jev-drone

Where it calls Jev

from typesafe_sdk import TypeSafeClient, Choice, Noul, Score

tactics.py:12

The link points at the commit we read, so the line number still holds.

What it asks Jev

These question sets are lifted from this project's source exactly as written.

import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    risk: { type: 'score', instructions: 'How dangerous is the drone\'s immediate situation?',
      criteria: ['clear and open', 'tight but manageable', 'about to hit something'] },
  },
});

import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    risk: { type: 'score', instructions: 'How close is this aircraft to hitting something?',
      criteria: ['open tunnel', 'must commit to a move now', 'impact unavoidable'] },
  },
});

Other projects in this category

• typesafe-mario ▶ — A TypeSafe/Jev agent that plays Super Mario Bros. from structured emulator state.
• typesafe-playground ▶ — Interactive experiments with TypeSafe Jev, from support routing to 3D driving simulations with real AI decisions and visible sensor inputs.
• live-jev ▶ — 2D autonomous car simulation in the browser, driven by TypeSafe's Jev decision model
• jev-autopilot ▶ — This demo uses Jev from TypeSafe AI to autonomously fly a drone in a random city from point A to point B, avoiding obstacles along the way. A trip costs $0.01.
• jev_speedway ▶ — Uses jev to control race cars on a virtual track
• chakravyuha-jev ▶ — Chakravyuha — a polar ring-maze where every move is a Jev (TypeSafe System One) decision. A fun experiment: the model picks each move, the walk grades it green or red, and the history page asks whether its confidence score can be trusted. BYOK, no build step.
• last-train-jev ▶ — A small detective escape room built with TypeSafe Jev, React, and Express.
• quackd — One CLI for all your robots. Connect them, command them, and let them work together, each with an LLM for a brain, Jev for cheaper steps. Microduck, Open Duck Mini, LeRobot, XLeRobot, AlohaMini, ToddlerBot or any ROS base. Claude, OpenAI, Gemini, Grok, or local via Ollama or vLLM. Simulator, .duck safety contracts, MCP, memory between runs, flocks.