diff --git a/README.ja.md b/README.ja.md
index 995b555..ef06952 100644
--- a/README.ja.md
+++ b/README.ja.md
@@ -1,14 +1,22 @@
+
+
+ 
+
+
Awesome JEV
用途に合う JEV プロジェクトを見つけ、何を解決し、どの実装が参考になるかを確認できます。
+自分のアプリに JEV を組み込みたいですか？ BeatAPI で無料の JEV API を使い始める →
+
プロジェクトを探す · 注目プロジェクト · Agent で検索 · 全プロジェクト · サイトで検索
English · 简体中文 · 日本語
-183 プロジェクト · 10 種類 · 更新 2026-09-22
各リポジトリ 50★以上 · ソース確認済み
+192 プロジェクト · 10 種類 · 更新 2026-09-23
各リポジトリ 50★以上 · ソース確認済み
@@ -28,13 +36,13 @@
| | |
| --- | --- |
| [](https://github.com/browser-use/jev-ultrafast) | [](https://github.com/BerriAI/litellm) |
-| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · ブラウザ・PC 操作 · 16.3K Star
1 回の JEV リクエストでブラウザ操作と対象 DOM を選択します。
[オリジナル事例 · 294 万表示](https://x.com/gregpr07/status/2100411066966749359) · [固定コミットの根拠](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · ルーティング · 59.4K Star
バックエンドモデルを選ぶ前にリクエストの複雑さを分類します。
[固定コミットの根拠](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
+| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · ブラウザ・PC 操作 · 18.2K Star
1 回の JEV リクエストでブラウザ操作と対象 DOM を選択します。
[オリジナル事例 · 294 万表示](https://x.com/gregpr07/status/2100411066966749359) · [固定コミットの根拠](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · ルーティング · 59.4K Star
バックエンドモデルを選ぶ前にリクエストの複雑さを分類します。
[固定コミットの根拠](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
| [](https://github.com/langchain-ai/langchain) | [](https://github.com/vercel/ai) |
-| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK 連携 · 146.8K Star
Python ワークフローで二値・分類・順序付き評価を扱います。
[固定コミットの根拠](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK 連携 · 26.9K Star
choice、score、yes/no を共通の evaluate インターフェースへ変換します。
[固定コミットの根拠](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
+| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK 連携 · 146.9K Star
Python ワークフローで二値・分類・順序付き評価を扱います。
[固定コミットの根拠](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK 連携 · 26.9K Star
choice、score、yes/no を共通の evaluate インターフェースへ変換します。
[固定コミットの根拠](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
| [](https://github.com/NandhaKishorM/laya) | [](https://github.com/can1357/jegrep) |
-| **[Laya](https://github.com/NandhaKishorM/laya)** · オープンモデル · 11.5K stars
多言語 System 1 の型付き判断を 1 回の前向き計算で行い、チェックポイントをルーティングします。
[Hugging Face モデル](https://huggingface.co/convaiinnovations/laya) · [固定コミット根拠](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)|
+| **[Laya](https://github.com/NandhaKishorM/laya)** · オープンモデル · 17.1K stars
多言語 System 1 の型付き判断を 1 回の前向き計算で行い、チェックポイントをルーティングします。
[Hugging Face モデル](https://huggingface.co/convaiinnovations/laya) · [固定コミット根拠](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)|
| [](https://github.com/trycua/cua) | [](https://github.com/agentgateway/agentgateway) |
-| **[Cua · JEV Use](https://github.com/trycua/cua)** · ブラウザ・PC 操作 · 25.8K Star
DOM や画面領域の説明から次のアクション ID を選択します。
[固定コミットの根拠](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 安全性・レビュー · 5.0K Star
ゲートウェイで脱獄、有害コンテンツ、秘密情報漏えいを判定します。
[固定コミットの根拠](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
+| **[Cua · JEV Use](https://github.com/trycua/cua)** · ブラウザ・PC 操作 · 26.0K Star
DOM や画面領域の説明から次のアクション ID を選択します。
[固定コミットの根拠](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 安全性・レビュー · 5.0K Star
ゲートウェイで脱獄、有害コンテンツ、秘密情報漏えいを判定します。
[固定コミットの根拠](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
| [](https://github.com/elvisun/newsjack) | [](https://github.com/OpenByteInc/QuantDinger) |
| **[NewsJack](https://github.com/elvisun/newsjack)** · Agent ワークフロー · 1.3K Star
大量のニュースを先に絞り込み、選ばれた機会だけを Agent が処理します。
[オリジナル事例 · 58.3 万表示](https://x.com/elvissun/status/2100951347080421409) · [固定コミットの根拠](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer) | **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · 業務特化ツール · 12.0K Star
一部の実取引エントリー前に、根拠とリスクの判断ゲートを追加します。
[固定コミットの根拠](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py) |
@@ -60,49 +68,50 @@ npx skills add BeatAPI/awesome-jev
-## 全 183 プロジェクト
+## 全 192 プロジェクト
