The Validation Dojo

Practice the one thing the AI can't do for you: deciding what to check before you trust the code. Every scenario is scored by the real Agent Karma engine — the same one in the extension.

• Actually fine white belt

  Sometimes the AI just... gets it right. Can you tell — without over-checking?

• The confident off-by-one white belt

  The AI wrote a paginator. It looks right. Did you check the edges?

• The API that doesn't exist green belt

  The AI called a tidy-looking method. Does it actually exist?

• The string-built query brown belt

  The AI wired up a DB lookup from user input. Anything jump out?

More scenarios coming. When you're ready for the real thing, install the extension.