Browse the live Awesome JEV gallery · Featured projects · All 148 projects · JSON catalogue · 简体中文 · 日本語

Awesome JEV

A source-reviewed gallery of JEV-related projects with 50+ GitHub stars (one editorial exception) — integrations, tools, open models, experiments, and ecosystem resources.

We list source-reviewed JEV-related repositories at or above 50 stars, plus one editorial exception: Jev Reviewer (30 stars). See where JEV chooses, scores, routes, or filters—while application code keeps control of execution.

At a glance

Source-reviewed projects	Repositories ≥1K stars	Use-case groups	Snapshot
148	34	10	2026-09-21

Browse by category

Browser & computer use · 14 · SDK integrations · 23 · Routing & optimization · 14 · Open models · 17 · Search & data · 9
Safety & review · 12 · Agent workflows · 11 · Interfaces & automation · 8 · Developer tools · 27 · Domain tools · 13

Featured project gallery

Jev Ultrafast · Browser & computer use · 13.8K stars Chooses a browser action and matching DOM target in one JEV request. Original case · 2.9M views · Fixed-commit evidence	LiteLLM · JEV Router · Routing & optimization · 59.3K stars Classifies request complexity before backend model routing. Fixed-commit evidence
LangChain · TypeSafe · SDK integration · 146.8K stars Exposes binary, classification, and ordinal JEV decisions inside Python workflows. Fixed-commit evidence	Vercel AI SDK · TypeSafe · SDK integration · 26.9K stars Maps choice, score, and yes/no questions onto a unified evaluate interface. Fixed-commit evidence
Laya · Open model · 6.7K stars Multilingual System 1 typed decisions in one forward pass, with a checkpoint router. Hugging Face model · Fixed-commit evidence	jegrep · Search & data Searches live code trees semantically without building an index. Fixed-commit evidence
Cua · JEV Use · Browser & computer use · 25.4K stars Selects an action ID from DOM or visual-region descriptions. Fixed-commit evidence	Agentgateway · JEV Guardrail · Safety & review · 5.0K stars Scores jailbreak, harmful-content, and secret-leakage checks at the gateway. Fixed-commit evidence
NewsJack · Agent workflow · 1.3K stars Filters hundreds of live news items before an agent handles the selected opportunities. Original case · 583K views · Fixed-commit evidence	QuantDinger · JEV Gate · Domain tool · 11.9K stars Adds an evidence and risk gate before selected live-trading entries. Fixed-commit evidence

Explore with filters and full descriptions →

All 148 projects

Browser & computer use (14)

• Cua · JEV Use · 25.4K stars — A computer-use example pairing Cua Driver observation and execution with bounded JEV choices. Source
• Jev Ultrafast · 13.8K stars — A browser agent that uses JEV to choose an action and matching DOM element, calling a text model only when input text is needed. Source
• Agent Desktop · JEV Skill · 1.4K stars — An optional JEV skill that chooses native desktop controls and actions from accessibility data. Source
• TypeSafe Computer Use · 693 stars — Builds candidate actions from OCR and UI state for Jev to control macOS, calling a text model only when writing is needed. Source
• Omg.dev · 533 stars — An omg.dev mobile testing script can use Jev to read the accessibility tree and choose the next interaction. Source
• Jev CU · 519 stars — Computer-use helpers that call JEV to decide the next UI action. Source
• Mobile JEV · 304 stars — Controls an Android phone through Mobilerun, with a web studio and CLI showing Jev decisions. Source
• JEV Browser Use · 291 stars — A Skill for Codex browser workflows: Jev chooses navigation, clicks, and scrolling while Codex handles text input and final checks. Source
• Third Hand · 270 stars — A Swift client stack that calls JEV while driving local computer-use style helpers. Source
• JEV Browser · 209 stars — Drives a browser from a task and URL, returning the final page, screenshot and action trace. Source
• JEV Voice Browser · 187 stars — Controls a Playwright browser by sending incremental speech transcripts to Jev. Source
• JEV Use · 79 stars — A macOS computer-use harness that drives Accessibility controls without a vision model. Source
• TypeSafe Adblock · 64 stars — An experimental Chrome extension that asks Jev whether candidate DOM elements are ads, then highlights or removes them. Source
• JEV Browser · 62 stars — A browser automation library, CLI, and MCP server pairing LLM plans with JEV actions. Source

