2026年9月16日 发布的项目

该浏览器智能体将可见页面控件整理成带索引的操作空间，用于执行自然语言目标。 Jev 在一次请求中选择操作及兼容元素；只有 TYPE_TEXT 的文本内容由独立的小型大语言模型生成。

SemIf 直接读取选项概率，处理路由、重试、证据判断等小型语义决策，无需生成回答句子。 它借鉴 Jev 的接口模式，不复现其未公开的模型或训练，也不隶属于 TypeSafe。

机器人随 Monad 区块更新仅挂单限价报价，并将盘口、订单与成交事件推送到仪表盘。 配置 MODEL=jev 和 TYPESAFE_AI_API_KEY 后由 Jev 选择买或卖；默认 mock 模型只是动量启发式。

typesafe-computer-use 根据文字目标操作 Mac，通过 OCR 和辅助功能控件读取桌面。 TypeSafe Jev 选择动作、目标和网站；代码提供事实，独立的写作模型负责自由文本。 需要 TypeSafe 密钥、屏幕录制和辅助功能权限。命令行支持先预演，再实际控制鼠标和键盘。

decider 根据状态和类型化问题，在一次前向计算中返回 Choice、Score 和 Noul 概率。 这是基于 Qwen 的独立 Jev 风格模型系列，训练数据并非从 Jev 蒸馏而来。 可调用 Python Decider 接口，或让 TypeSafe SDK 连接兼容的本地 HTTP 端点。

Jev Review 扫描 Git 差异或完整代码库，检查正确性、安全性、可靠性、兼容性和测试覆盖。 Jev 选择证据、判断问题机制并评估严重程度，满足条件时通过 Choice 选择审查者；代码负责阈值与流程策略。 使用 Node.js 24+、Git 和 TypeSafe 密钥，生成 JSON 报告或在本地查看已保存审查。

jev-router 通过本地代理启动原生编程 CLI，并为每个新用户轮次选择模型档位。 用 npm 安装 jev-router，配置 JEV_API_KEY，在上游 CLI 已安装并登录后运行 jev-claude 或 jev-codex。

TypeSafe Mario 是实验性 NES 控制循环，使用模拟器遥测及 RAM 提取的游戏观察。 准备 Python 3.13、TypeSafe 密钥和本地游戏环境，先检查 state-demo，再运行 play 命令。

在 perch.yaml 中定义语义要求，并按文件或方法限定范围，再查看代码检查结果。 System One 客户端针对给定状态提交类型化问题，默认使用 jev-latest 并调用 TypeSafe API。

Camera-only autonomous drone in MuJoCo with a small judgment model (TypeSafe Jev) in the loop at 2.5Hz

Guardrails for Pi that steer instead of interrupt: enforces your project rules on every write, holds only hard-to-undo actions (3 per 1,000 calls, measured on real sessions), catches unverified "done" claims and stuck loops, and trims tool output. Built on pi-typesafe.

Typesafe.ai System One Model Jev navigating a Neo4j graph by using a classifier over neighbouring relationships

TypeSafe Jev as a decision layer for the Pi coding agent: a measured tool-call gate plus jev_ask for typed, calibrated answers

A calibrated context sieve for Claude Code: every tool result is judged by a System One model before it enters context.

A small, type-safe client for asking AI questions about your data, powered by TypeSafe Jev.

Browser automation where an LLM plans and Jev (Typesafe System One) decides. Library, CLI and MCP server.

Codebase search powered by Jev from @typesafe-ai

Typed semantic decisions for Unix pipelines and CI, powered by TypeSafe AI Jev.

JEV-inspired parallel decisions for CUDA LLMs. One context, many decisions. vLLM API, game-agent examples, and reproducible benchmarks.

TypeSafe decisions for Pi: batched evaluation tool, terminal playground, and typed API for extension authors

Typed JSON inference with DiffusionGemma, with Every and Jev benchmark results

This is a LLM Gateway that mimics typesafe ai structured output. Like an imposter Jev.

Local bilingual probability decisions from context, questions, and candidate answers. Independent research preview inspired by TypeSafe Jev.

TypeSafe Jev controls original StarCraft shareware through keyboard and mouse with recorded action probabilities.

Agent-ergonomic CLI for TypeSafe's Jev: fast calibrated judgments (pick, rate, check, rank, triage, guard) from the shell

Unofficial study: Jev-style parallel typed decisions on stock 1.5B-8B models on an Apple Silicon laptop. Benchmarks, research notes, and a Hugging Face Space demo.

MCP server for TypeSafe Jev: typed classify, score, check, match and screen for any agent, with confidence on every answer

Typed, confidence-aware agent skill routing with TypeSafe Jev.

Community TypeSafe AI playground: 110 use cases, games, dilemmas and model challenges, with editable prompts, A/B comparisons and a mobile-friendly UI.

TypeSafe structured-output provider for RubyLLM 2

BrighTO LLM Router: free open-source, ultra-fast self-hosted Rust LLM gateway for OpenAI/Anthropic APIs, SystemOne/JEV/DJEV decisions, Ollaya/Laya, load balancing, fallback routing, team keys and token budgets.

TypeSafe (Jev) skill routing for Hermes Agent: names the one skill worth loading, before the model call. Opt-in, stdlib only, ~$0.001 per routed turn.

Async LangGraph workflow that gets a typed Jev Choice (invoice or general) and routes each inbound email to the matching handler.

Community .NET SDK for the TypeSafe AI System One API — typed noul, choice, and score questions with structured, confidence-scored answers. Not affiliated with TypeSafe AI.

Can a decision model beat dedicated rerankers? TypeSafe Jev vs Cohere Rerank 4 vs ZeroEntropy zerank-2 vs a chat-model baseline: 14 datasets, every raw API response, bootstrap ranges on every gap.

Hybrid coding harness: System 2 writes, System 1 (Jev) runs reflexes.

Ruby client for typesafe.ai

macOS computer use driven by Jev (TypeSafe System One) as the decision maker

Benchmarks and a playground for TypeSafe's Jev (System One) model: chess, and who-is-the-player-talking-to for speech-to-text game NPCs

A word-level language model whose output layer is Jev: n-gram drafter, Noul chunk verification, bits-per-token eval

Jev (TypeSafe) vs Claude Haiku 4.5 on 2 000 phishing emails: accuracy, calibration, latency, cost. Reproducible benchmark.

Discriminative Monte Carlo Tree Search using System One and Harnesses

Go client for TypeSafe's System One API and its model Jev: typed judgments and calibrated probabilities instead of generated text

Experimental multi-horizon BTC signal generator using TypeSafe Jev probabilities and Binance market data.

DeepSeek Harness (dsh) 插件:论文伴读工作台 —— PDF 转录 / 检索 / 流式问答 + 内置阅读器页面。

Eight minimal working examples of TypeSafe's Jev (a System One model) applied to mechanical and electrical engineering: CAD/CAE/CAM routing, FEM result triage, DFM screening, BOM alignment, hallucination-proof extraction. Zero dependencies.

Jev-style calibrated decision model (Choice/Score/Noul) on Qwen3.5-0.8B

Ask Jev typed questions from the shell: noul, choice, and score answers as numbers, not prose

Independent Jev 1.13.0 behavior study: report, controlled prompt experiments, raw results, and offline verification.

Drives a real browser with Jev making every decision and Vercel's agent-browser performing every action, with a benchmark.