Factual answer: claims carry checkable sources

The job

Jev judge eval (1 boolean) over a fixed state : runnable via experimental_evaluate.

The prompt

Copy this into Muse and replace anything in [square brackets] with your own details.

// Factual answer: claims carry checkable sources — Jev judge eval (typesafe-ai/jev).
// An agent answers a warranty question; the judge checks every factual claim is backed by a cited source.
// Model: typesafe-ai/jev via the Vercel AI Gateway ($0.042/1M input tokens).
// Auth: set AI_GATEWAY_API_KEY (vck_…) in the environment — never paste the key here.
// For zero-data-retention / no-training workloads, add
// providerOptions: { gateway: { … } } to the evaluate call below.

import { experimental_evaluate } from 'ai';

const result = await experimental_evaluate({
  model: 'typesafe-ai/jev',
  state: "Customer: Is my battery covered under warranty?\nAgent: Yes — batteries are covered for 2 years from purchase (Warranty Policy §4.2, https://example.com/warranty#4.2). Your order #88121 shipped 14 months ago, so you are inside the window. Start a claim at https://example.com/claims.",
  questions: {
    "claims_sourced": {
      "type": "boolean",
      "instructions": "True only if each factual claim (coverage period, order age, claim path) is backed by a cited source or link."
    }
  },
});

console.log(JSON.stringify(result, null, 2));