Launches on September 25, 2026

Java 21 client for TypeSafe's Jev evaluator: typed questions, typed answers, no decisions made for you

用 TypeSafe Jev 给 Gmail / Outlook 邮件自动分类：域名规则定平台，模型判类型，邮箱规则管新邮件

Experiments with TypeSafe Jev for typed, confidence-aware model routing across Claude and Kimi.

Reference coding agent wiring Jev (System One) and Opus 5.5 (System Two)

Context-aware profanity detection for JavaScript and TypeScript, powered by Jev AI with optional fast word-list matching.

Two-stage session and passage retrieval with optional Jev reranking

RAG evaluation with typed decisions: sentence-level hallucination verdicts with offsets and citations, passage and answer relevance, in one Jev call per answer. Benchmarked against Ragas, DeepEval, an LLM judge and HHEM.

Jev (TypeSafe AI System One Model) 技术趋势调研：GitHub 生态、工程应用与 Agent Harness 落地建议

Pseudonymisation wrapper for Jev queries

Technical case study exploring Jev (jevtypesafeai.com) as a semantic decision layer for ad detection in a Chrome extension.

A simple Python train station assistant built with Jev that uses natural-language input and Noul confidence scoring to route users based on whether they have lost something.

Delegate browser operations to an ultra-fast MCP server, executing full page tasks in seconds with one tool call.

Response validation, observation and evaluation utilities for the official TypeSafe SDK.

Control Chrome by voice: open sites, search, click links and fill in forms. Chrome extension powered by Jev.

jev-vs-laya

You vs Jev: race across Wikipedia by clicking links; Jev picks the closest of dozens each hop, one real typed decision, no hallucination. Single-file, no build. Play free: jevtypesafeai.com/games/jev-wikiracer

Decision contracts, adversarial testing, tracing, stability, and privacy controls for TypeSafe Jev

Pi extension that selects model-compatible reasoning effort with TypeSafe Jev

kblam: the knowledge base for LLM-assisted mereology. it's a flat-file research/knowledge storage system with enforced organization (using heuristics, semantic similarity, and Jev evaluation) to prevent document drift, internal contradictions, incorrect data gleaned from reading superseded sections out of context, and other LLM bullshit. not done

Local computer use on macOS with Laya decisions — read any window's Accessibility tree, let a local System One model pick the control, execute via cua-driver. No screenshots, no vision, $0. Sibling of laya-browser-agent.

series of scripts i use to test laya (and supposedly any jev like endpoint). its just curling things so the idea is this helps me get comfortable with the shapes of the api requests

A thin shim for Laya/Laya-MLX

LoopGrid for Jev - Signed, tamper-evident evidence for live TypeSafe AI Jev decisions through Vercel AI Gateway.

Hooks and MCP that make Claude Code, Claude Desktop and Codex delegate decisions to TypeSafe JEV through a local gateway.

A code review check that learns from reviewer rulings without fine-tuning. Jev (TypeSafe AI) judges each diff hunk against your team's guidelines; a Meko datapack (powered by YugabyteDB) holds the guidelines, the rulings, and a trace of every decision. Companion repo for the blog post "Jev and Meko Make Code Review Feedback Stick".

Talk to your Mac. It does the thing. A voice assistant for macOS that acts instead of searching.

Automate real Android tasks on live phones with Jev, Mobilerun, and TypeSafe—no ADB needed.

Mule 4 connector for TypeSafe's System One API (Jev decision models): typed yes/no, choice and score answers that drive Mule flows. Think: Smart if-statements.

Manual, evidence-first TypeSafe/Jev decision tools for Pi

Semantic checks for Pi file edits and new-file placement, powered by TypeSafe Jev (System One).

Cook hands-free on any recipe page: say "next", "how much butter?" or "set a timer". Chrome extension powered by Jev.

Sev-X: Kev-descended decision models (DDP, eval context, SRA program)

Labels LinkedIn posts as human, unclear or AI slop as you scroll, with the reasons on hover. Chrome extension powered by Jev.

Sniff your Markdown drafts against house rules with local regex checks and optional AI judgment calls.

The Party watches you work. A joke productivity monitor and a real field test of Jev (TypeSafe System One).

Claim-by-claim fact-checking for AI output, designed for CI. PASS/FAIL/REVIEW with calibrated probabilities and source evidence.

From API key to typed decisions, testable policy, confidence fallbacks, and shadow-mode validation

X draft gate powered by TypeSafe Jev (System One): hard secret checks + Noul scores before you publish.

Pre-signature firewall for AI agent x402 payments: signs only what the user promised. ETHGlobal Tokyo 2026.