This repository contains multiple Model Context Protocol (MCP) servers built with FastMCP, providing various functionalities including authentication, financial data, portfolio management, and data visualization.

Purpose: Provides Azure Entra ID authentication for secure user identification and access control.

Features:

• Azure Entra ID integration
• OAuth 2.0 authentication flow
• User information retrieval (name, email, job title, office location)

Tools:

• get_user_info - Fetch authenticated user information from access token

Purpose: Fetch real-time stock quotes and historical price data from Yahoo Finance.

Features:

• Real-time stock quotes
• Historical OHLCV (Open, High, Low, Close, Volume) data
• Flexible date range and interval configurations

Tools:

• ping - Test connectivity
• get_quote - Fetch latest quote for a ticker symbol
• get_history - Download historical price data with customizable periods/dates

Purpose: Comprehensive investment portfolio management with persistent storage using Azure Cosmos DB.

Features:

• Portfolio holdings management (CRUD operations)
• Transaction history tracking
• Watchlist management
• Cash balance tracking
• API key-based authentication
• Azure Cosmos DB integration

Tools:

• add_to_portfolio - Purchase stocks and add to holdings
• remove_from_portfolio - Sell stocks and remove holdings
• update_position - Update existing holdings
• get_holdings - View all portfolio holdings
• get_transaction_history - View transaction history
• add_to_watchlist / remove_from_watchlist - Manage watchlist
• get_watchlist - View watchlist

Purpose: Create various data visualizations including charts, graphs, and plots.

Features:

• Multiple chart types (scatter, line, histogram, heatmap, pie)
• Relationship graphs with networkx
• Automatic plot saving and display
• Customizable styling and configurations

Tools:

• create_relationship_graph - Directed relationship graphs
• create_scatter_plot - Scatter plots with labels
• create_classification_plot - Category-colored scatter plots
• create_histogram - Distribution histograms
• create_line_plot - Time series and line charts
• create_heatmap - 2D matrix heatmaps
• create_pie_chart - Pie charts for proportions

• Python 3.10 or higher
• pip or uv package manager
• Docker Desktop (for Portfolio MCP only)
• Azure Entra ID credentials (for Auth MCP only)

1. Clone the repository:

git clone <repository-url>
cd ms-mcp

2. Install dependencies:

# Install all dependencies
pip install -r requirements.txt

# Or install per server using uv
cd auth_mcp && uv pip install -e .
cd localFinance && uv pip install -e .
cd PortfolioMcp && uv pip install -e .
cd vis_mcp && uv pip install -e .

1. Configure Azure Entra ID credentials in auth_mcp/main.py:

   • tenant_id
   • client_id
   • client_secret

2. Start the server:

cd auth_mcp
python main.py

The server runs on HTTP at http://localhost:8010

cd localFinance
python main.py

The server runs using stdio transport (standard input/output for MCP communication).

1. Start Azure Cosmos DB Emulator:

cd PortfolioMcp
docker-compose up -d

Wait for the Cosmos DB emulator to be healthy (takes ~60 seconds).

2. Configure environment variables (create .env file in PortfolioMcp/):

# Database Configuration
COSMOS_ENDPOINT=https://localhost:8081
COSMOS_KEY=C2y6yDjf5/R+ob0N8A7Cgv30VRDJIWEHLM+4QDU5DE2nQ9nDuVTqobD4b8mGGyPMbIZnqyMsEcaGQy67XIw/Jw==
COSMOS_DATABASE_NAME=portfolio_db

# Authentication
API_KEY=your-secure-api-key-here

# Logging
LOG_LEVEL=INFO

3. Start the server:

cd PortfolioMcp
python server.py

The server initializes database containers and repositories on first run.

cd vis_mcp
python server.py

The server runs using stdio transport. Generated plots are saved to the system temp directory and automatically opened.

Add to your .vscode/mcp.json or MCP client settings:

Windows:

{
  "mcpServers": {
    "mslab/auth": {
      "command": "${workspaceFolder}/auth_mcp/.venv/Scripts/python.exe",
      "args": ["main.py"],
      "cwd": "${workspaceFolder}/auth_mcp",
      "transport": "http",
    },
    "mslab/finance": {
      "command": "${workspaceFolder}/localFinance/.venv/Scripts/python.exe",
      "args": ["main.py"],
      "cwd": "${workspaceFolder}/localFinance"
    },
    "mslab/portfolio": {
      "command": "${workspaceFolder}/PortfolioMcp/.venv/Scripts/python.exe",
      "args": ["server.py"],
      "cwd": "${workspaceFolder}/PortfolioMcp",
      "env": {
        "API_KEY": "your-secure-api-key-here"
      }
    },
    "mslab/visualization": {
      "command": "${workspaceFolder}/vis_mcp/.venv/Scripts/python.exe",
      "args": ["server.py"],
      "cwd": "${workspaceFolder}/vis_mcp"
    }
  }
}

