Rules first
Numeric targets are selected and validated in code, so the model cannot invent an impossible mechanic.
Try the mission director prototype. The objective is set by game-safe rules; the copy can be generated by Claude API when a server key is configured. The result simulator lets you test mission completion without changing the existing game.
A real API request requires a running backend, an API key stored server-side, and a private preview token. This form never asks for your Anthropic API key.
Choose a chapter and press “Generate a mission” to explore the prototype.
Enter a sample game result to test the mission rules. These values are not pulled from the public game leaderboard.
Numeric targets are selected and validated in code, so the model cannot invent an impossible mechanic.
Claude creates a title and short narrative tailored to the chapter, difficulty and selected language.
Mission success is computed from the submitted run data. Production integration will validate player events on the game server.
Daily challenges are the first experiment in a larger roadmap for AI-enhanced storytelling, localization and mobile releases.