Agent tooling

typesafe-mod

@BeLazy167★ 2TypeScriptMITupdated 2026-09-17▶ runnable

Claude Code mod that routes decisions to TypeSafe's Jev model: ranks installed skills per prompt, and answers the agent's own this-or-that questions when confident.

BeLazy167/typesafe-mod ↗

Where it calls Jev

endpoint: 'https://api.typesafe.ai/v1/systemone',

hooks/suggest.ts:27

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 hooks/suggest.ts:188

  1. needs_skilltrue/false

    Does answering `state.user_request` call for following a written procedure,

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: {
    needs_skill: { type: 'boolean', instructions: 'Does answering `state.user_request` call for following a written procedure,' },
  },
});

Other projects in this category