種類別に全カタログを確認できます。具体的な目的がある場合は、上の用途別ガイドから始めてください。
-[ブラウザ・PC 操作 · 16](#browser-computer-use) · [SDK・フレームワーク · 30](#sdk-integrations) · [ルーティング · 16](#routing-optimization) · [オープンモデル · 21](#open-models) · [検索・データ · 15](#search-data)
-[安全性・レビュー · 14](#safety-review) · [Agent ワークフロー · 14](#agent-workflows) · [UI・自動化 · 9](#interfaces) · [開発者ツール · 35](#developer-tools) · [業務特化ツール · 13](#domain-tools)
+[ブラウザ・PC 操作 · 17](#browser-computer-use) · [SDK・フレームワーク · 31](#sdk-integrations) · [ルーティング · 16](#routing-optimization) · [オープンモデル · 23](#open-models) · [検索・データ · 15](#search-data)
+[安全性・レビュー · 16](#safety-review) · [Agent ワークフロー · 15](#agent-workflows) · [UI・自動化 · 9](#interfaces) · [開発者ツール · 35](#developer-tools) · [業務特化ツール · 15](#domain-tools)
-### ブラウザ・PC 操作 (16)
+### ブラウザ・PC 操作 (17)
-- **[Cua · JEV Use](https://github.com/trycua/cua)** · 25.8K Star — Cua Driver の観察と実行を、制限された JEV の選択肢と組み合わせたコンピューター使用例。 [根拠](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
-- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 16.3K Star — JEV を使用してアクションと一致する DOM 要素を選択し、入力テキストが必要な場合にのみテキスト モデルを呼び出すブラウザ エージェント。 [根拠](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
-- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.4K Star — アクセシビリティ データからネイティブ デスクトップ コントロールとアクションを選択するオプションの JEV スキル。 [根拠](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
-- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 765 Star — JEV の OCR および UI 状態から候補アクションを構築して macOS を制御し、書き込みが必要な場合にのみテキスト モデルを呼び出します。 [根拠](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
-- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 549 Star — 次の UI 操作を JEV に決めさせる Computer Use 補助ツール。 [根拠](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
+- **[Cua · JEV Use](https://github.com/trycua/cua)** · 26.0K Star — Cua Driver の観察と実行を、制限された JEV の選択肢と組み合わせたコンピューター使用例。 [根拠](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
+- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 18.2K Star — JEV を使用してアクションと一致する DOM 要素を選択し、入力テキストが必要な場合にのみテキスト モデルを呼び出すブラウザ エージェント。 [根拠](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
+- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.5K Star — アクセシビリティ データからネイティブ デスクトップ コントロールとアクションを選択するオプションの JEV スキル。 [根拠](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
+- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 843 Star — JEV の OCR および UI 状態から候補アクションを構築して macOS を制御し、書き込みが必要な場合にのみテキスト モデルを呼び出します。 [根拠](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
+- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 581 Star — 次の UI 操作を JEV に決めさせる Computer Use 補助ツール。 [根拠](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
- **[Omg.dev](https://github.com/BennyKok/omg.dev)** · 535 Star — omg.dev モバイル テスト スクリプトは、JEV を使用してアクセシビリティ ツリーを読み取り、次のインタラクションを選択できます。 [根拠](https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts)
-- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 334 Star — Codex ブラウザ ワークフローのスキル: JEV はナビゲーション、クリック、スクロールを選択し、Codex はテキスト入力と最終チェックを処理します。 [根拠](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
-- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 329 Star — Web スタジオと CLI で JEV の決定を表示し、Mobilerun を通じて Android スマートフォンを制御します。 [根拠](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
-- **[Third Hand](https://github.com/shhivv/third-hand)** · 283 Star — ローカル Computer Use 補助を動かしつつ JEV を呼ぶ Swift クライアント。 [根拠](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
-- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 231 Star — タスクと URL からブラウザを起動し、最終ページ、スクリーンショット、アクション トレースを返します。 [根拠](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
-- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 216 Star — 増分音声トランスクリプトを JEV に送信することで、Playwright ブラウザを制御します。 [根拠](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
-- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 127 Star — TypeSafe Policy 実装を備えた高速 Computer Use アクション層。 [根拠](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
