Stop letting your AI agent hallucinate option prices

What it does

AlphaGBM Skills is a collection of agent-ready tools that inject real market data into AI coding assistants. Each skill queries live or built-in mock data to return verifiable metrics—implied volatility, Greeks, VRP, skew, and quantitative scores—rather than generating plausible-sounding fiction. The package covers US, Hong Kong, China ETFs, and commodities, with a free tier that runs on demo data for tickers like AAPL and NVDA without an API key.

The interesting bit

The project treats AI agents as first-class users, not afterthoughts. Skills chain together—stock analysis feeds options scoring, which feeds strategy builders and P&L simulators—while a “research brain” workspace auto-refreshes company profiles and audits your investment theses weekly. It also quantifies famous investor philosophies (Buffett, Howard Marks, David Tepper) into one-call scoring tools, which is either brilliant or slightly unhinged.

Key highlights

• Modular skills spanning options analysis, risk management, macro tracking, and backtesting.
• Built-in demo dataset for AAPL, NVDA, SPY, TSLA, and META works offline without API keys.
• Real-time coverage includes US equities, HK stocks, China ETFs, and commodity futures options.
• Quantitative scoring models like G = B + M (Gain = Basics + Momentum) and per-ticker FearScore composites.
• Free tier offers limited daily analysis calls; paid tiers scale to thousands per month.

Caveats

• The README contradicts itself on the exact skill count, citing both 26 and 29.
• Live data depends on the alphagbm.zeabur.app endpoint and an API key; the free tier is capped at 2 stock and 1 options analysis per day.
• The hero screenshot is currently a placeholder flagged with a TODO in the source.

Verdict

Traders and quant-curious developers who want their AI assistant to cite real IV ranks instead of inventing them should look here. If you don’t trade options or already trust your broker’s app, this is just elaborate plumbing.