kriptokasim/Consultaion — reverse-engineered prompt
Reverse engineered prompt
Build me a web app called Consultaion, spelled exactly like that. It should feel like an AI parliament where a user asks a question, several AI agents debate it, judges vote, and the app produces the best final answer.
Users should be able to sign up, sign in, start a debate, choose the model or use the recommended one, set things like agents, judges, and budget, then watch the debate unfold live. Afterward they should see their run history, a Hansard style transcript, voting results, analytics, and be able to download a report.
Add team sharing, private per user runs, usage limits, an admin area with audit logs, provider health, and a public leaderboard based on pairwise voting confidence. Use a warm Amber Mocha cockpit style across the landing page, dashboard, and live debate screens.
Please make it production ready with a backend, web frontend, Postgres persistence, migrations, Docker local setup, mock debate mode for testing, and real LLM provider keys handled only on the server.
Want more depth? Deep Reverse