What it does AI-Trader is a live trading platform where both humans and AI agents can publish signals, mirror trades, and paper-trade across stocks, crypto, forex, and prediction markets. The twist: agents onboard themselves by reading a SKILL.md file and registering automatically—no human wiring required.

The interesting bit The project flips the typical integration model. Instead of you plugging an API into your bot, you tell your agent to read a skill definition and it joins the platform, installs dependencies, and starts trading. The README claims compatibility with “OpenClaw, nanobot, Claude Code, Codex, Cursor, and more”—though what “support” means in practice is left vague.

Key highlights

• Agent self-onboarding via https://ai4trade.ai/SKILL.md
• Paper trading with $100K simulated capital and Polymarket integration
• Copy trading: follow top performers and mirror positions automatically
• Cross-broker sync (Binance, Coinbase, Interactive Brokers mentioned)
• FastAPI backend split from background workers for stability; React frontend
• Modular skill definitions in skills/ for trading, copy-trading, and trade-sync roles

Caveats

• The “100% Fully-Automated” claim in the title is marketing; the platform facilitates automation but still requires broker connections and human oversight for real money
• No performance metrics, backtests, or risk disclosures visible in the README
• “Agent-native” is a neat onboarding trick, but the actual trading intelligence still depends on whatever model the user brings

Verdict Worth a look if you’re experimenting with LLM agents in financial contexts or building a social trading product. Skip it if you want proven alpha strategies—the smarts are yours to provide.