Multi-Agent LLM tutoring system using RAG to retrieve top-N ranked context from PDFs, web, Wikipedia, and scholarly sources. Domain-specialized tutor agents generate responses, debate via an agentic protocol, and a judge agent selects the most accurate, context-grounded answer.