public-dot-com

**Quick judgment** This is a powerful, production-grade integration for **live trading on Public.com**, not a toy. Functionality coverage is excellent, but the financial and operational risks are substantial and require careful, expert use and strict isolation. **What it’s good for** - Letting an AI agent **inspect** your Public.com accounts: accounts list, portfolio, orders, and transaction history. - Getting **market data and options analytics**: quotes, instruments, option chains/expirations, and greeks. - Running **preflight cost and buying-power checks** before trades. - **Placing and cancelling live orders** for equities, options, and crypto when a human explicitly confirms details. - Exploring **advanced options strategies** using the provided options automation library and Public.com SDK-based workflows. **Key risks & concerns** - **Real-money trading risk:** The skill can place and cancel live orders. Any misinterpretation, hallucinated symbol, wrong quantity, or incorrect side (BUY vs SELL) can cause **financial loss**. - **Complex options workflows:** OSI symbol construction, expirations, and multi-leg strategies are error-prone; bugs or model mistakes can materially change payoff profiles. - **Account & key security:** Requires a **Public.com API secret** stored in OpenClaw config. Misconfigured environments or shared instances increase the blast radius if something goes wrong. - **Asynchronous order status:** Orders and cancellations are asynchronous; relying on them as “done” without re-checking status can lead to unexpected positions. **Safety recommendations** - Prefer **read-only style usage** (accounts, portfolio, history, quotes, chains, greeks) unless you fully understand the risks. - If you do trade, **run in as isolated an instance as possible**, ideally with a **small or dedicated Public.com account** and strict limits. - Always have the agent **echo back full order details** (symbol, type, side, order type, quantity/amount, prices, time-in-force, session) and require explicit human approval before calling `place_order.py` or `cancel_order.py`. - Use **preflight checks** before any order, and re-verify positions/orders afterwards. - Treat all outputs as **informational only**, not investment advice or recommendations.