Grade: changelog headline strength

The job

Jev judge eval (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.

// Grade: changelog headline strength — Jev judge eval (typesafe-ai/jev).
// An agent writes a changelog headline; the judge scores it.
// 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: "Headline: Various improvements and bug fixes.\nBody: Export to CSV is 10x faster, SSO is now self-serve, and the mobile crash on iOS 17 is fixed.",
  questions: {
    "headline": {
      "type": "score",
      "instructions": "Score this changelog headline against its body. Higher is better.",
      "criteria": [
        "Generic filler that says nothing (e.g. various improvements).",
        "Hints at content but buries the lead.",
        "Names the biggest change but undersells the rest.",
        "Specific, benefit-led, and earns the click."
      ]
    }
  },
});

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