jev-playground
Runnable Jev cookbooks, a plain-English guide and sixteen interactive simulations of how a typed decision model changes architecture.
Overview
jev-playground by exponen-agi is a working engineer's playground for Jev. It combines runnable Python cookbooks, a plain-English guide including a section on where Jev breaks, and sixteen interactive browser simulations designed so non-technical colleagues can follow the flow. The simulations need no install or API key and call nothing; each links to the Python it is built from.
Simulations cover customer conversations (a support triage cascade where most tickets never reach an expensive model, shared-inbox sorting, checking a chatbot reply before sending), sales and orders (lead scoring inside a form submit, WhatsApp order intake where Jev judges and a regex reads the number), back-office work (invoice matching against open orders, bookkeeping where calibration decides which rows a person reads), and screening and routing around existing OpenAI, Claude or Gemini systems, such as gating documents before RAG. Users can drag thresholds and watch routing change.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which desk should this support ticket go to?(inferred) | — |
| Yes / no | Is this chatbot draft safe to send?(inferred) | yes · no |
| Yes / no | Should this document enter the RAG index?(inferred) | yes · no |
How it works
- Open a live simulation in the browser
- Drag the confidence floor or approval limit
- Watch items route to code, models or humans
- Open the linked cookbook to run the real Python
Strengths
- No key or install needed to understand the patterns
- Covers many business processes
- Explicit section on failure cases
Caveats
- Simulations do not call the model
- Cookbooks need a TypeSafe key to run live
Getting started
Open https://exponen-agi.github.io/jev-playground/ or see the Setup section to run the cookbooks.At a glance
- Type
- App
- Status
- beta
- Language
- Python
- Runtime
- Cookbooks plus static web simulations
- Has code
- yes
Use cases and integrations
How curated lists describe it
A simulated playground along with actual code samples scenario-based examples.
hellogumbo/awesome-jev
Frequently asked questions
Do the simulations call Jev?
No. They are interactive illustrations that link to the real Python cookbooks.
Who is it for?
Engineers and non-technical colleagues who want to see where a decision model fits.
Does it cover limitations?
Yes. The guide has a 'Where it breaks' section.
Related projects
Python sandbox turning Jev issue-category and reproduction judgments into testable routing proposals, with offline replay.
ProjectprototypePythonAgent skill and CLI that chains Jev question design: find decision calls, check fit, draft, lint, calibrate thresholds and plan a cascade.
Agent skillprototypePythonTool that lets code and agents ask Jev typed questions via reusable jevels, mapping answers to act, mark or fall_back.
ProjectprototypeTypeScriptWorked Jev examples for ticket triage and reranking, runnable through OpenRouter without an early-access key, with offline tests.
ProjectprototypePythonRust crate and jud CLI where a decision is a .jud rubric file answered by a System One model, type-checked and testable without a key.
PluginbetaRustResearch notes on Jev plus a live Cloudflare Worker playground that triages support messages by department, urgency and frustration.
AppprototypeTypeScript