Guardrails und Sicherheit

Das Framework verbindet Modellschlussfolgerungen und Werkzeugnutzung mit Kontextverwaltung, Ereignisströmen, Speicher und Berechtigungssteuerung. JevClassifierModel integriert den Jev-Klassifikator über das TypeSafe SDK in dieses allgemeine Agentenframework.

Die native CLI bietet interaktive Shell, Skriptanfragen und Einbettung sowie Skills, MCP-Server und Unteragenten. Jev erhält Prüfrichtlinie, Kontext und ausstehende Aktion und liefert clear oder caution an die Berechtigungsverarbeitung.

awesome-jev-by-typesafe sammelt belegte Anwendungsfälle, Prompts, Muster und Einstiegscode als unabhängige Community-Ressource. Es erklärt Choice, Score und Noul für Klassifikation, Routing und Prüfung; der Code behält die Ablaufkontrolle. Python- und TypeScript-Beispiele stehen bereit. Offline-Policytests benötigen keinen Schlüssel; Live-Beispiele erfordern TYPESAFE_API_KEY.

Jev Review untersucht Git-Diffs oder ganze Codebasen auf Korrektheit, Sicherheit, Zuverlässigkeit, Kompatibilität und Testabdeckung. Jev wählt Belege, klassifiziert Mechanismen, bewertet den Schweregrad und wählt bei Bedarf per Choice einen Reviewer. Der Code setzt Schwellenwerte und Ablaufregeln um. Mit Node.js 24+, Git und TypeSafe-Schlüssel JSON-Berichte erzeugen oder gespeicherte Prüfungen lokal ansehen.

Real-time Discord moderation bot: Jev evaluates messages and metadata in parallel to catch phishing, spam, and social engineering with a progressive escalation ladder.

Auto mode for every coding agent, built on Jev: risk-scores every tool call with session context (deny / ask / allow), flags prompt injection in results, checks skills and plugins. Claude Code, Codex, Copilot, Gemini, Cursor, pi, OpenCode, ACP.

Typed-judgment admission control at the traffic edge: three-layer prompt-injection and abuse filter for nginx/OpenResty, powered by TypeSafe Jev. Fail-open, cached, hot-reloadable.

Self-evolving cognitive decision engine & TypeSafe Jev alternative. Deliberative decision flywheel ('explore first, crystallize later' System 2→1) with 100% option-order invariance. Typed probabilistic API (Choice, Noul, Score) for Open LLMs, Laya (ModernBERT), & commercial Jev with calibrated logprobs and adaptive safety guards.

A Clash Royale bot with no trained policy: Jev (TypeSafe System One) makes every decision from the live game state

Typed judgment tools for MCP agents. TypeSafe's Jev model as verify, screen, find, classify, rerank, decide, compare, extract, review, gate, and score: the model judges, policy decides auto, review, or escalate.

位于 nginx 与大模型后端之间的前置过滤反向代理：逐请求提取用户输入交给 JEV 判定，有害拦截、正常透明放行

A minimal permission gate for Pi in the style of Claude Code's auto mode

A curated, evidence-led directory of DeepSeek Harness (DSH) plugins: verified loadable extensions, skills, and permission-aware installation guidance.

Compile agent policy prose into deterministic verdict programs: narrow evidence questions for the model, the verdict computed in code. Install: npm i -g jev-compiler

Input moderation for Mastra agents on TypeSafe Jev — one file

Decision maker for Ruby: turn decision model answers into policy verdicts

Fast System-1 Decision, Arbitration & Safety Engine for Autonomous AI Agents (Powered by TypeSafe Jev)

基于 Jev 的聊天决策辅助：读当前聊天窗口，只返回类型化判定与置信度，不生成回复。QQ / 飞书 / 抖音实测可用。· An on-device chat co-pilot built on Jev — typed verdicts instead of prose, no package allowlist.

An automated Discord moderation bot, powered by Jev.

Experimental Jev permission gate for Claude Code via OpenRouter, with reproducible latency and cost benchmarks

Effect-based safety gate for AI coding agents' shell commands (OpenCode, Antigravity): fast structural rules, then TypeSafe's Jev or a chat model judges what a command does. Certified with Jev at zero dangerous commands allowed.

Shift every LLM call to the cheapest model that can handle it. Routing decided by TypeSafe Jev in ~180 ms. No training data. Policy in plain YAML. TypeScript and Python.

Blind security benchmarks for Jev, TypeSafe's System One model: prompt injection and vulnerable code detection, built on jev-go

A governed execution layer between typed-decision models (TypeSafe Jev) and agent/AI tooling like Claude Code, MCP, and AI SDKs, turning probabilistic decisions into deterministic, policy-controlled, and auditable actions.