innovatics-zeel/Retail_Poc — reverse-engineered prompt

Build me a demo retail intelligence app for US apparel marketplaces. I want it to scrape product data, starting with Nordstrom men’s T shirts and women’s dresses, save everything into a local PostgreSQL database, and show the results in a clean Streamlit dashboard.

The dashboard should have tabs for basic business stats, charts for prices, ratings, brands and attributes, a Claude powered chat where I can ask questions about the live product data, simple daily trend forecasts, and ranked buying or sourcing recommendations. For each recommendation I should be able to accept it, modify it, or dismiss it, and that feedback should be saved.

Please make the scraper setup modular so new marketplaces or categories can be added without rewriting the whole pipeline. Include safe demo defaults, environment variable setup for database and Claude keys, database migrations, a manual scrape command, and an optional daily scheduler. Look up current docs online if you need to.