Audit: release notes audience + quality

The job

Jev judge eval (1 choice + 1 score) over a fixed state : runnable via experimental_evaluate.

The prompt

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

// Audit: release notes audience + quality — Jev judge eval (typesafe-ai/jev).
// Draft release notes for a mobile update; the judge classifies and grades them.
// 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: "Notes: v3.8 — Dark mode is here (Settings → Appearance). Fixed the crash when exporting large boards. Note: iOS 15 is no longer supported; you need iOS 16+.",
  questions: {
    "audience": {
      "type": "choice",
      "instructions": "Who are these release notes written for?",
      "criteria": {
        "end_users": "Plain-language notes a non-technical user can act on.",
        "developers": "API, SDK, or integration changes for engineers.",
        "internal": "Jargon-heavy notes only the team would understand."
      }
    },
    "notes_quality": {
      "type": "score",
      "instructions": "Score these release notes. Higher is better.",
      "criteria": [
        "Missing what changed or who is affected.",
        "Lists changes but no action or requirement stated.",
        "Clear changes with requirements, slightly terse.",
        "Clear changes, exact navigation, and upgrade requirements."
      ]
    }
  },
});

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