Official

system-one-adapter-python

@typesafe-ai177PythonMITupdated 2026-09-18runnable

Drop-in TypeSafeClient replacement backed by LLM APIs

typesafe-ai/system-one-adapter-python

Where it calls Jev

from typesafe_sdk import (

src/system_one_adapter/__init__.py:3

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/test_client_with_fake_model.py:25

  1. positivetrue/false

    The review is positive.

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: {
    positive: { type: 'boolean', instructions: 'The review is positive.' },
  },
});

from tests/test_client_with_live_apis.py:42

  1. positivetrue/false

    The book review is positive.

  2. ratingscore

    How favorable the reviewer's overall assessment is.

    • 0The reviewer condemns the book and urges readers to avoid it.
    • 1The reviewer is mostly critical and does not recommend the book.
    • 2The reviewer expresses mixed or neutral feelings about the book.
    • 3The reviewer praises the book overall while noting meaningful flaws.
    • 4The reviewer offers unreserved praise and an emphatic recommendation.
  3. genrechoice

    Which genre this review is about.

    • fictionA novel or short story.
    • nonfictionA book based on facts, real events, or ideas.

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: {
    positive: { type: 'boolean', instructions: 'The book review is positive.' },
    rating: { type: 'score', instructions: 'How favorable the reviewer\'s overall assessment is.',
      criteria: ['The reviewer condemns the book and urges readers to avoid it.', 'The reviewer is mostly critical and does not recommend the book.', 'The reviewer expresses mixed or neutral feelings about the book.', 'The reviewer praises the book overall while noting meaningful flaws.', 'The reviewer offers unreserved praise and an emphatic recommendation.'] },
    genre: { type: 'choice', instructions: 'Which genre this review is about.',
      criteria: { fiction: 'A novel or short story.', nonfiction: 'A book based on facts, real events, or ideas.' } },
  },
});

from tests/test_client_with_live_apis.py:69

  1. instruction_probechoice

    Return the only marker whose state is ACTIVE.

    • marker_fenThe marker_fen catalog entry.
    • marker_torThe marker_tor catalog entry.
  2. criteria_probechoice

    Return the correct opaque handling route for the parcel.

    • route_7qUse when the handling class is CLASS_CRYSTAL.
    • route_2mUse when the handling class is CLASS_STEEL.

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: {
    instruction_probe: { type: 'choice', instructions: 'Return the only marker whose state is ACTIVE.',
      criteria: { marker_fen: 'The marker_fen catalog entry.', marker_tor: 'The marker_tor catalog entry.' } },
    criteria_probe: { type: 'choice', instructions: 'Return the correct opaque handling route for the parcel.',
      criteria: { route_7q: 'Use when the handling class is CLASS_CRYSTAL.', route_2m: 'Use when the handling class is CLASS_STEEL.' } },
  },
});

Other projects in this category