Automatically reply to buyer inquiries on Xianyu using AI or keyword rules so your store never goes silent overnight.
Manage multiple Xianyu seller accounts and product listings from a single React dashboard.
Automatically ship digital goods and confirm receipt on orders without manual steps.
Schedule items to be re-listed on a timer so they stay visible in Xianyu search results.
Requires MySQL 8.0, Redis 7, and Python 3.11 on Linux or 3.12 on Windows, Docker Compose is recommended for production, documentation is in Chinese.
xianyu-auto-reply is an automated customer service and store management system built for Xianyu, a second-hand marketplace operated by Alibaba, similar to eBay or Facebook Marketplace but widely used in China. The system connects to Xianyu's messaging service via WebSocket, receives buyer inquiries in real time, and replies automatically using either keyword matching or an AI assistant backed by the OpenAI API. The project is structured as a microservice application with six components: a web backend built in Python and FastAPI, a WebSocket service that handles real-time messaging using Playwright for browser automation, a scheduler for timed tasks like re-listing items and rate-limit tracking, a MySQL database, a Redis cache, and a React-based frontend dashboard. These components can be deployed together via Docker Compose, run locally for development, or packaged as a standalone Windows EXE. Beyond automated replies, the system handles a broad range of store operations: fetching and managing orders, managing product listings, automatic shipment of digital goods (card codes), automatic confirmation of receipt, automatic buyer reviews, scheduled re-listing of items to keep them visible in search results, batch publishing of new listings, and automatic cookie refresh to maintain account login state. Multiple Xianyu seller accounts can be managed simultaneously. The admin dashboard includes user account management with multiple roles, a data statistics panel, a risk control log, and support for multi-level reseller structures. An optional promotions module handles affiliate link integration and commission tracking. Deployment requires Python 3.11 on Linux or Python 3.12 on Windows, along with MySQL 8.0 and Redis 7. Docker Compose is the recommended method for production. The project documentation is written in Chinese.
← zhinianboke on gitmyhub — every repo by this author, as a profile.
Verify against the repo before relying on details.