Make it work.
Know why it works.
Your first mission: repair a game that gives away its answer. Describe the fix, try it, and prove it with a test.
Save the mystery.
The answer escaped.
A tiny space mystery should keep its answer hidden until a player chooses a place and locks the guess. This version reveals it immediately.
What should unlock the answer?
Clear rules. Better requests.
Specify observable behavior. These controls define the rules this training simulator can build. Your wording is yours.
Need a hint?
Think about three moments: before a choice, before a commitment, and after the result. “Make it better” does not tell a builder what to do at any of those moments.
Looks right? Prove it.
Your rules are applied. Play the preview, then run five checks against the same game logic. A failed check tells you what to change.
These checks cover this mission’s rules. Passing them is not a claim that a real app is ready for production.
You repaired the mystery.
You made a request precise, checked the result, and repaired behavior. Those are useful habits when working with an AI coding assistant.
What gave you evidence that the repair worked?
Describe → build → test → improve.
A next challenge to try yourself
Take your downloaded brief to an AI coding tool you already use. Ask it to implement the game and write checks. Compare its result with your requirements, including repeated clicks and an empty selection. This prototype does not connect to those tools.
Build something.
Build your judgment, too.
This is an early learning-game experiment from HumanAI. Future ideas include scores, timers, and original variations. Only Mission 001 is available in this prototype.