advocaat

Where it calls Jev

/** Env: `TYPESAFE_BASE_URL`. Default: `https://api.typesafe.ai` (gateway: `https://ai-gateway.vercel.sh/v4/ai`) */

src/api.ts:162

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: {
    urgency: { type: 'score', instructions: 'Urgency?',
      criteria: ['low', 'high'] },
  },
});

Other projects in this category

• Jev-cu ▶ — 把 Computer Use 的「下一步点哪里」交给 Jev（TypeSafe System One）：Jev 从界面文字候选中选元素、动作、完成度与风险，Codex Computer Use 负责读取界面与执行，本地策略门槛拦截敏感操作。只传文字，不传截图。
• ruby_decision_model ▶ — Ruby client for decision models such as Typesafe Jev
• jev-axi ▶ — Agent-ergonomic CLI for TypeSafe's Jev: fast calibrated judgments (pick, rate, check, rank, triage, guard) from the shell
• mario-jev ▶ — A uv-managed Python prototype that plays NES Super Mario Bros. (level 1-1 by default).
• jev-game-tools ▶ — Experiments in driving real-time games with TypeSafe Jev: a fast
• typesafe-jev-workflow ▶ — A small async LangGraph workflow that sends a mocked email to TypeSafe's Jev model,
• jev_jsonschema ▶ — Run a JSON Schema through TypeSafe's Jev API, and get JSON back.
• is-odd-jev ▶ — Check whether a number is odd, using a System One model, with a calibrated probability.