What is the Here we go Jev playground?

A one-page playground and test bench for Jev, TypeSafe's System One model. Jev evaluates a state against typed questions and returns structured answers with calibrated probabilities instead of text; the page puts the runs, the state and questions, an inspector and the request console together.

How do I start the playground?

bun install, copy .env.example to .env, then bun run dev and open http://127.0.0.1:4173. Set OPENROUTER_API_KEY in the env file to call the real Jev; the Mock backend works without any key, so you can try the UI first.

What is the LLM baseline backend?

It asks an ordinary model the same questions, one call per question, through OPENROUTER_BASELINE_MODEL, so you can compare Jev's structured answers side by side with an LLM's. It needs the OpenRouter key and is meant for exactly that comparison, not for production use.

Do API keys ever reach the browser?

No — the two Jev backends proxy the real model server-side, so the browser never holds the keys. The TypeSafe backend uses TYPESAFE_API_KEY instead of an OpenRouter one, and swapping backends leaves the rest of the UI untouched.

What can the mock backend simulate?

Seeded answers plus the failure modes you want to test against: 401, 422, 429 and 529. It is how the UI, the scripts and the official SDKs can be exercised without spending anything or needing a key.

What do the experiments cover?

bun run experiments runs E1 and E3 through E6, and bun run e2 is option-order sensitivity against real Jev; you can pass a subset like e1 e5. The experiments need the server running and call real APIs for a fraction of a cent per run.

What is the lab notebook?

docs/lab-notebook.md is a record of the sources, vendor claims and every experiment run locally, with results. DESIGN.md documents the visual system the playground implements, and AGENTS.md the conventions for contributors and coding agents.

Jev directory

CommunityJev Demos & Experiments

Here we go Jev

A one-page playground and test bench for TypeSafe Jev: pick a backend, load an example and run typed questions against Jev through OpenRouter or TypeSafe, an LLM baseline, or a keyless mock — all speaking the same /v1/systemone contract, with an inspector for the answer distributions.

See highlights Open github.com

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

creativoma/here-we-go-jev

User repository on GitHub

View on GitHub

Local playground and test bench for TypeSafe's Jev System One model: typed questions, calibrated-probability answers, and side-by-side comparison with an LLM baseline.

Forks: 0
Last pushed: Updated Sep 2026

Repo stats from the GitHub API, cached Oct 2026.

Tagscommunity catalog-import logicrw playground test-bench openrouter

Works forJev for Evaluation Jev for Classification

Used inJev for Software & DevTools

Languages: TypeScript
Platforms: Browser OpenRouter
Capabilities: Open sourceRuns locallyRequires API key

Highlights

Four backends speak one /v1/systemone contract: Jev via OpenRouter, Jev via TypeSafe, an LLM baseline, and a keyless mock that simulates 401, 422, 429 and 529.
The two Jev backends proxy the real model server-side so keys never reach the browser; point an SDK at any backend by setting TYPESAFE_BASE_URL to it.
Quick start: bun install, copy .env.example to .env, bun run dev, then open http://127.0.0.1:4173 — the mock backend works without any key.
Scripts cover smoke tests against a live backend, the check gate, experiments E1 and E3–E6, and an option-order sensitivity run against real Jev.
docs/lab-notebook.md records the sources, vendor claims and every experiment run locally, with results.

Quickstart

bash

bun install
cp .env.example .env
bun run dev

Watch out

Grounded on the full repo README and GitHub metadata read on 2026-10-15. The README publishes no accuracy or quality numbers for the playground itself; the smoke test and the experiments call real APIs and cost a fraction of a cent per run, and the experiments need the local server running.

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.

Per-Tick Decisions in Games

Sep 28, 2026

A game tick is a state plus a legal action list. How Clash Royale, Minecraft, and the earlier Mario, Pokémon, StarCraft, and drone loops ask Jev one step at a time.

Jev Demos & Experiments#gaming#agents#control

Read article

From the Jev AI model community

Posts from builders shipping with Jev right now.

Follow @typesafeai