Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

here-we-go-jev

One-page playground and test bench for Jev with TypeSafe, OpenRouter, LLM-baseline and mock backends behind one /v1/systemone contract.

ServiceprototypeMIT
View on GitHub ↗

Overview

Here we go Jev is a one-page playground and test bench for TypeSafe's Jev. You pick a backend, load an example and run it to see structured answers with probabilities. Every backend speaks the same /v1/systemone contract, so the UI, scripts and official SDKs can switch between them by changing the base URL.

Backends include Jev via OpenRouter, Jev via TypeSafe, an LLM baseline that asks an ordinary model the same questions one call per question for side-by-side comparison, and a mock that returns seeded answers and can simulate 401, 422, 429 and 529 errors. The Jev backends proxy the model server-side so keys never reach the browser. Scripts include a smoke test and experiments that call real APIs at a fraction of a cent per run.

What the decision model decides

TypeQuestionAnswers
ChoiceExample typed questions loaded into the playground(inferred)—

How it works

  1. Choose a backend (Jev, LLM baseline or mock)
  2. Load or write a request
  3. Run and inspect answers
  4. Point SDKs at the same backend routes

Strengths

  • Mock backend works without keys and simulates errors
  • Side-by-side LLM baseline
  • Same contract across backends

Caveats

  • Playground rather than production tooling

Getting started

bun install, cp .env.example .env (set OPENROUTER_API_KEY for real Jev), bun run dev, then open http://127.0.0.1:4173.

At a glance

Type
Service
Status
prototype
Language
TypeScript
Runtime
Bun web app
Has code
yes
Has tests
yes
License
MIT

Use cases and integrations

How curated lists describe it

  • The smoke test and the experiments call real APIs and cost a fraction of a cent per run. The experiments need the server running.

    logicrw/awesome-jev-projects

Frequently asked questions

Can I use it without a key?

Yes, with the mock backend.

Can it compare Jev with an LLM?

Yes, the LLM baseline asks the same questions one call each.

Can SDKs use it?

Yes, set TYPESAFE_BASE_URL to a backend route.

Related projects

  • jev-test

    Small harness calling Jev via OpenRouter's Decisions API, with 17 doc-derived cases and a latency and cost benchmark against two LLMs.

    BenchmarkprototypePython
  • typesafe.pro

    Server behind api.typesafe.pro, a free anonymous System One-compatible endpoint that forwards to TypeSafe on the operator's key.

    Serviceproduction
  • jev-builder

    Browser form for building Jev requests from pasted text without hand-writing JSON, then rerunning to see how steady the answers are.

    ProjectbetaJavaScript
  • jev-repl

    Terminal REPL for shaping System One requests before writing code, with lessons, a sketch editor, conversation turns and offline simulation.

    ServicebetaTypeScript
  • jev.nu

    Nushell module for TypeSafe's System One API: pipe content in, ask noul, choice and score questions, and filter on the numbers.

    ServicebetaNushell
  • jev-ai (alexsasha2750)

    Static browser interface for making structured Jev decisions through the Pollinations API, with multilingual READMEs.

    ServiceprototypeJavaScript