_{↑ Back to categories}

SDK integrations (23)

• LangChain · TypeSafe · 146.8K stars — An optional JEV classifier integration for Python LangChain workflows. Source
• Composio · TypeSafe Provider · 30.3K stars — A TypeSafe provider that uses JEV to choose among tools and bounded argument options. Source
• Vercel AI SDK · TypeSafe · 26.9K stars — The AI SDK TypeSafe provider exposes JEV through the shared evaluate interface. Source
• Pydantic AI · TypeSafe · 20.1K stars — A TypeSafe model integration that maps supported structured outputs to JEV questions. Source
• Eliza · TypeSafe Client · 19.4K stars — An optional TypeSafe HTTP adapter in the Eliza agent operating system. Source
• LangChain.js · TypeSafe · 18.2K stars — A TypeSafeClassifier integration for JavaScript LangChain workflows. Source
• Rig · TypeSafe Crate · 8.7K stars — An experimental Rust crate for expressing JEV questions and answers with Rust types. Source
• Ax · TypeSafe Client · 2.9K stars — A TypeSafe integration for boolean, finite-class, and native JEV signatures. Source
• Vellum Assistant · JEV · 1.3K stars — An optional JEV provider that submits conversation state and explicit questions. Source
• Req LLM · 580 stars — A TypeSafe provider for calling Jev through ReqLLM’s evaluate interface in Elixir. Source
• Simple JEV · 428 stars — Adapter turning open LLM endpoints into Jev-compatible classification services without training a separate classifier head. Source
• Classifier.dev · 396 stars — A zero-shot classification service with JEV as its primary typed-decision backend. Source
• Instructor Php · 327 stars — A TypeSafe Decision driver within Instructor PHP’s Polyglot module. Source
• Openai Scala Client · 248 stars — A dedicated TypeSafe module in a Scala client that supports multiple AI providers. Source
• Openjev · 245 stars — An independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model. Source
• Pi Fabric · 241 stars — Pi’s programmable runtime includes an optional Jev loop for observing state, making decisions and running bounded actions. Source
• System One Adapter (Python) · 218 stars — A drop-in TypeSafeClient replacement that backs System One calls with ordinary LLM APIs. Source
• TypeSafe SDK Js · 196 stars — The JavaScript and TypeScript SDK published by TypeSafe, with typed Jev requests and answers. Source
• TypeSafe SDK Python · 171 stars — Official TypeSafe Python SDK with synchronous and asynchronous clients for Jev System One, plus question and answer types. Source
• Runline · 162 stars — A TypeSafe plugin exposing Jev decisions as callable actions in Runline Agent JavaScript. Source
• AI · 133 stars — A Jev forwarding endpoint in the Hack Club AI proxy, using its authentication, limits and usage logging. Source
• Effect Agent · 121 stars — An Effect Agent TypeSafe decision provider for typed question sets and optional model selection. Source
• Advocaat · 88 stars — A small TypeScript client for asking Jev multiple typed questions about the same data. Source

_{↑ Back to categories}

Routing & optimization (14)

• LiteLLM · JEV Router · 59.3K stars — LiteLLM can use JEV inside its complexity-based model router. Source
• Oh My Pi · JEV Judgments · 32.2K stars — A coding agent with an optional TypeSafe judgment provider for bounded workflow decisions. Source
• OpenChamber · JEV Router · 10.2K stars — An optional automatic model router that classifies a message before model selection. Source
• Firstmate · JEV Dispatch · 6.9K stars — An agent crew dispatcher that can match task briefs to rules with JEV. Source
• Atomic · 809 stars — An optional Jev decision backend in the Atomic coding Agent for bounded structured choices such as routing. Source
• Vexjoy Agent · 421 stars — An optional Jev routing path that matches VexJoy requests to specialist Agents, skills and workflows. Source
• Hermes JEV Skills · 335 stars — JEV-powered routing, memory, compaction, skill selection, and computer/browser use for Hermes agents. Source
• WrongStack · 329 stars — An optional Jev dispatch classifier for choosing among WrongStack specialist Agents. Source
• JEV Router · 283 stars — A Claude Code and CLI proxy that asks Jev to score task complexity and pick a model from the account’s available set. Source
• Skillbox · 220 stars — A self-hosted, versioned Agent skills library with optional Jev recommendations. Source
• JEV Codex Router · 125 stars — Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode. Source
• JevRouter · 125 stars — A lightweight JEV router for models, tools, and subagents. Source
• Grok Bot JEV · 68 stars — A JEV decision layer for Grok Bot with usage gates and skill templates. Source
• Agent Router · 58 stars — A CLI that routes tasks to coding agents, models, and reasoning effort with JEV. Source

