What it is
A live Three.js drone simulation with a Python/FastAPI backend where TypeSafe's jev-latest chooses each navigation action. It is for developers exploring typed AI decision-making in a simulated flight environment.
How it uses Jev
Jev receives structured state of nine candidate movements and answers a Choice question to select the next action, plus an independent Noul question estimating if the direct route is blocked. Python validates the typed response and rechecks the chosen movement; if blocked or confidence below 0.12, it hovers.
Primitives:choicenoul
Technique worth stealing
Geometric candidate generation with swept collision clearance and typed model validation.
Try it
Run uv sync, npm install, set TYPESAFE_API_KEY, then npm run build and uv run uvicorn backend.app:app --host 127.0.0.1 --port 8000.
Evidence
Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.
- Jev-centricyes0.91
- Shows a System One patternyes0.74
- Handles uncertaintyno0.04
- Measuredno0.05
- Runnableyes0.95
- Worth recommendingunclear0.50
- Model replicano0.04
- Problem scopescore on a 0–2 scale0.01
- About Jevyes0.98
Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.