Audit: shift handoff note completeness

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.

// Audit: shift handoff note completeness — Jev judge eval (typesafe-ai/jev).
// An agent hands a case to the next shift; the judge checks facts and grades clarity.
// 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: "Handoff: Case C-209 (Acme, $40k ARR): outage workaround holding since 14:00; customer expects a permanent fix ETA by Friday 12:00. Next step: Bella to confirm the DB migration window tonight. Risk: they mentioned evaluating a competitor if this slips again.",
  questions: {
    "next_step_present": {
      "type": "boolean",
      "instructions": "True only if the handoff names a concrete next step with an owner."
    },
    "handoff_clarity": {
      "type": "score",
      "instructions": "Score the clarity of this handoff note. Higher is better.",
      "criteria": [
        "Unusable — missing context and next actions.",
        "Readable but a key fact (deadline, owner, or risk) is missing.",
        "Complete facts, slightly disorganized.",
        "Context, status, deadline, owner, and risk in one scan."
      ]
    }
  },
});

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