Code to the Moon introduces Jev through planet and atmosphere questions in its playground. He shows how context changes a decision and explains that the Mario example receives game state rather than screen images.
Original by Code to the MoonGetting startedBeginner7 min 25 secPublished
Choose the answer shape first: Choice selects from named options, Noul gives a yes-or-no probability, and Score places an answer on an ordered scale.
Supply context in state. A hypothetical move to Mars changes the planet answer, and different question types in one request share the same state.
Check what the model actually receives. The Mario example uses structured game information from memory, not the pixels shown to the viewer.
Worth knowing
Playground examples and creator commentary, not a test of general accuracy or game-playing ability. The planet story is hypothetical and the atmosphere scores are model outputs, not scientific measurements. The creator says he has not tried the two mentioned Rust clients. His latency, training and generalization claims were not independently checked; no client or game was executed for these notes.