adityajha2005/agents — reverse-engineered prompt

Build me a Python framework for realtime voice AI agents that work with LiveKit rooms and feel ready for real use, not just a toy demo. I want to be able to create an agent with simple instructions, plug in speech to text, a language model, and text to speech, then have it join a session, talk naturally, use basic tools, and know when the user is done speaking. Please include a clean server entrypoint, session management, examples for a simple voice assistant and a multi agent handoff flow, plus support for exchanging data with clients and the option to connect to phone calls through LiveKit. It should be easy to swap model providers, and there should be a small plugin setup for common providers. Add a built in testing story so agent behavior can be checked automatically. Keep it open source friendly and runnable on my own server. If anything is unclear, look up the current LiveKit agent docs online.