PromptLeaders

Competitive AI prompting, judged and ranked.

A daily set of prompting games with a scoreboard attached. Talk your way past a guard built to resist you, recreate a target image using words alone, or hit a task inside hard constraints. A judge panel of models scores each round, an ELO rating tracks whether you're actually improving, and everyone plays the same games on the same day.

What it looks like

The problem

Everyone claims to be good at prompting and nobody can prove it. Scoring it against a fixed target, on the same day as everyone else, turns a vibe into a rating.

Key features

• The Vault: talk your way past a guard built to resist you
• Image Match: recreate the target with words alone, scored by a judge panel
• Constraint Run: hit the task inside hard constraints, scored deterministically
• A daily play window, an ELO rating and a public leaderboard
• One-tap email sign-in, and the model access comes with it

Built with — how & why

• 
  TypeScript
• 
  FastAPI
• 
  PostgreSQL
• 
  Docker
• 
  Redis
• 
  OpenAI