polymarket-copy-trading-skill
github.com/polytrackers/polymarket-copy-trading-skill
MCP server enabling copy trading and paper trading for Polymarket via AI agents, with whale tracking and $10k mock wallets
AI-agent-driven copy trading via MCP protocol: tracks whale wallets, generates copy signals, executes paper trades via $10k mock wallets
This repository implements an MCP (Model Context Protocol) server that allows AI agents like Claude to interact with Polymarket for copy trading and paper trading. The skill-based architecture ('npx skills add polytrackers/polymarket-copy-trading-skill') makes it easily installable. The core functionality includes whale wallet tracking, copy signal generation based on whale activity patterns, and a $10k mock wallet system for paper trading. This is particularly interesting for the AI agent ecosystem — it bridges the gap between natural language trading commands and Polymarket's CLOB API. The MCP protocol enables structured tool definitions that LLMs can use deterministically. At 143KB with clean structure, the codebase is focused and well-organized. The whale tracking algorithm likely uses on-chain data aggregators or Polymarket's API to identify large traders and their position changes. Suitable for developers building AI trading agents or those wanting to experiment with agent-driven trading strategies. The mock wallet system allows safe testing. No malware indicators; the npm/npx distribution model provides supply chain security. The concept is forward-looking as AI agent trading becomes more prevalent.