furidngrt/ARBITRAGE-PEDIA — reverse-engineered prompt

Build me a Node.js app called ARBITRAGE PEDIA that helps me find marketplace arbitrage opportunities in Indonesia.

I want it to scan Tokopedia, Shopee, and Bukalapak for used or underpriced items, especially electronics, gaming gear, phones, laptops, monitors, keyboards, consoles, and tablets. It should filter out bad listings, broken items, spare parts, low rated sellers, and suspicious sellers. Then it should compare market prices, estimate fees and shipping, calculate expected profit, and only recommend items that look worth flipping.

Use an LLM through an OpenAI compatible API to reason about each candidate. Send the best opportunities to Telegram with a clear reason, estimated profit, confidence score, and Buy or Skip buttons. Do not auto buy anything, keep buying manual for safety.

Include a dry run mode with mock data, simple config for budget and profit rules, a command line interface with status and reports, and a learning log that improves rules after sales are recorded. Look up current docs online if needed.