-- **[JEV Use](https://github.com/savka777/jev-use)** · 86 Star — 視覚モデルなしで Accessibility 要素を操作する macOS Computer Use ツール。 [根拠](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
-- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 76 Star — アクセシビリティと WebMCP を対象に固定条件の JEV 実験を行うブラウザー Agent ベンチマークです。 [根拠](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
-- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 68 Star — 候補となる DOM 要素が広告であるかどうかを JEV に尋ね、それらをハイライト表示または削除する実験的な Chrome 拡張機能。 [根拠](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
-- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 67 Star — LLM の計画と JEV の操作を組み合わせるブラウザー自動化ライブラリ、CLI、MCP サーバー。 [根拠](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 387 Star — Codex ブラウザ ワークフローのスキル: JEV はナビゲーション、クリック、スクロールを選択し、Codex はテキスト入力と最終チェックを処理します。 [根拠](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
+- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 356 Star — Web スタジオと CLI で JEV の決定を表示し、Mobilerun を通じて Android スマートフォンを制御します。 [根拠](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
+- **[Third Hand](https://github.com/shhivv/third-hand)** · 288 Star — ローカル Computer Use 補助を動かしつつ JEV を呼ぶ Swift クライアント。 [根拠](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
+- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 247 Star — 増分音声トランスクリプトを JEV に送信することで、Playwright ブラウザを制御します。 [根拠](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
+- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 240 Star — タスクと URL からブラウザを起動し、最終ページ、スクリーンショット、アクション トレースを返します。 [根拠](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
+- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 128 Star — TypeSafe Policy 実装を備えた高速 Computer Use アクション層。 [根拠](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
+- **[fastbrowse · Jev](https://github.com/agent-labs-dev/fastbrowse)** · 97 Star — ページ上の操作候補を索引化し、Jev が次の操作を選び、LLM が計画・読解するブラウザ Agent。 [根拠](https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py)
+- **[JEV Use](https://github.com/savka777/jev-use)** · 92 Star — 視覚モデルなしで Accessibility 要素を操作する macOS Computer Use ツール。 [根拠](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
+- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 79 Star — アクセシビリティと WebMCP を対象に固定条件の JEV 実験を行うブラウザー Agent ベンチマークです。 [根拠](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
+- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 76 Star — LLM の計画と JEV の操作を組み合わせるブラウザー自動化ライブラリ、CLI、MCP サーバー。 [根拠](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 71 Star — 候補となる DOM 要素が広告であるかどうかを JEV に尋ね、それらをハイライト表示または削除する実験的な Chrome 拡張機能。 [根拠](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
[↑ 検索入口に戻る](#discovery)
-### SDK・フレームワーク (30)
+### SDK・フレームワーク (31)
-- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.8K Star — Python LangChain ワークフロー用のオプションの JEV 分類子統合。 [根拠](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
+- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.9K Star — Python LangChain ワークフロー用のオプションの JEV 分類子統合。 [根拠](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
- **[Composio · TypeSafe Provider](https://github.com/ComposioHQ/composio)** · 30.3K Star — JEV を使ってツールと許可済みの引数候補を選ぶ TypeSafe プロバイダー。 [根拠](https://github.com/ComposioHQ/composio/blob/4b5920bf7aa55c8a44657b060d4bd25ce7b13a9a/ts/packages/providers/typesafe/src/decide.ts)
- **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · 26.9K Star — 共通の評価インターフェイスから JEV を利用できる、AI SDK 向け TypeSafe プロバイダー。 [根拠](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts)
- **[Pydantic AI · TypeSafe](https://github.com/pydantic/pydantic-ai)** · 20.1K Star — サポートされている構造化出力を JEV 質問にマッピングする TypeSafe モデル統合。 [根拠](https://github.com/pydantic/pydantic-ai/blob/c4898abb54dc25ae6f6aef208a4c0661b30a455e/pydantic_ai_slim/pydantic_ai/models/typesafe.py)
- **[Eliza · TypeSafe Client](https://github.com/elizaOS/eliza)** · 19.4K Star — Eliza エージェント オペレーティング システムのオプションの TypeSafe HTTP アダプター。 [根拠](https://github.com/elizaOS/eliza/blob/ebc808e3a67fb941e29153d89fc896524d32fe3c/packages/agent/src/services/typesafe/client.ts)
