Agent tooling

super-jev

@Kevthetech143★ 5PythonMITupdated 2026-09-20▶ runnable

A small, extensible decision-to-action harness for TypeSafe Jev

Kevthetech143/super-jev ↗

Where it calls Jev

return this.transport('https://api.typesafe.ai/v1/systemone', {

src/jev.ts:208

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 test/jev-pin.test.ts:26

  1. q1true/false

    is it true?

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: {
    q1: { type: 'boolean', instructions: 'is it true?' },
  },
});

from test/loop.test.ts:72

  1. oktrue/false

    Is this a greeting?

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: {
    ok: { type: 'boolean', instructions: 'Is this a greeting?' },
  },
});

Other projects in this category