User repository on GitHub
Demos with Jev, TypeSafe's System One model, one folder per demo. Code behind the videos on The Agentic Enterprise.
- Forks
- 4
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Demos & Experiments
Jev Demos — mani-aiml
Five self-contained Jev demos, one folder each: Jev fetching for Claude, Jev and Laya grading 500 AI answers, a prompt-injection review gate, and two fine-tuned decider comparisons against Jev.
- Stars
- 7
- Language
- Python
- License
- MIT
- Published by
- Community
- Author
- mani-aiml
- Last verified
- Oct 2026
- Languages
- Python
- Capabilities
- Open sourceRequires API key
Highlights
- One folder per demo, each self-contained with its own README, requirements, and key names.
- jev-fetches-claude-writes runs eight lookup tasks twice, with Claude alone and with Jev fetching while Claude writes.
- laya-vs-jev-judge has Jev and Laya grading the same 500 AI-written answers zero-shot, including cascade maths.
- decision-model-prompt-injection tests a Jev and Claude Haiku 4.5 review gate against twelve hand-written emails and LLMail-Inject attacks.
- laya-finetune and strands-decider rescore the same 500 items; checkpoints are not included and shipped data is synthetic.
Watch out
MIT for everything, but these are demos, not production pieces: several folders ship code only, without checkpoints or data. The prompt-injection folder downloads real attack emails at run time, and you supply the keys each folder's README names.
Questions about this entry
Short answers drawn from the linked project or page.
Five self-contained demo folders with one README each: jev-fetches-claude-writes, laya-vs-jev-judge, decision-model-prompt-injection, laya-finetune, and strands-decider, plus the data and notebooks each demo needs.
jev-fetches-claude-writes: it runs eight tasks with Claude alone and with Jev fetching the lookups while Claude writes, and the two agents, tools, data, offline tests, and notebook are included.
All shipped data is synthetic, except decision-model-prompt-injection, which holds no data and downloads public LLMail-Inject attack emails at run time; strands-decider reads laya-finetune/data/ instead of carrying its own.
cd into its folder, create a venv, pip install -r requirements.txt, copy .env.example to .env, and add the keys that folder's README names; the README shows this recipe once for all folders.
The demos are the code behind videos on The Agentic Enterprise on YouTube, and each folder keeps its own architecture notes in its README rather than one shared document.
No: laya-finetune says the fine-tuned checkpoint is not included, and strands-decider ships code only and reads laya-finetune/data/, with no checkpoints at all.
MIT covers everything in the repository, and the one third-party dataset, Microsoft's LLMail-Inject attack corpus fetched by the prompt-injection demo at run time, is also MIT.
Other listings of this project
Forks and copies of this project in the directory. This entry is the canonical one.
0GitHub stars
A two-package collection of TypeSafe System One demos: jev_common is a shim for the API that loads .env and records results, and maze_lookahead checks spatial reasoning by testing how far the model can think ahead in a maze.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
Bud-ro/jev-demosDart
Bud-ro/jev-demosDartMore Jev resources like this
97GitHub stars
A Python experiment that turns Jev into a chat model by repeatedly asking which symbol comes next, including alphabets, sampling modes, and a bench command.
Jev Demos & Experiments#community#python#demos
Community
kyle-pena-nlp/jevchatPython
kyle-pena-nlp/jevchatPython7GitHub stars
A benchmark harness and web UI that cross-compares dozens of zero-shot information-extraction and classification systems — extractor encoders, rerankers, and typed-decision engines like Jev — on accuracy, latency, and cost.
Jev Benchmarks & Evaluations#community#python#benchmark
Community
umstek/zero-shot-ie-benchPython
umstek/zero-shot-ie-benchPython1GitHub stars
A dated harness for running open System One typed-decision models — Laya, Decider, imajev, Winnow-12B, decider-12b — on a Snapdragon X Elite NPU, GPU, and CPU, benchmarked against TypeSafe's Jev API.
Jev Benchmarks & Evaluations#community#python#snapdragon
A localization pre-review gate that pairs source and target strings and judges meaning, terminology, tone, and placeholders as separate questions, so one failing dimension blocks a release without blending scores.
Jev Tools & CLIs#community#python#localization
Community
29.7kGitHub stars
The Cua computer-use platform documents a jev-use recipe where TypeSafe Jev picks bounded action IDs and Cua Driver executes clicks and fills in the browser.
Jev Demos & Experiments#community#browser#computer-use
Community
trycua/cuaRust
trycua/cuaRust103GitHub stars
A Playwright-based CodexQA skill where Jev chooses indexed controls on a page-built index while the browser executes the action.
Jev Demos & Experiments#community#browser#playwright
Community
openqa-cn/jev-browserTypeScript
openqa-cn/jev-browserTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
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 community
Posts from builders shipping with Jev right now.