This project provides an AI-powered customer service agent system specifically designed for the Xianyu platform, enabling 24/7 automated customer interaction. It targets sellers on Xianyu looking to automate responses, manage negotiations, and provide consistent support, thereby saving time and improving efficiency.

How It Works

The system employs a multi-agent architecture where different "experts" (e.g., pricing, technical support, general customer service) handle specific tasks. An LLM-based prompt and rule routing system directs incoming queries to the appropriate expert agent. It features context-aware dialogue management, storing conversation history to provide LLMs with complete context for more natural and relevant responses. The pricing agent implements a tiered price reduction strategy for automated negotiation.

Quick Start & Requirements

• Install: Clone the repository, install dependencies with pip install -r requirements.txt.
• Prerequisites: Python 3.8+, NodeJS 18+, OpenAI API key, Xianyu cookies.
• Setup: Create a .env file for API keys and cookies.
• Run: Execute python main.py.
• Docs: Custom prompts can be edited in the prompts/ directory.

Highlighted Details

• Context-aware dialogue memory management.
• Multi-expert agent system with LLM-based intent recognition and dynamic dispatch.
• Automated negotiation with tiered price reduction strategies.
• Integrated web search for technical support queries.

Maintenance & Community

The project welcomes contributions via Issues and Pull Requests. Contact information for contributors is provided for inquiries and collaboration.

Licensing & Compatibility

The project is presented for learning and exchange purposes. The README includes a disclaimer that the development team may stop updates or delete the project at any time. Specific licensing details are not provided, and commercial use implications are unclear due to the project's stated nature and potential for discontinuation.

Limitations & Caveats

The project is explicitly stated to be for learning and exchange only, with a disclaimer about potential discontinuation. There is no mention of official support, testing, or a clear roadmap for future development, raising concerns about its long-term viability and stability for production use.