Summary

This project provides an enterprise-grade multi-agent intelligent customer service system for financial/e-commerce scenarios. It targets developers preparing for AI/backend interviews, offering robust multi-agent orchestration, layered memory, standardized tool interaction (MCP), and RAG. Implemented in Python, Java, and Go, it includes comprehensive interview materials to enhance candidate profiles.

How It Works

A Supervisor agent orchestrates specialized sub-agents (intent routing, knowledge retrieval, ticket handling, compliance) via a directed graph, enabling central control, parallel execution, and breakpoint recovery. A three-layered memory system (work, short-term Redis, long-term vector DB) manages context and knowledge. The Model Context Protocol (MCP) standardizes AI agent interaction with external tools. Retrieval-Augmented Generation (RAG) ensures accurate knowledge retrieval. Full-link tracing via OpenTelemetry visualizes agent interactions.

Quick Start & Requirements

• Primary Install/Run Command: Docker Compose for one-click setup is recommended; manual setup for Python, Java, or Go is supported.
• Prerequisites: Mandatory OpenAI API Key (or LLM provider equivalent); Docker recommended.
• Links: Repository: https://github.com/bcefghj/smart-cs-multi-agent.git; API Docs: http://localhost:8000/docs; Tracing UI: http://localhost:16686.

Highlighted Details

• Supervisor Orchestration: Centralized coordination, parallel scheduling, human-in-the-loop, breakpoint recovery.
• Layered Memory: Work, short-term (Redis), and long-term (vector DB) memory.
• MCP Tool Protocol: Standardized JSON protocol for AI-tool interaction.
• RAG Implementation: Advanced knowledge retrieval.
• Full-Link Tracing: OpenTelemetry for agent interaction visibility.
• Compliance Module: Financial scenario checks.
• Multi-Language Support: Python (LangGraph), Java (Spring AI), Go (Eino).

Maintenance & Community

No specific details on maintenance, contributors, sponsorships, or community channels are provided.

Licensing & Compatibility

• License Type: MIT License.
• Compatibility: Permissive MIT license allows free use, modification, and distribution, including commercial and closed-source integration.

Limitations & Caveats

The Go implementation has 3/5 production maturity stars, potentially less mature than Python (5/5) or Java (4/5). The system requires paid LLM API access. Docker simplifies setup, but manual configuration may be complex. The project is primarily an interview preparation tool, focusing on architectural demonstration.