joseaburt/zep — reverse-engineered prompt

Build me a simple working example repo for Zep that shows how an AI assistant can remember useful context over time and use it in later conversations.

I want it to be easy to run locally with Python and uv. It should load a Zep Cloud API key from an env file, add sample chat messages, a few bits of business data, and some app events, then ask an agent a question that proves it can pull the right context back. Please include one clean example for an AutoGen style agent integration, plus clear sample scripts that someone can run from the terminal.

Keep the code organized like a real examples and integrations repo, with short comments and a README that explains setup, what each example does, and how Zep is adding relationship aware context. Don’t use the old community edition path except as a legacy note. Look up current Zep docs online if you need to.