Launches on September 20, 2026

Benchmarks of TypeSafe Jev on Finnish text: reranking, code search without an index, intent classification, citation checking and lecture-transcript reranking.

LiteLLM pre-call hook that routes coding-agent requests to a cloud or local model lane, judged by TypeSafe jev

判断特化AI Jev (TypeSafe AI) に架空の営業電話ログ60本を判定させた検証コードとデータ

Live Sub-25ms Sales AI Teleprompter & Objection Anticipation Engine in Rust. Anticipates prospect subtext and serves instant Chris Voss & Oren Klaff counter-scripts with two-way mic testing.

Jev based Resume scanner

参考 TypeSafe 的 jev 实现一个支持 jev 的 skill

Spec-compliant universal agent skill: fast TypeSafe Jev semantic judgments (intent detection, code ranking, quality gates, safety checks) via a bundled stdlib-only CLI

Open-source Jev harness for TypeSafe, OpenJev, LocalJev, Ollama, vLLM, LM Studio, llama.cpp. Typed decisions, calibration, verification, abstention, CI gates.

Tiny inputs. Instant decisions. A small experimental playground for exploring fast, probabilistic decisions with Jev.

Auditing LLM deep-research reports with TypeSafe's Jev: source authority, relevance, and whether cited figures mean what the report says

Tiny Jev (typesafe) call stub — sample rows + Choice/noul questions

Jev(TypeSafe AI System One Model) 스터디 — 타입화된 결정·RLCD·confidence-gated routing을 한국어 노트와 TypeScript 목업으로 정리

Hands-on measurements of TypeSafe's Jev via OpenRouter: integration, Chinese-language behaviour, and game loops. Every figure in the reports is reproducible by the scripts here. Reports in Chinese. ~1,430 API calls, ~$0.15.

Autonomous dinosaur ecosystem where code runs the physics and TypeSafe's Jev decides each creature's intent from prose laws. Pixel-art web UI with a live Jev monitor (JSON in/out per tick), interchangeable brains (hybrid / full Jev / offline). Experimental Jev integration.

One graph, one policy, two judgment engines. This experiment tests TypeSafe’s published claims about Jev in a support-triage workflow, with a confidence gate that sends uncertain cases to a human.

🐠 A multiplayer fish game where every AI decision is a TypeSafe Jev (System One) call — flee, hunt, roam, with live confidence bars. Node + WebSocket + Three.js.

Support ticket router: Express + node:sqlite, mock and Jev decision engines

Autonomous settlement circuit breaker protecting ledgers against CVE/RCE balance bypasses, nocturnal draining, and prompt injection attacks using TypeSafe Jev.

TypeSafe’s Jev AI + an n-ary memory tree = efficient agentic memory routing and retrieval.

Reads a clinical note and nudges for the orders it commits to but the chart does not carry out. Built on TypeSafe jev typed judgments.

HiQS' unofficial starter code for Jev model with examples and toolkit based on our testing

Fast 'System One' reflex for reasoning LLMs: typed probabilistic second opinions from Jev via OpenCode Zen, with PII-minimizing state redaction.

Controle o Windows por voz: transcricao local (faster-whisper GPU) + decisoes tipadas do Jev (TypeSafe AI)

Benchmark: TypeSafe Jev vs traditional LLM on support-ticket routing accuracy, latency, and cost

Event-driven trajectory watchdog for coding agents: per-event LLM judgments (Jev, Claude, GPT) accumulated into quarantine decisions

Live Wikipedia vandalism watch judged by Jev, TypeSafe AI's typed-decision model. Two files, no packages.

Supervised Windows UI Automation with TypeSafe Jev: bounded choices, local policy, verified outcomes.

Jev Workbench: a C++ desktop workspace with Python diagnostics and structured TypeSafe evaluations

Tag X/Twitter accounts with TypeSafe Jev and hide posts by tag (Chrome MV3 / WXT).

Semantic YouTube recommendation filtering powered by Jev

Open, auditable reproduction of TypeSafe Jev: a 2B probabilistic decision model benchmarked across six public suites with weights, code, and receipts.

100% clean-room, copyright-free 2D platformer benchmarking TypeSafe Jev AI in real-time control (60 FPS)

Calibrated decision layer and risk gate for AI coding agents. Routes yes/no, routing and scoring judgments to the Jev System One model at ~$0.0000123 per decision, then enforces the result deterministically through Claude Code PreToolUse hooks. Fitted thresholds, not guessed.

Ask Jev about an image. A CNN reports what it sees with a calibrated confidence or an abstention; Jev judges what it means.

Monorepo for jevfanity, a profanity detector using Jev by TypeSafe AI

Official Model Context Protocol (MCP) server for JevGuard and TypeSafe AI

Experimental System One implementation of the SingGuard-NSFA agent-security taxonom.

jevhunt project

把判断和写作分开的本地 agent：Jev 负责所有能用类型表达的判断，语言模型只写该写的。零第三方依赖。A local agent harness that puts a System One model in front of the LLM.

Jev-style Choice, Score and Noul decisions from ordinary models. An optional DSH plugin.

A malicious instruction protection skill and prehook for zcode

Evaluating TypeSafe's System One model Jev through camera-based food allergen estimation (Flutter, iOS/Android)

Evidence-backed Jev opportunity mapping for public GitHub repositories

Typed questions over your own state, not a chat. A front end for TypeSafe's System One API.

Four linters in this ecosystem are powered by Jev. None lints the Jev questions themselves. Nine rules from TypeSafe's own documented failure modes, no API key, no labelled data, 0.03s.

Rust client for TypeSafe AI's Jev (System One) model: sans-IO core, typed questions, WASI transports

What Jev answers when the only options are true and false. 3,539 statements put to TypeSafe's System One model, one subject swapped at a time — open dataset, runner and every raw response.

Jev (TypeSafe System One) plays Minesweeper, scored against an exact probability oracle. Its judgments came out worse than random — and 86% of the model calls turned out to be unnecessary.

Un laboratorio para construir algo creativo, divertido y útil con Jev de TypeSafe AI.

A vacation-themed educational playground for TypeSafe Jev, built with TypeScript, Hono, and Cloudflare Workers.