Jev ecosystem projects — page 84

S1Rank: Can a System-One decision model (TypeSafe Jev) rerank? Benchmark, raw responses, and paper.

Give Claude Code a second opinion it can ask for in half a second. Reusable "battery" checks plus optional risk and finish gates, powered by TypeSafe's Jev.

Open self-hosted System One decision API (Jev wire-compatible) + agent skill — 87.9% JevBench, beats Jev on accuracy and score

Jev-powered model router for AI coding agents (omp + Claude Code): small jobs go to smaller, cheaper models.

Rank your agent skills against a request so your agent can suggest which to invoke; powered by Jev

A Chrome extension that pauses you before clicks you might regret, scored by Jev.

Free, bring-your-own-key Chrome extension that estimates whether text was written by AI, using TypeSafe's Jev decision model. No server.

Chrome extension that subtly marks obviously AI-written text on web pages, scored by TypeSafe Jev

Snake played by TypeSafe's Jev decision model, coached by Claude

Ask typed questions in TypeSafe Jev format (noul, choice, score) and run them on Jev or any LLM - probabilities, agreement and type violations side by side.

A small Spring Boot ai app that picks the right OpenAI model for each request, depending upon the toughness of the question..

Strands agents that make small local LLMs (Qwen3 0.6B, MiniCPM5 2B, Qwen3.5 4B) answer like Jev, a System 1 model: typed yes/no, choice and score answers with probabilities. Benchmarked against Jev on the jevals suite, via logprob readout and written-out JSON probabilities.

Sub2API plugin that adapts jev-* requests to the JEV Decision API

Research: what a fast, cheap classification model (TypeSafe's Jev) is good for inside a real decision system, tested on Polymarket prediction markets with Claude for research. Shadow-first; findings in docs/FINDINGS.md. Not advice.

Type-safe Swift client package (TypeSafe library)

High-speed dual-process browser agent combining sub-150ms Jev reflex decisions with Stagehand stealth automation and Gemini/Claude System 2 fallback.

An unofficial Go module for interacting with Jev and other System One models

One C# IDecider, three engines: TypeSafe Jev (cloud), any local model in a constrained-decoding harness, and a fake for tests. 'System One' bounded decisions in .NET.

After-queue Postfix spam classification with TypeSafe Jev, subject tagging, and recipient-scoped filtering.

Android app that reads visible Bilibili comments and shows a floating sentiment distribution, powered by TypeSafe Jev.

Transaction Assessment & Logic Orchestration System

Tavli (Greek backgammon) played against Jev, TypeSafe's System One decision model. Includes a tutorial on building with typed decision models.

Thunderbird add-on that flags phishing mail impersonating Japanese organizations. Optional TypeSafe Jev support.

Keeps a test-writing agent and a code-writing agent on the same plan, using TypeSafe (Jev) judgments: requirement coverage and failing-test blame routing.