MCP-Data-Service-for-Financial-Analytics

MCP-Data-Service-for-Financial-Analytics

A Model Context Protocol (MCP) server implementation providing real time and historical financial data analytics through natural language queries. Developed with FastMCP and deployed using Railway, this service integrates CoinGecko and yfinance APIs to deliver comprehensive financial insights with automatic chart generation and IST timezone support.

Overview

This asynchronous MCP server enables large language model clients to perform advanced financial analysis via natural language queries. The service automatically retrieves data, computes relevant metrics, generates visualizations, and formats responses for conversational interaction.

Key Features

• Real-time and historical data retrieval for stocks and cryptocurrencies
• Secure access using Bearer token authentication with RSA key pairs
• Performance optimizations, including data downsampling and in-memory rendering
• Automatic IST timezone conversion for Indian market data
• Chart generation for graphical analysis of price trends
• Additional tools: vintage photo filter and daily horoscope service

Architecture

The system adheres to the Model Context Protocol specification, exposing discoverable tools to MCP clients.

MCP Client MCP Server Data Sources (CoinGecko, Yahoo Finance)

Clients issue natural language queries, which the server maps to specific tools. The server fetches data, processes it, and returns formatted text and images.

Quick Start

Prerequisites

• Python 3.8 or higher
• Railway account for deployment
• MCP-compatible client (e.g., Puchai)

Local Setup

1. Clone the repository:

   git clone https://github.com/harshitkapoor03/MCP-Data-Service-for-Financial-Analytics.git
   cd MCP-Data-Service-for-Financial-Analytics
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Create a .env file with the following variables:

   AUTH_TOKEN=YOUR_TOKEN
   MY_NUMBER=YOUR_NUMBER
   

4. Run the server:

   python mcp-server.py
   

   The server will be accessible at http://localhost:8086.

Deployment on Railway

1. Connect the GitHub repository to Railway.
2. Set environment variables (AUTH_TOKEN, MY_NUMBER) in the Railway dashboard.
3. Deploy the application; Railway will provide a public URL.

Configuration

Authentication

The server uses a custom BearerAuthProvider:

from fastmcp.server.auth.providers.bearer import BearerAuthProvider, RSAKeyPair
from mcp.server.auth.provider import AccessToken

class SimpleBearerAuthProvider(BearerAuthProvider):
    def __init__(self, token: str):
        key_pair = RSAKeyPair.generate()
        super().__init__(public_key=key_pair.public_key, jwks_uri=None, issuer=None, audience=None)
        self.token = token

    async def load_access_token(self, token: str) -> AccessToken | None:
        if token == self.token:
            return AccessToken(token=token, client_id="client", scopes=["*"], expires_at=None)
        return None

Connecting an MCP Client

Configure your MCP client (for example, Puchai) to connect to the deployed server:

{
  "servers": {
    "financial-analytics": {
      "url": "https://.up.railway.app/mcp",
      "auth": {
        "type": "bearer",
        "token": "730E024D"
      }
    }
  }
}

Available Tools

Stock Time Series Analysis

Cryptocurrency Time Series Analysis

Vintage Photo Filter

Daily Horoscope

Data Sources

• yfinance for real-time and historical stock data
• CoinGecko API for cryptocurrency market data

Performance Optimizations

• Data point downsampling for efficient chart rendering
• In-memory chart generation to eliminate disk I/O
• Caching of repeated API calls
• IST timezone conversion for accurate time display

File Structure

├── mcp-server.py       # Server implementation
├── requirements.txt    # Dependencies
├── env.txt             # Environment variable template
├── README.md           # Project documentation
└── .railway/           # Deployment configuration

License

This project is licensed under the MIT License.