Agent tooling

claude-jev

@buchmark★ 3TypeScriptMITupdated 2026-09-18▶ runnable

Claude Code plugin that scores review findings, debug hypotheses and design options with TypeSafe's Jev — calibrated probabilities instead of one more opinion.

buchmark/claude-jev ↗

Where it calls Jev

import { TypeSafeClient } from '@typesafe-ai/sdk';

src/infrastructure/typesafe-jev.ts:1

The link points at the commit we read, so the line number still holds.

What it asks Jev

These question sets are lifted from this project's source exactly as written.

from tests/typesafe-jev.test.ts:37

  1. is_urgenttrue/false

    Does this convey urgency?

  2. frustrationscore

    How frustrated is the customer?

    • 0 — Calm
    • 1 — Frustrated
    • 2 — Very angry

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    is_urgent: { type: 'boolean', instructions: 'Does this convey urgency?' },
    frustration: { type: 'score', instructions: 'How frustrated is the customer?',
      criteria: ['Calm', 'Frustrated', 'Very angry'] },
  },
});

from tests/typesafe-jev.test.ts:135

  1. plaintrue/false

    Is it fine?

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    plain: { type: 'boolean', instructions: 'Is it fine?' },
  },
});

from tests/use-cases.test.ts:23

  1. qtrue/false

    Is it fine?

Run it on your own text

The code
import { experimental_evaluate as evaluate } from 'ai';

const { answers } = await evaluate({
  model: 'typesafe-ai/jev',
  state,
  questions: {
    q: { type: 'boolean', instructions: 'Is it fine?' },
  },
});

Other projects in this category