_{↑ Back to categories}

Open models (17)

• Laya · 6.7K stars — Multilingual non-autoregressive System 1 decision engine: typed choice/score/noul in one forward pass, with a router across checkpoints. Source
• SemIf · 2.7K stars — An independent open-model implementation of the semantic-if interface pattern. Source
• NanoJev · 1.7K stars — A 0.6B open replica of the JEV interface with parallel decisions and a training pipeline. Source
• Kev 0.5B · 1.5K stars — A tiny trainable Jev-like decision model based on Qwen2.5-0.5B. Source
• Nimble · 1.4K stars — An open toolkit for local typed decisions, data curation, and evaluation. Source
• Jevlike · 1.1K stars — An independent starter model that scores a changing list of text or visual options in one pass. Source
• LocalJev · 657 stars — A local Jev-compatible server with benchmarks across small local models. Source
• Splash · 506 stars — A local Apple Silicon inference engine oriented around decision-style models. Source
• Von · 297 stars — An open, local System One decision model with a JEV-compatible interface. Source
• OpenJEV SGLang · 245 stars — A JEV-compatible API endpoint built on open models with a prefill-only serving path. Source
• OpenJEV Verdict 2.0 · 221 stars — A calibrated ~151M non-autoregressive local decision engine aiming at JEV-style typed answers. Source
• JEV Visual · 192 stars — An educational JEV-like visual inference experiment on Apple Silicon with shared context scoring. Source
• Jeff · 183 stars — A self-hosted Jev-compatible service powered by GliFormer. Source
• Decider · 180 stars — One-pass typed decisions with calibrated probabilities, fine-tuned from Qwen3.5-2B. Source
• djev-spark · 152 stars — A DGX Spark container recipe for running DiffusionGemma NVFP4 structured decisions. Source
• Reflex · 91 stars — A small open decision model that recreates the JEV/System One interface on Qwen. Source
• Open JEV · 82 stars — An open JEV-style implementation with custom fine-tuning support. Source

_{↑ Back to categories}

Search & data (9)

• Kody · 659 stars — Optional second-stage search: widen the hybrid pool, then Score-rerank candidates with Workers AI typesafe/jev. Source
• Jev Experiments · 352 stars — Small Jev experiments including intent search and agent-assist flows. Source
• JEV Search · 352 stars — Searches the web with Jev choosing sources and time ranges, then ranking returned links. Source
• Tax Document Classifier · 314 stars — Classifies extracted tax-document pages against a fixed IRS form catalogue. Source
• Pg JEV · 259 stars — Adds natural-language filtering, classification, and ranking of rows to PostgreSQL queries. Source
• jev-semgrep · 122 stars — Meaning-based grep with multilingual AND, OR, NOT, and probability thresholds. Source
• Pg TypeSafe · 81 stars — A pre-alpha PostgreSQL C extension for calling Jev from SQL for classification, yes/no judgments, and scoring. Source
• jegrep · 57 stars — Semantic grep for live code trees: describe what you need, get files and original line ranges without building an index. Source
• Neo4JEV · 57 stars — A Neo4j graph navigator that uses JEV to classify the next relationship. Source

_{↑ Back to categories}

Safety & review (12)

