

The following files should remain private:

.env
expenses.db
.venv/
__pycache__/

API keys and other secrets should be stored using environment variables instead of being hard-coded in the source code.

🎯 Learning Goals

This project was built to understand:

• How the Model Context Protocol works

• How AI assistants interact with external tools

• How to build MCP servers using FastMCP

• How to connect LLMs with external data

• How tool-based AI workflows work

• How to deploy an MCP server remotely

• How MCP can be integrated with Claude

🚧 Future Improvements

• Add authentication and authorization

• Add monthly spending analytics

• Add budget tracking

• Add richer financial insights

• Improve error handling and input validation

• Add automated testing

• Use a production-grade database

• Add more financial management tools

👩‍💻 Author

Khushi Sonwane

Artificial Intelligence & Robotics Student

Interested in Generative AI, AI Agents, MCP, RAG, and AI Research.