Mohit-Nanda-Krishna/CustomerSupportAgent ? reverse-engineered prompt

Build me a Python customer support agent that can handle common customer questions in a smart, multi step way. I want it to use LangGraph and LangChain, keep track of the conversation so it remembers earlier messages, and pull relevant answers from a knowledge base with RAG before replying. It should feel like a real support assistant that can reason through a question, use the right context, and give a clear final answer instead of just guessing.

Please organize it cleanly so it is easy to extend later, with separate pieces for the agent logic, memory, retrieval, and shared state. Add a simple app entry point so I can run it locally, and include a few tests for the main flow. If you need current setup details for the libraries, look them up online and use sensible defaults. I mainly want a solid working prototype that shows the full support workflow end to end.