What it does
The included Pokémon archive can be browsed without model calls; functional Python nodes currently require the supported macOS sandbox.
Agent tooling
Ask a coding model to create, test, and optionally evolve a Jev decision harness for a particular task.
Only you can see your notes.
Screenshot unavailable. Open the experiment ↗
The included Pokémon archive can be browsed without model calls; functional Python nodes currently require the supported macOS sandbox.
For a game-playing assistant, start with the saved Pokémon example. Ask your developer to study how one AI writes the playing plan as code. Once that plan is fixed, Jev, a second AI, only needs to choose moves as the code runs. You could borrow this approach instead of asking an AI to write a new plan on every turn.