• Agentgateway · JEV Guardrail · 5.0K stars — A JEV webhook guardrail example for inspecting model requests and responses. Source
• Latitude · JEV Preclassifier · 4.7K stars — An optional JEV preclassifier for selecting which conversation checks should run. Source
• Agent · 618 stars — An optional Jev command-risk advisor inside a native macOS Agent, with a TypeSafeKit client. Source
• jev-align · 253 stars — A CLI for calibrating Jev classifiers from human examples with GEPA. Source
• Abide · 199 stars — A coding-agent rule layer for constraints that ordinary linters cannot express. Source
• Interlinked CLI · 178 stars — Interlinked adds optional Jev judgments and evidence checks to local coding-agent checks. Source
• Unclutter · 152 stars — A browser extension using Jev to identify ads, promotions, and subscription popups and save reusable hiding rules. Source
• Pi JEV · 131 stars — A Pi extension that flags tool risks before execution and checks output for secrets and failure types. Source
• Pi Warden · 118 stars — Adds checks for project rules, out-of-scope actions, repeated failures, and completion claims to Pi Agents. Source
• Supercov · 85 stars — A code-quality and coverage tool that uses JEV for bounded quality checks. Source
• Captaincore · 71 stars — Jev commands in the WordPress toolkit CaptainCore answer structured questions and prioritize malware scanner findings for review. Source
• Oxlint Plugin JEV · 52 stars — An Oxlint plugin that evaluates semantic lint rules with JEV. Source

_{↑ Back to categories}

Agent workflows (11)

• Jev Model Router · 30.9K stars — A Claude Code mod that classifies subagent model and reasoning-effort needs. Source
• Openwork · 23.7K stars — An open-source cowork-style agent workspace that can run JEV-backed reviews inside CI and skill workflows. Source
• Eve · 5.3K stars — Vercel’s open agent framework for packaging skills, evals, and tool-using workflows. Source
• NewsJack · 1.3K stars — An open-source PR workflow that screens a live news feed for timely brand opportunities. Source
• Smithers · 418 stars — An agentic TypeScript workflow framework with a first-class JEV classify path. Source
• Compact Adviser · 157 stars — A coding-agent skill/plugin that advises when to compact context to save tokens. Source
• JEV Pruner · 130 stars — A Claude Code plugin that trims long Bash output with TypeSafe JEV before the model sees it. Source
• Skillranker · 107 stars — A Rust CLI that ranks agent skills against live session context with JEV. Source
• Bluenoise · 90 stars — An X/Twitter filtering extension using local rules by default, with optional Jev checks for unmatched replies. Source
• jev-gateway · 79 stars — A gateway for inserting Jev decisions into Claude Code and Codex tool flows. Source
• Save Token JEV Clean · 62 stars — A context cleaner that asks JEV which history to retain, truncate, or drop. Source

_{↑ Back to categories}

Interfaces & automation (8)

• json-render · JEV Compose · 17.7K stars — A JEV UI-composition experiment that selects from predefined components and properties. Source
• Aiavatarkit · 678 stars — An optional AIAvatarKit component uses Jev to judge turn endings from speech transcripts. Source
• TipTour macOS · 627 stars — A local macOS computer-use companion driven by Jev or Gemini Live. Source
• Notra · 199 stars — A product that turns work artifacts into publishable content with agent skill hooks. Source
• OpenWhisper · 187 stars — A dictation and meeting-notes app with optional Jev checks for topic changes, note-taker instructions and sensitive text. Source
• Youtube Sponsor Detection · 81 stars — YouTube extension detecting sponsored segments from live audio and transcripts using Jev, skipping promotional blocks automatically. Source
• Jevmeter · 77 stars — Creates edited videos with score meters by asking Jev to rate transcript sentences against selected rubrics. Source
• JEV Chat · 71 stars — A tool-using chat interface where JEV chooses only from code-supplied reply and tool options. Source

_{↑ Back to categories}

Developer tools (27)

