Launches on September 23, 2026

LLM-as-a-judge evaluation harness: a confidence-gated cascade (fast typed judge → Claude) that scores LLM outputs and gates CI on a regression diff. TypeScript, real recorded transcripts, real benchmark numbers.

Eval harness for TypeSafe Jev decision models: golden-dataset evaluation, calibration metrics, and schema-guarantee tests for typed decisions (Choice / Score / Noul)

Standalone TypeSafe Jev evaluations on 2026 Korean CSAT and SAT Practice Test #11

Worked Python examples for TypeSafe AI's Jev, plus a 500-run stability study with stored data and generated reports

Terminal UI for trying Jev / TypeSafe System One questions one at a time.

Measured: asking TypeSafe Jev N questions in one call bills the state once. 2,976 real requests, raw data, exact billing check.

Chrome extension that uses TypeSafe AI's Jev (System One) to filter ads and uninteresting content out of Xiaohongshu feeds. BYOK, fail-open, self-serve dislike rules. Weibo & Bilibili adapters on the way.

用 Jev 把一句自然语言设计指令拆成几道选择题，再由代码在 Figma 里画出来 · Jev decides, code draws: a Figma plugin that turns one sentence into multiple-choice questions for Jev (System One)

Chrome extension: priority score badges for Gmail Primary inbox using TypeSafe Jev (patient-related mail excluded)

Jev 특강 실습 코드(1~3강): TypeSafe Jev를 Claude Code·Codex·OpenClaw에 붙이는 스킬·훅·라우터·활용 예제

AI inbox sorting for any IMAP mailbox, powered by Jev (TypeSafe). Plain-English categories, preview-first, never deletes.

Claude Code skill: find every internal link your site is missing, decided by TypeSafe's Jev model. One command, about $1, interactive report.

Real-time AI post classifier for X (Twitter) using TypeSafe AI's Jev model

jev-kit: Jev-powered triage across the dev loop (test/build/error) - RETRY/FIX_CODE/FIX_ENV on real failure logs

Local playground for decision models in the shape of TypeSafe's Jev, running the open openjev replication (Qwen3.5 as an NLI cross-encoder). CPU-only, no API key.

O Jev (TypeSafe AI) como guardrail em tempo real e como pré-avaliador de agentes conversacionais de negócio. Next.js + LangChain via Vercel AI Gateway, com estudo de calibração em português.

Five production tools for the lifecycle of Jev and compatible typed decision models.

MCP service exposing TypeSafe Jev/System One bounded-decision primitives over remote HTTPS/MCP

Claude Code skill that uses Jev (TypeSafe System One) to route tasks to the best Claude model — Opus, Sonnet, Haiku, or Fable — by capability, cost, and latency.

model-agnostic typed-decision judge plugin for Paseo

Typed distribution patterns for Jev responses: descriptive shapes, composable predicates, and exhaustive matching.

Let jev play game.

Jev-supervised madmax conductor for pi: tiny slices, parallel pi-subagents workers in worktrees, TypeSafe Jev makes routing/review/QA/supervision decisions.

A text box that becomes what you mean — every keystroke classified by TypeSafe's Jev model. Next.js + Cloudflare Workers.

A playground for Jev by TypeSafe AI. Try a prompt, change the criteria, and see what Jev decides.

Classify and tag Jev projects, System One models, and real-world use cases with Jev. GitHub Stars rank; humans review.

Measures whether TypeSafe's Jev decision model flips actions on identical replays near shipped thresholds across 12,705 calls. Replication data, harness, and independent verification.

Jev Resume Screener: open-source AI resume screening with TypeSafe's Jev decision model. YES/NO verdicts, 0-100 fit scores, best-candidate pick, and a Jev vs LLM comparison.

RPG pixel art 8-bit donde los NPC toman decisiones con Jev (TypeSafe) vía Vercel AI Gateway

Unofficial local Jev Router adapter for the Codex desktop app on macOS

Shadow-mode eval of TypeSafe Jev as an AIOps triage router vs Qwen3.8-27B and Claude Sonnet 5

Internet search and page reading for coding agents, with TypeSafe Jev ranking

An interactive image classification workbench powered by Jev, with a virtual conveyor belt and visual results.

claude code plugin that measures jev before trusting it: background verdicts, joined with the classifier's real denials, and a report

Skill do Claude Code: o Jev (TypeSafe via OpenRouter) escolhe o modelo e classifica pilhas; o Claude escreve.

Agent skill: decide when and how to integrate TypeSafe's Jev decision model (OpenRouter, TypeSafe, Vercel AI SDK, harness patterns, calibration). Claude Code plugin + npx skills.

Semantic skill routing for Pi, powered by Jev.

TypeSafe cookbook skill suggestion (Jev) as a stdio MCP server for Hermes: one Choice call ranks the roster + gates, a second reranks the top-3; suggests at most one skill.

Jev skill suggestion demo: rank and gate which agent skill to load for a request (TypeSafe Choice + Noul)

Local System One-compatible decision server for Laya backends

Small apps built on Jev (TypeSafe AI): Are they into you? and Finder, on Cloudflare Workers.

A small demo showing a cheaper way to route support tickets.

Single-file Bun proxy: TypeSafe Jev API → local Laya (or any Jev-like backend)

Pre-registered test of the Jev AI decision model as a crypto trader on Base: 40,228 decisions, worse than a coin flip. Interactive write-up EN/PT.

Jev(TypeSafe AI System One) 한국어 쿡북 — mdBook

Ticket router built on TypeSafe's Jev, the companion code for a DataCamp tutorial. Six typed questions in one call, routing policy in plain Python.

Jev (TypeSafe AI の System One / 判断モデル) をフォームとJSONで分かりやすく試せるローカル用プレイグラウンド

typesafe.ai JEV decision model quick demo

Local Vampire Survivors bridge and TypeSafe Jev agent with recorded gameplay, action probabilities, latency, tokens, and cost.

Jev (TypeSafe) vs GPT-4o-mini: classifying YouTube comments with a Judgement Model — demo for CT-YT-2026-043