RoshaanK20/GODMODE3 — reverse-engineered prompt
Reverse engineered prompt
Build me a privacy focused AI chat web app called G0DM0D3. I want it to run mostly as a static browser app, where users bring their own OpenRouter API key, save it only in their browser, and can chat with lots of different AI models from one clean interface.
The main experience should feel like a cyber terminal, with themes like Matrix, Hacker, Glyph, and Minimal. Add a settings panel for API key, privacy choices, model mode, data import and export, and chat history. Users should be able to pick normal chat, a classic mode that sends the same prompt to several model and prompt combos, and a bigger comparison mode that asks many models, scores the answers, and shows the best one.
Include optional research features for red teaming, like input perturbation, response cleanup modules, and automatic sampling settings. Make telemetry opt out, dataset publishing opt in with a clear warning, and keep chat history in local storage. Make it responsive and easy to self host.
Want more depth? Deep Reverse