• Fast Jev Compaction · 5.7K stars — Compacts Claude Code tool history while keeping retained text verbatim. Source
• TypeSafe Skills · 1.4K stars — Official agent skills for building against TypeSafe’s System One / JEV API. Source
• Celesto · 946 stars — A Celesto PR-review example prepares sandbox checks and compares a general model with Jev on candidate findings. Source
• AI CLI · 808 stars — A terminal generator CLI that ships with an agent skill for creating content from the command line. Source
• Awesome JEV (yibie) · 805 stars — A large community directory of public JEV projects, integrations, and discussions. Source
• Awesome JEV by TypeSafe · 742 stars — An evidence-backed playbook of JEV use cases, patterns, prompts, and starter code. Source
• Awesome JEV Tools · 599 stars — A curated directory focused on tools built for JEV and TypeSafe System One. Source
• Foreman · 446 stars — An independent supervisor loop that reads worker diffs, logs and tests, asks Jev Nouls about stuck / off-track / verify, then applies a Python policy. Source
• Jev Review · 438 stars — Reviews a Git diff or a codebase in stages and displays review leads in a local dashboard. Source
• Awesome TypeSafe · 414 stars — A curated list of official and community resources for TypeSafe, System One, and JEV. Source
• Awesome JEV (cobanov) · 292 stars — A source-backed list of projects built with JEV / TypeSafe System One. Source
• Orchestkit · 280 stars — OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold. Source
• Awesome JEV Projects · 264 stars — A source-backed JEV ecosystem radar with plain-language discovery and GitHub sync. Source
• JEV MCP · 196 stars — Eight MCP tools for evidence checks, content screening, search, reranking, classification, comparison and extraction. Source
• JEV Review · 191 stars — A local MCP code-quality reviewer returning structured scores to coding Agents. Source
• Awesome JEV (fatwang2) · 185 stars — A source-backed JEV directory plus a reusable JEV-only GitHub review workflow. Source
• Awesome JEV (AnotiaWang) · 181 stars — A curated list of JEV applications, libraries, and System One resources. Source
• TypeSafe MCP · 173 stars — An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions. Source
• Perch · 167 stars — Semantic code linting that asks JEV / System One whether findings should fire. Source
• Awesome JEV Gallery · 131 stars — A gallery of System One papers, open reproductions, and independent evaluations. Source
• Building with JEV Skill · 128 stars — An agent skill for writing and improving programs that call JEV / System One. Source
• Taskuary · 116 stars — An optional Jev judgment module in Taskuary for checking user-defined conditions on task state. Source
• Awesome JEV TypeSafe · 115 stars — A TypeSafe JEV resource list centered on typed, confidence-aware decisions. Source
• Awesome Jev · 109 stars — The source catalogue behind awesomejev.com, covering repos, integrations, demos, and research. Source
• Stanley Code · 106 stars — Bounded JEV workflows for coding agents with a dedicated adapter. Source
• JEV Shell History · 80 stars — Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context. Source
• Awesome JEV (AppitStudio) · 66 stars — A curated JEV resource list with runnable typed-decision examples. Source

_{↑ Back to categories}

Domain tools (13)

• AI Hedge Fund · JEV Adapter · 63.6K stars — An educational hedge-fund prototype with an optional JEV adapter for structured strategy judgments. Source
• QuantDinger · JEV Gate · 11.9K stars — An open-source trading OS with a JEV decision gate before selected live entries. Source
• JEV Trader · 1.7K stars — A market-making experiment that can request one JEV trade decision per Monad block. Source
• Astra + JEV Minecraft Agent · 395 stars — A Minecraft agent where a planner sets objectives and JEV selects legal game actions. Source
• TypeSafe Mario · 314 stars — An experimental NES Mario controller that gives Jev structured emulator RAM and telemetry instead of screenshots. Source
• Jevpilot · 143 stars — A browser driving simulator where Jev chooses among locally generated paths and speeds. Source
• JEV Eval Agent · 99 stars — Agent tool evaluation harness comparing standard LLM tool selection against Jev routing across 100 mocked tools. Source
• Formanator · 98 stars — A CLI and MCP client for benefit claims with optional JEV receipt classification. Source
• JEV Drone · 96 stars — A MuJoCo drone experiment deriving scene features from camera buffers for Jev tactical advice. Source
• Dasheng · 90 stars — A reading-practice app that combines streaming ASR with per-word JEV judgments. Source
• JEV Trade · 68 stars — A live Hyperliquid trading agent with JEV as a bounded decision layer. Source
• Prism Liquidity Agent · 54 stars — An autonomous liquidity agent that uses JEV inside its rebalancing decision service. Source
• Jev Reviewer · 30 stars — A browser app that pulls verbatim quotes from trial reports to fill systematic-review extraction forms. Source

_{↑ Back to categories}

BeatAPI

Browse the live gallery · Create a BeatAPI key · Contribute

JEV is live on BeatAPI: use the same BeatAPI key and USD balance to call POST /v1/systemone with model jev-1.13.

---

_{Curated by BeatAPI · Live gallery · License & notices}