A small desktop console and CLI for Jev, TypeSafe's System One model. It lets a human use the API by hand: paste a state, add a few typed questions, press one button, and read the probabilities and confidences without writing code for every question.

Which question types does it support?

All three Jev primitives: noul (is this true?) returns a probability between 0 and 1; choice (pick one of N options) returns the chosen option with the full distribution and confidence; score (where on this scale?) returns a score that may land between levels.

How do I install and run it?

Python 3.10+ with tkinter is all you need — no third-party packages, since it uses only the standard library's urllib. Set TYPESAFE_API_KEY as an environment variable or in a .env file, then run python jev_gui.py for the window or python jev_cli.py --help for the CLI.

No. Jev cannot write prose, so neither does the console; it only sends typed questions and shows the numbers that come back. The README also says not to use it for arithmetic, counting or date ordering, because TypeSafe's own guidance keeps those in code.

Where does it look for the API key?

In order: TYPESAFE_ENV_FILE, then ./.env next to the scripts, then ./.env in the current working directory, then ~/.hermes/.env. Create a key at console.typesafe.ai/keys first. You can sanity-check which models your key can call with python jev_cli.py --models.

What does the CLI accept?

Repeatable --noul, --choice and --score flags, --state or --state-file for the material, --questions for a JSON file, plus --model, --json, --dry-run, --models and --timeout. Both the GUI and CLI share one request path in jev_core.py.

Any tips from the author on using Jev by hand?

Questions run in parallel over the same state, so adding questions costs input tokens but almost no extra latency — ask everything you might need in one call. Accuracy falls as state fills with irrelevant detail, one narrow judgment per question works best, and Jev reads literally.

Jev directory

CommunityJev Tools & CLIs

Jev Console

A small desktop console and CLI for Jev: paste a state, add noul, choice and score questions by hand, and read calibrated probabilities back without writing code. Standard library only, with a tkinter window whose labels are in Chinese.

See highlights Open github.com

Stars: 1
Language: Python
License: MIT
Reading time: 3 min
Published by: Community
Author: wenchenxi
Last verified: Oct 2026

wenchenxi/jev-console

User repository on GitHub

View on GitHub

Local console + CLI for TypeSafe's Jev (System One) model: send a state and typed questions, get calibrated probabilities back.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity catalog-import logicrw gui cli

Works forJev for Classification Jev for Scoring

Used inJev for Software & DevTools

Languages: Python
Platforms: CLI
Capabilities: Open sourceRuns locallyRequires API key

Highlights

Noul returns a probability between 0 and 1; Choice returns the chosen option with the full distribution and confidence; Score may land between levels.
Standard library only (urllib) — typesafe-sdk is not needed; one POST evaluates all questions in parallel with exponential-backoff retry on 429 and 529.
The window runs from python jev_gui.py; the CLI takes repeatable --noul, --choice and --score plus --state, --state-file, --questions, --json and --dry-run.
Key lookup order: TYPESAFE_ENV_FILE, then ./.env next to the scripts, then ./.env in the working directory, then ~/.hermes/.env.
Not a chat client, not for arithmetic, counting or date ordering, and not affiliated with TypeSafe AI — it calls only the documented endpoint.

Quickstart

bash

python jev_gui.py
python jev_cli.py --state-file s.txt \
    --choice "dept=Which team should handle this|billing:payments;technical:bugs;sales:pricing" \
    --score  "frustration=How frustrated is the customer|calm;annoyed but civil;angry"

Watch out

Copy reviewed against the repo README on 2026-10-15; the UI is in Chinese because it was written for personal use, and no accuracy or performance benchmarks are published. The usage notes — filter state before sending, one narrow judgment per question, write the exact condition — are the author's own observations, not measured results.

Questions about this entry

Short answers drawn from the linked project or page.

From the Jev guides

Original write-ups on what this entry is for.

Jev at home: voice pipelines, Assist routing, and sentinels

Oct 16, 2026

How Home Assistant conversation agents, macOS voice pipelines, and an advisory sentinel put a fast typed decision in the loop — one call per utterance, thresholds in code, and a daily token budget.

Jev Tools & CLIs#home-assistant#voice#automation

Read article

Search visibility, audited by decisions

Oct 16, 2026

Jev-based SEO tools split the crawl from the call: deterministic crawlers count and parse, while typed judgments rank fixes, place internal links, score search results, and track AI-answer mentions — with costs and caveats attached.

Jev Tools & CLIs#seo#search-visibility#audits

Read article

Jev on the Gateways: Vercel, Workers AI, and OpenRouter

Oct 12, 2026

The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.

Jev Tools & CLIs#gateway#vercel#cloudflare

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai

Jev Console

Highlights

Quickstart

Questions about this entry

What is Jev Console?

Which question types does it support?

How do I install and run it?

Is it a chat client?

Where does it look for the API key?

What does the CLI accept?

Any tips from the author on using Jev by hand?

More Jev resources like this

From the Jev guides

From the Jev AI model community