The problem
Cross-border e-commerce sellers who source from 1688.com (China's B2B wholesale platform) and list on eBay, Etsy, Amazon, and AliExpress spend hours on manual work: translating product titles, filtering out branded items, checking images for watermarks, and uploading to each platform one by one.
I was doing this myself and decided to build AutoStore to automate it.
What it does
AutoStore is a free macOS app + Chrome extension that runs the full pipeline:
- Discovers products on 1688 across 450+ blue-ocean categories
- Brand filter — 548-brand blacklist scans titles/descriptions/specs to avoid IP violations
- OCR image check — Vision framework detects watermarks and Chinese text in product photos
- On-device AI translation — bundled LLM translates titles and descriptions (no API key, no cost)
- Bulk upload — drives your already-logged-in Chrome sessions on each platform via extension
Technical approach
The key insight: instead of trying to automate platform logins (which triggers bot detection), we drive the seller's existing authenticated browser session through a Chrome extension. The platform sees a normal human user, not a bot.
All data is stored locally — MySQL for products, macOS Keychain for credentials. No cloud dependency, PIPL compliant.
Stack
- Mac app: SwiftUI, Vision framework (OCR), URLSession
- Chrome extension: Playwright-based automation injected into seller dashboards
- Local LLM: bundled llama.cpp server for on-device translation
- Database: MySQL (local)
It's free
No subscription, no commission, no usage limits. Download at autostore.spriterock.com
Happy to answer questions about the architecture, the 1688 pipeline, or the brand safety system.
Top comments (0)