agentuniverse-ai/agentUniverse — reverse-engineered prompt

Build me a Python project called agentUniverse that makes it easy to create and run both single agents and teams of agents powered by large language models. I want it to feel like a practical framework, not just a demo, with a clean starter app, sensible configuration, and examples I can run quickly to see results.

The main value should be multi agent collaboration for complex tasks. Please include ready to use collaboration patterns like one where agents plan, execute, review, and present, and another for data heavy work that mixes facts, expert opinion, and final writing. It should be easy to switch between popular model providers just by changing config and API keys.

I also want the framework to support real world agent building features like tools, knowledge base or RAG style retrieval, testing, optimization, and basic evaluation. Add good docs and sample apps so a beginner can follow along from install to first working agent. If anything is unclear, check the current docs online and fill in the missing pieces sensibly.