Options Sentinel—Options Trading via AI & MCP

Options Sentinel is an AI-driven options trading system built around one core principle: AI can recommend, but it cannot authorize. The system converts live market data into a structured context (price, volume, volatility, and regime detection), then runs a multi-agent debate — a Bull Agent and Bear Agent argue opposing theses, and a Decision Agent synthesizes their reasoning into a single trade proposal. Before anything reaches a broker, every proposal passes through the Risk Gate: a deterministic, non-AI validation layer that checks position size, account equity, and max allowed risk against hard-coded rules. Trades that exceed risk thresholds are blocked outright, regardless of how confident the AI agents were. Approved trades are routed through an MCP (Model Context Protocol) server, which acts as the controlled interface between Options Sentinel and Alpaca's paper trading API — giving the AI system tool access without direct, unrestricted account control. Every decision is logged in a trade journal (bull/bear confidence, risk outcome, and consensus) and monitored continuously, so the system tracks not just what it traded, but why. The result is an autonomous trading loop — Observe → Debate → Decide → Validate → Execute → Monitor → Learn — designed for safety and auditability rather than unrestricted AI autonomy.