Xianyu Automatic Publishing System

Problem Being Addressed

Publishing secondhand items on Xianyu often requires repeating browser-based steps: logging in, searching for matching products, downloading main images, cleaning up titles and descriptions, and filling in condition and price. This skill targets that repetitive listing workflow by automating QR-code login, product information enrichment, image collection, and publish actions, while saving cookies to a local file for later runs.

How It Works

The implementation is split into multiple Python scripts. main.py is the full automation entry point. login.py performs QR-code login and stores the session cookie in cookies.json. publish.py handles the publishing action, get_product_images.py searches for matching products and downloads images, product_info.py parses product descriptions and fills missing information, and stealth.py provides anti-detection behavior. The expected input format is brand model condition price, where price and condition can be omitted and estimated by the system. Image downloading prefers the detail page and requires a natural width of at least 400px; if the CDN request fails, it falls back to a thumbnail. Listings whose titles contain broken-device keywords such as cracked screen or water damage are skipped automatically.

Boundaries and Caveats

The browser opens in non-headless mode, so it should not be closed while the flow is running. A greyed-out publish button usually means images have not finished uploading; retry after a short wait. Expired cookies require re-running the login script. If search returns no results, the account may be rate-limited; waiting a few hours or performing some normal manual interactions on Xianyu first may be necessary.