farukgnc/AI-Debate — reverse-engineered prompt

Reverse engineered prompt

Build me a simple command line app where I can make two AI models debate each other on any topic.

When I run it, it should walk me through picking two models, setting each one’s personality and which side they’re arguing for, then choosing the topic, number of rounds, and an optional word limit. I want the debate to feel natural, like normal paragraphs, but each turn should still clearly respond to the other side, add a new point, and end with a challenge or question. Please make it keep enough recent context so the models don’t just repeat themselves, and add some guardrails against generic restating.

It should support Gemini, Groq, and Kimi if I have those keys, and optionally let me connect an OpenAI key at runtime or from a .env file so I can use a Codex or OpenAI model too. Only validate the keys that are actually needed for the models I pick. At the end, let me save the full transcript. Look up current API docs online if needed.

Want more depth? Deep Reverse