Linux/macOS:

{
  "mcpServers": {
    "mslab/auth": {
      "command": "${workspaceFolder}/auth_mcp/.venv/bin/python",
      "args": ["main.py"],
      "cwd": "${workspaceFolder}/auth_mcp",
      "transport": "http",
    },
    "mslab/finance": {
      "command": "${workspaceFolder}/localFinance/.venv/bin/python",
      "args": ["main.py"],
      "cwd": "${workspaceFolder}/localFinance"
    },
    "mslab/portfolio": {
      "command": "${workspaceFolder}/PortfolioMcp/.venv/bin/python",
      "args": ["server.py"],
      "cwd": "${workspaceFolder}/PortfolioMcp",
      "env": {
        "API_KEY": "your-secure-api-key-here"
      }
    },
    "mslab/visualization": {
      "command": "${workspaceFolder}/vis_mcp/.venv/bin/python",
      "args": ["server.py"],
      "cwd": "${workspaceFolder}/vis_mcp"
    }
  }
}

Note: The ${workspaceFolder} variable automatically resolves to your workspace root directory. Make sure each server has its own .venv directory with dependencies installed.

Create and configure virtual environments for each server:

Windows:

# Navigate to each server directory and set up venv
cd auth_mcp
python -m venv .venv
.venv\Scripts\activate
pip install -e .
deactivate

cd ../localFinance
python -m venv .venv
.venv\Scripts\activate
pip install -e .
deactivate

cd ../PortfolioMcp
python -m venv .venv
.venv\Scripts\activate
pip install -e .
deactivate

cd ../vis_mcp
python -m venv .venv
.venv\Scripts\activate
pip install -e .
deactivate

Linux/macOS:

# Navigate to each server directory and set up venv
cd auth_mcp
python -m venv .venv
source .venv/bin/activate
pip install -e .
deactivate

cd ../localFinance
python -m venv .venv
source .venv/bin/activate
pip install -e .
deactivate

cd ../PortfolioMcp
python -m venv .venv
source .venv/bin/activate
pip install -e .
deactivate

cd ../vis_mcp
python -m venv .venv
source .venv/bin/activate
pip install -e .
deactivate

ms-mcp/
├── README.md                      # This file
├── requirements.txt               # All dependencies
├── auth_mcp/                      # Authentication MCP
│   ├── main.py                    # Server entry point
│   └── pyproject.toml             # Project metadata
├── localFinance/                  # Finance data MCP
│   ├── main.py                    # Server entry point
│   ├── finance_tools.py           # Yahoo Finance integration
│   ├── pyproject.toml             # Project metadata
│   └── requirements.txt           # Dependencies
├── PortfolioMcp/                  # Portfolio management MCP
│   ├── server.py                  # Server entry point
│   ├── docker-compose.yml         # Cosmos DB emulator
│   ├── pyproject.toml             # Project metadata
│   ├── config/                    # Configuration modules
│   ├── database/                  # Database layer
│   ├── models/                    # Domain models
│   ├── services/                  # Business logic
│   └── tools/                     # MCP tools
└── vis_mcp/                       # Visualization MCP
    ├── server.py                  # Server entry point
    ├── config.py                  # Plot styling
    ├── plot_utils.py              # Utilities
    ├── pyproject.toml             # Project metadata
    ├── README.md                  # Detailed docs
    └── tools/                     # Visualization tools

• Auth MCP: Never commit Azure credentials to version control. Use environment variables or secure secret management.
• Portfolio MCP: Change the default API key before production use. The Cosmos DB emulator key is for development only.
• All MCPs: Run in trusted environments when exposing sensitive financial or user data.

Test each server's connectivity:

# Finance MCP
curl -X POST http://localhost:<port> -d '{"tool": "ping"}'

# Portfolio MCP (with API key)
curl -H "Authorization: Bearer your-api-key" http://localhost:<port>

• fastmcp - MCP server framework
• yfinance - Yahoo Finance data
• azure-cosmos - Azure Cosmos DB client
• matplotlib, networkx, plotly - Visualization libraries
• uvicorn - ASGI server
• pydantic - Data validation

Each MCP server is designed to be modular and independent. Feel free to:

• Add new tools to existing servers
• Create additional MCP servers
• Enhance authentication and security
• Improve data persistence and caching

[Specify your license here]

• Model Context Protocol Documentation
• FastMCP Framework
• Azure Cosmos DB
• Yahoo Finance API (yfinance)