A scalable, Dockerized multi-agent system built with Python and LangChain, designed to intelligently route and process user queries through specialized agents.
This system demonstrates a three-agent architecture:
Router Agent: Classifies incoming messages and directs them to the appropriate downstream agent.
Knowledge Agent: Answers questions using a RAG (Retrieval-Augmented Generation) pipeline with web-sourced data.
Customer Support Agent: Handles user-specific inquiries with customizable tools.
✅ LLM-powered routing (LangChain) ✅ RAG pipeline for grounded responses ✅ Personality layer for human-like interactions ✅ FastAPI endpoint for seamless integration ✅ Dockerized for easy deployment
Language: Python
Frameworks: LangChain, FastAPI
Data: Vector DB (e.g., FAISS), Web Scraping (BeautifulSoup/Scrapy)
Infrastructure: Docker
bash
git clone https://github.com/Fugant1/AgentSwarm.git
cd agent_swarm
docker build -t agent_swarm .
docker run -p 8000:8000 agent_swarm
API Endpoint: POST /chat (payload: {"message": "Your query", "user_id": "123"})
agent_swarm/
├── agents/ # Router, Knowledge, Customer agents
├── api/ # FastAPI endpoint
├── data/ # RAG data storage
├── tests/ # Unit/ec2e
└── utils/ # RAG, personality layer
Modular Design: Easily extendable with new agents/tools.
Production-Ready: Dockerized with comprehensive tests.
LLM Best Practices: Implements RAG and structured workflows.