I spent the last 6 months building a bridge between Claude Desktop and my local Python quantitative models.

It uses the Model Context Protocol (MCP) to let Claude 'consult' a regime-switching LSTM model before answering questions about stocks.

The Backtest: 3-year run (2022-2025) vs S&P 500. It generated +9.2% alpha by moving to cash during the 2022 bear market https://www.investbuddy.ai/benchmarks/voo_benchmark_results.....

Repo/Install: npx -y investbuddy-mcp-server

Would love feedback