Many small sellers on marketplace platforms spend disproportionate time on repetitive customer messages, order confirmations, and manual post-sale workflows. This project bundles multi-account management, a keyword-based + AI reply pipeline, and browser automation so a single self-hosted service can automate messaging, shipping confirmation, and basic operational tasks for Xianyu accounts.
What Sets It Apart
- Combines browser automation (Playwright/DrissionPage) with an AI reply engine and priority-based keyword rules — so you can fallback from deterministic keyword replies to context-aware AI only when needed, reducing false responses.
- All-in-one deployability with Docker Compose and a FastAPI backend — so teams can run a single containerized stack (web UI, WebSocket status, SQLite storage) without stitching multiple services together.
- Multi-account isolation and operations-first features (timed/one-click item ‘refresh’, automated matching for auto-delivery) — so operators can scale from a single account to many while keeping data separated and configurable.
- Learning/research license and explicit non-commercial restriction — so it’s suitable for experimentation and development but requires caution before any production or commercial use.
Who It's For & Trade-offs
Great fit if you need a self-hosted automation layer for Xianyu that mixes rule-based and AI-driven replies and you are comfortable running Docker/Playwright. It’s useful for developers, hobbyist operators, and teams testing automation workflows. Look elsewhere if you require guaranteed compliance with marketplace TOS, enterprise-grade security, or horizontally scalable DB backends — the project uses SQLite and browser automation which can be brittle at large scale and may raise platform policy/legal considerations.
Where It Fits
This repo sits between lightweight chatbot scripts and full SaaS conversational platforms: it emphasizes operational automation (account/task management, auto-shipping) plus an AI-reply engine rather than providing a managed LLM service. Use it as a research/prototyping base or an internal ops tool with careful attention to platform rules and privacy.
