rajeevpareek/credit_risk_mcp_server

Credit Risk MCP Server

A Model Context Protocol (MCP) server that provides credit risk portfolio analytics and connects to a simulated Risk Data Mart. This server enables AI assistants to perform sophisticated credit risk analysis through a standardized interface.

Overview

The Credit Risk MCP Server demonstrates how MCP can be used to integrate credit risk data and analytics into AI workflows. It provides access to portfolio data including auto loans, mortgages, credit cards, personal loans, and commercial loans with comprehensive risk metrics.

Features

• Portfolio Search & Filtering: Search portfolios by name, product type, or risk metrics
• Risk Analytics: Calculate delinquency rates, default rates, and risk-adjusted returns
• Portfolio Rankings: Identify top-performing or highest-risk portfolios
• Product Type Analysis: Aggregate statistics by loan product categories
• Derived Metrics: Calculate expected loss, portfolio quality scores, and exposure metrics

Dataset

The server includes 8 sample portfolios across multiple product types:

• Auto Loans (Prime & Subprime)
• Mortgages (Prime & Jumbo)
• Credit Cards
• Personal Loans
• Commercial Loans
• Commercial Real Estate

Each portfolio includes:

• Volume and book value
• Delinquency and default rates
• Average credit scores
• Geographic regions
• Vintage information

Installation

# Clone the repository
git clone <repository-url>
cd credit-risk-mcp-server

# Install dependencies
pip install mcp pydantic

# Make the script executable
chmod +x credit_risk_mcp.py

Usage

Running the Server

python credit_risk_mcp.py

Available Tools

1. search_portfolios: Search by portfolio name or product type
2. get_portfolio_by_id: Retrieve detailed portfolio information
3. filter_by_risk_metrics: Filter by delinquency/default rate thresholds
4. get_top_portfolios: Rank portfolios by book value, volume, or risk
5. analyze_by_product_type: Aggregate statistics by product category
6. get_highest_risk_portfolios: Identify highest-risk portfolios
7. calculate_portfolio_metrics: Compute risk-adjusted metrics

Available Resources

• risk://portfolio/all - Complete portfolio dataset
• risk://portfolio/summary - Aggregate summary statistics

Example Queries

Find high-risk portfolios:

"Show me portfolios with delinquency rates above 5%"

Compare product types:

"What's the average default rate for auto loans vs mortgages?"

Calculate risk metrics:

"Calculate the expected loss for portfolio ID 3"

Top performers:

"Show me the top 3 portfolios by book value"

Configuration with Claude Desktop

Add to your Claude Desktop configuration file:

MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json

Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "credit-risk": {
      "command": "python",
      "args": ["/path/to/credit_risk_mcp.py"]
    }
  }
}

API Reference

Tool: search_portfolios

{
  "query": "string (portfolio name or product type)"
}

Tool: filter_by_risk_metrics

{
  "max_delinquency_rate": "number (optional)",
  "max_default_rate": "number (optional)"
}

Tool: get_top_portfolios

{
  "metric": "book_value | volume | lowest_delinquency | lowest_default",
  "limit": "integer (default: 5)"
}

Tool: calculate_portfolio_metrics

Returns:

• Risk-adjusted value
• Exposure per account
• Delinquency dollar amount
• Expected loss
• Portfolio quality score

Development

Requirements

• Python 3.8+
• mcp
• pydantic

Extending the Server

To add more portfolios, edit the CREDIT_RISK_PORTFOLIO list in the source code. Each portfolio should include:

• portfolio_id
• portfolio_name
• product_type
• volume
• book_value_usd
• delinquency_rate
• default_rate
• avg_credit_score
• avg_loan_size
• vintage
• geographic_region

Use Cases

• Credit Risk Assessment: Analyze portfolio risk profiles
• Portfolio Optimization: Identify underperforming segments
• Regulatory Reporting: Generate risk metrics for compliance
• Investment Analysis: Evaluate portfolio quality and returns
• Stress Testing: Assess portfolios under different risk scenarios

License

MIT License

Contributing

Contributions are welcome! Please submit pull requests or open issues for bugs and feature requests.

Disclaimer

This is a demonstration server with simulated data. Do not use for actual credit risk decisions without proper validation and compliance review.