About this project

Market Agent is a data collection and strategy signal analysis project for Polymarket prediction markets, designed with a dual-engine architecture: Sniper and Radar. Sniper uses Puppeteer to read topics tagged with [poly] from Central-Bank's Issues, performing precise searches on Polymarket; Radar uses Axios to call APIs and scan the global Top 1000 markets, excluding content already collected by Sniper, and takes the Top 3 from each sector. Collected results are output to the data/strategy/ and data/trends/ directories respectively, after which the Harvest process copies them to the Central-Bank repository, verifies file sizes, and cleans up local data. The project includes a built-in strategy engine that tags market signals from four perspectives: Taleb, Soros, Munger, and Naval. Trigger conditions involve metrics such as price range, liquidity, 24-hour trading volume, daily change, spread, and the TECH category; signals that do not match any strategy are marked as RAW_MARKET. Radar's sector classification includes POLITICS, ECONOMY, CRYPTO, TECH, GEOPOLITICS, WORLD, FINANCE, and CLIMATE-SCIENCE. The automation process runs on ubuntu-latest via GitHub Actions, using Node.js 20, triggered at minute 15 of every hour: first executing Sniper collection, waiting 10 seconds, then running Radar scanning, and finally automatically triggering Harvest. Environment variables support MY_PAT (a GitHub Token for reading and writing Central-Bank) and optional PUPPETEER_EXECUTABLE_PATH (custom Chrome path). The project works in conjunction with the Central-Bank and Refinery-Engine repositories, which serve as raw data storage and command center, and data cleaning and AI auditing, respectively.