Review: newsletter draft quality + CTA

The job

Jev judge eval (1 boolean + 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.

// Review: newsletter draft quality + CTA — Jev judge eval (typesafe-ai/jev).
// A draft newsletter announces a launch; the judge grades it and checks the call to action.
// 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: "Draft: Big news — Boards 2.0 is live. What is new: 10x faster exports, offline mode, 40 templates. Try it: one click migrates your boards in minutes. PS: reply and tell us what to build next — we read everything.",
  questions: {
    "draft_quality": {
      "type": "score",
      "instructions": "Score this newsletter draft. Higher is better.",
      "criteria": [
        "Buried lede, no links, no reason to care.",
        "Announces the news but no clear action or benefit.",
        "Clear news plus one action, slightly flat voice.",
        "Headline news, concrete benefits, one crisp CTA, human voice."
      ]
    },
    "cta_present": {
      "type": "boolean",
      "instructions": "True only if the draft contains an explicit call to action telling the reader what to do next."
    }
  },
});

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