- **[LangChain.js · TypeSafe](https://github.com/langchain-ai/langchainjs)** · 18.2K Star — JavaScript の LangChain ワークフロー向け TypeSafeClassifier 連携。 [根拠](https://github.com/langchain-ai/langchainjs/blob/206d8b992bcf90ce7d46f2158f1ad85fc1d826c0/libs/providers/langchain-typesafe/src/classifier.ts)
- **[Rig · TypeSafe Crate](https://github.com/0xPlaygrounds/rig)** · 8.7K Star — JEV の質問と回答を Rust タイプで表現するための実験的な Rust クレート。 [根拠](https://github.com/0xPlaygrounds/rig/blob/2d16c1b25f6749b3a2cd841beddf767106495069/crates/rig-typesafeai/src/wire.rs#L40)
-- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.2K Star — Bifrost は共通の Decision 操作を通じて TypeSafe をネイティブ Provider として公開します。 [根拠](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
-- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 6.9K Star — セルフホスト AI ゲートウェイに、JEV 専用チャネルとネイティブ Decisions ルートを搭載。 [根拠](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
+- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.3K Star — Bifrost は共通の Decision 操作を通じて TypeSafe をネイティブ Provider として公開します。 [根拠](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
+- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 7.0K Star — セルフホスト AI ゲートウェイに、JEV 専用チャネルとネイティブ Decisions ルートを搭載。 [根拠](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
- **[Open Connector · TypeSafe](https://github.com/oomol-lab/open-connector)** · 5.9K Star — Open Connector は Provider Runtime 経由で TypeSafe 評価を公開します。 [根拠](https://github.com/oomol-lab/open-connector/blob/4e6d8533ae436cba4781b6fb69d1bf31e5fb5ffe/src/providers/typesafe_ai/runtime.ts)
- **[Kiln · JEV Adapter](https://github.com/Kiln-AI/Kiln)** · 5.1K Star — Kiln は互換 JSON Schema を JEV 質問へ変換し、構造化された単発タスクを実行します。 [根拠](https://github.com/Kiln-AI/Kiln/blob/5992e2e51ec6782fce33f277145954ba3ae8467e/libs/core/kiln_ai/adapters/model_adapters/jev_adapter.py)
- **[Ax · TypeSafe Client](https://github.com/ax-llm/ax)** · 2.9K Star — ブール型、有限クラス、およびネイティブ JEV 署名の TypeSafe 統合。 [根拠](https://github.com/ax-llm/ax/blob/5c43344f9ef3016db576fa2c3b59d48ef21b4d71/src/ax/ai/typesafe/client.ts#L1)
@@ -110,20 +119,21 @@ npx skills add BeatAPI/awesome-jev
- **[Vellum Assistant · JEV](https://github.com/vellum-ai/vellum-assistant)** · 1.3K Star — 会話状態と明示的な質問を送信するオプションの JEV プロバイダー。 [根拠](https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts)
- **[Laravel AI · TypeSafe](https://github.com/laravel/ai)** · 1.2K Star — Laravel AI に、JEV を使う TypeSafe 分類プロバイダーが組み込まれています。 [根拠](https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php)
- **[Req LLM](https://github.com/agentjido/req_llm)** · 581 Star — Elixir の ReqLLM 評価インターフェイスから JEV を呼び出す TypeSafe プロバイダー。 [根拠](https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md)
-- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 458 Star — 専用の分類ヘッドを学習せずに、公開 LLM エンドポイントを JEV 互換の分類サービスへ変換するアダプター。 [根拠](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
-- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 406 Star — JEV を主要な型付き判断バックエンドに使うゼロショット分類サービス。 [根拠](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
+- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 485 Star — 専用の分類ヘッドを学習せずに、公開 LLM エンドポイントを JEV 互換の分類サービスへ変換するアダプター。 [根拠](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
+- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 411 Star — JEV を主要な型付き判断バックエンドに使うゼロショット分類サービス。 [根拠](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
- **[Instructor Php](https://github.com/cognesy/instructor-php)** · 327 Star — Instructor PHP の Polyglot モジュール内の TypeSafe Decision ドライバー。 [根拠](https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md)
-- **[Openjev](https://github.com/razorback16/openjev)** · 286 Star — JEV API と互換性のある独立型 System One 意思決定サーバーで、オープンな DiffusionGemma モデルを実行します。 [根拠](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
-- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 248 Star — 複数の AI プロバイダーをサポートする Scala クライアントの専用 TypeSafe モジュール。 [根拠](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
-- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 244 Star — Pi のプログラム可能なランタイムには、状態の監視、意思決定、制限されたアクションの実行のためのオプションの JEV ループが含まれています。 [根拠](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
-- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 243 Star — 通常の LLM API で System One 呼び出しを受ける、TypeSafeClient 差し替えアダプタ。 [根拠](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
-- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 214 Star — TypeSafe によって公開された JavaScript および TypeScript SDK (型指定された JEV リクエストと回答を含む)。 [根拠](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
-- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 188 Star — JEV System One の同期および非同期クライアントを備えた公式 TypeSafe Python SDK、および質問と回答のタイプ。 [根拠](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[Openjev](https://github.com/razorback16/openjev)** · 319 Star — JEV API と互換性のある独立型 System One 意思決定サーバーで、オープンな DiffusionGemma モデルを実行します。 [根拠](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
+- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 266 Star — 通常の LLM API で System One 呼び出しを受ける、TypeSafeClient 差し替えアダプタ。 [根拠](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
+- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 253 Star — Pi のプログラム可能なランタイムには、状態の監視、意思決定、制限されたアクションの実行のためのオプションの JEV ループが含まれています。 [根拠](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
+- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 249 Star — 複数の AI プロバイダーをサポートする Scala クライアントの専用 TypeSafe モジュール。 [根拠](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
+- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 226 Star — TypeSafe によって公開された JavaScript および TypeScript SDK (型指定された JEV リクエストと回答を含む)。 [根拠](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
+- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 205 Star — JEV System One の同期および非同期クライアントを備えた公式 TypeSafe Python SDK、および質問と回答のタイプ。 [根拠](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[AI SDK Python · Jev Evaluate](https://github.com/vercel-labs/ai-python)** · 185 Star — typesafe-ai/jev 経由で型付き質問を送る実験的 evaluate を備えた Vercel Python AI SDK。 [根拠](https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py)
- **[Runline](https://github.com/Michaelliv/runline)** · 163 Star — Runline Agent JavaScript で呼び出し可能なアクションとして JEV 決定を公開する TypeSafe プラグイン。 [根拠](https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md)
-- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 137 Star — NeuroLink は TypeSafe プロバイダーと JEV 分類ルーティングを提供します。 [根拠](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
+- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 138 Star — NeuroLink は TypeSafe プロバイダーと JEV 分類ルーティングを提供します。 [根拠](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
- **[AI](https://github.com/hackclub/ai)** · 133 Star — Hack Club AI プロキシの JEV 転送エンドポイント。認証、制限、使用状況ログを使用します。 [根拠](https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md)
- **[Effect Agent](https://github.com/danieljvdm/effect-agent)** · 121 Star — 型付き質問セットと任意のモデル選択に対応する、Effect Agent 向け TypeSafe 意思決定プロバイダー。 [根拠](https://github.com/danieljvdm/effect-agent/blob/88005e497e9b627eeb16d670f278903c57601da9/README.md)
-- **[Advocaat](https://github.com/pithings/advocaat)** · 89 Star — 同じデータについて複数の型付き質問を JEV に送るための、小さな TypeScript クライアント。 [根拠](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
+- **[Advocaat](https://github.com/pithings/advocaat)** · 90 Star — 同じデータについて複数の型付き質問を JEV に送るための、小さな TypeScript クライアント。 [根拠](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
[↑ 検索入口に戻る](#discovery)
@@ -132,49 +142,51 @@ npx skills add BeatAPI/awesome-jev
### ルーティング (16)
- **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · 59.4K Star — LiteLLM は、複雑さベースのモデル ルーター内で JEV を使用できます。 [根拠](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70)
-- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.4K Star — 境界の明確なワークフロー判断に、任意の TypeSafe 意思決定プロバイダーを使えるコーディング Agent。 [根拠](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
-- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.2K Star — モデルを選択する前にメッセージを分類するオプションの自動モデル ルーター。 [根拠](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
-- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 6.9K Star — JEV を使用してタスクの概要をルールに一致させることができるエージェント クルー ディスパッチャー。 [根拠](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
+- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.8K Star — 境界の明確なワークフロー判断に、任意の TypeSafe 意思決定プロバイダーを使えるコーディング Agent。 [根拠](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
+- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.3K Star — モデルを選択する前にメッセージを分類するオプションの自動モデル ルーター。 [根拠](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
+- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 7.0K Star — JEV を使用してタスクの概要をルールに一致させることができるエージェント クルー ディスパッチャー。 [根拠](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
- **[Agent Router · TypeSafe](https://github.com/theagentrouter/agent-router)** · 2.1K Star — Agent Router は Envoy ベースのゲートウェイに TypeSafe System One トランスレータを備えます。 [根拠](https://github.com/theagentrouter/agent-router/blob/7d7c07ffdb14241362def2f7d3f76cbd06d518cc/internal/translator/typesafe_systemone.go)
- **[LLMGateway · System One](https://github.com/theopenco/llmgateway)** · 1.7K Star — LLMGateway は型付き Schema と Provider マッピングを備えたネイティブ System One ルートを実装。 [根拠](https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts)
-- **[Atomic](https://github.com/bastani-inc/atomic)** · 812 Star — ルーティングなど、範囲を限定した構造化選択に使える Atomic コーディング Agent の任意 JEV バックエンド。 [根拠](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Atomic](https://github.com/bastani-inc/atomic)** · 814 Star — ルーティングなど、範囲を限定した構造化選択に使える Atomic コーディング Agent の任意 JEV バックエンド。 [根拠](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 600 Star — Hermes エージェント向けに、JEV 駆動のルーティング・記憶・圧縮・Skill 選択・PC/ブラウザ操作。 [根拠](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
- **[Vexjoy Agent](https://github.com/notque/vexjoy-agent)** · 423 Star — VexJoy リクエストを専門エージェント、スキル、ワークフローに照合するオプションの JEV ルーティング パス。 [根拠](https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md)
-- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 401 Star — Hermes エージェント向けに、JEV 駆動のルーティング・記憶・圧縮・Skill 選択・PC/ブラウザ操作。 [根拠](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
-- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 331 Star — WrongStack の専門 Agent から担当を選ぶための、任意で使える JEV ディスパッチ分類器。 [根拠](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
-- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 316 Star — Claude Code および CLI プロキシ。JEV にタスクの複雑さをスコアリングし、アカウントの利用可能なセットからモデルを選択するよう依頼します。 [根拠](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
-- **[Skillbox](https://github.com/kitze/skillbox)** · 223 Star — オプションの JEV 推奨事項を備えた自己ホスト型のバージョン管理されたエージェント スキル ライブラリ。 [根拠](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
-- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 184 Star — JEV で Codex の各ターンを分類し、ローカルルールに従ってモデル、推論強度、速度モードを選びます。 [根拠](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
-- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 145 Star — モデル、ツール、サブ Agent 向けの軽量 JEV ルーター。 [根拠](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
-- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 74 Star — 使用量ゲートと Skill テンプレートを備えた Grok Bot 向け JEV 判断レイヤー。 [根拠](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
-- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 63 Star — JEV でタスクをコーディング Agent、モデル、推論強度へ振り分ける CLI。 [根拠](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
+- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 349 Star — Claude Code および CLI プロキシ。JEV にタスクの複雑さをスコアリングし、アカウントの利用可能なセットからモデルを選択するよう依頼します。 [根拠](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
+- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 332 Star — WrongStack の専門 Agent から担当を選ぶための、任意で使える JEV ディスパッチ分類器。 [根拠](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
+- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 229 Star — JEV で Codex の各ターンを分類し、ローカルルールに従ってモデル、推論強度、速度モードを選びます。 [根拠](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
+- **[Skillbox](https://github.com/kitze/skillbox)** · 229 Star — オプションの JEV 推奨事項を備えた自己ホスト型のバージョン管理されたエージェント スキル ライブラリ。 [根拠](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
+- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 173 Star — モデル、ツール、サブ Agent 向けの軽量 JEV ルーター。 [根拠](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
+- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 77 Star — 使用量ゲートと Skill テンプレートを備えた Grok Bot 向け JEV 判断レイヤー。 [根拠](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
+- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 68 Star — JEV でタスクをコーディング Agent、モデル、推論強度へ振り分ける CLI。 [根拠](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
[↑ 検索入口に戻る](#discovery)
-### オープンモデル (21)
+### オープンモデル (23)
-- **[Laya](https://github.com/NandhaKishorM/laya)** · 11.5K Star — 多言語・非自己回帰の System 1 意思決定エンジン。1 回の前向き計算で型付き choice/score/noul を出し、ルータがチェックポイントを選びます。 [根拠](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
-- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.4K Star — semantic-if インターフェイス パターンの独立したオープンモデル実装。 [根拠](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
-- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 2.6K Star — Qwen2.5-0.5B に基づく、小さなトレーニング可能な JEV のような意思決定モデル。 [根拠](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
-- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 1.9K Star — 並列決定とトレーニング パイプラインを備えた JEV インターフェイスの 0.6B オープン レプリカ。 [根拠](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
-- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.5K Star — ローカルでの型付き意思決定、データキュレーション、評価に使えるオープンツールキット。 [根拠](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
+- **[Laya](https://github.com/NandhaKishorM/laya)** · 17.1K Star — 多言語・非自己回帰の System 1 意思決定エンジン。1 回の前向き計算で型付き choice/score/noul を出し、ルータがチェックポイントを選びます。 [根拠](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
+- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 4.5K Star — Qwen2.5-0.5B に基づく、小さなトレーニング可能な JEV のような意思決定モデル。 [根拠](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
+- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.9K Star — semantic-if インターフェイス パターンの独立したオープンモデル実装。 [根拠](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
+- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 2.0K Star — 並列決定とトレーニング パイプラインを備えた JEV インターフェイスの 0.6B オープン レプリカ。 [根拠](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
+- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.6K Star — ローカルでの型付き意思決定、データキュレーション、評価に使えるオープンツールキット。 [根拠](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
- **[Jevlike](https://github.com/vinnylarouge/jevlike)** · 1.2K Star — テキストまたはビジュアル オプションの変更リストを 1 パスでスコアリングする独立したスターター モデル。 [根拠](https://github.com/vinnylarouge/jevlike/blob/94f5fd1b0b11d52bbdfdf4e0ee6aa96b568f8452/README.md)
-- **[LocalJev](https://github.com/githubnext/localjev)** · 704 Star — 小規模なローカル モデルにわたるベンチマークを備えたローカル JEV 互換サーバー。 [根拠](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
-- **[Splash](https://github.com/incoai/splash)** · 578 Star — 意思決定系モデル向けの Apple Silicon ローカル推論エンジン。 [根拠](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
-- **[Von](https://github.com/wfzyx/von)** · 374 Star — JEV 互換インターフェースを持つローカルのオープン System One 判断モデル。 [根拠](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
-- **[Decider](https://github.com/Mapika/decider)** · 279 Star — Qwen3.5-2B 微調整による、1 パスの型付き意思決定と校正確率。 [根拠](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
-- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 273 Star — TypeSafe 互換の System One インターフェースを公開する独立ローカル判断モデルです。 [根拠](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
-- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 258 Star — オープンモデル上の JEV 互換 API。prefill-only の提供経路を使います。 [根拠](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
-- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 246 Star — 約 151M の校正済み非自己回帰ローカル意思決定エンジン。JEV 風の型付き回答を目指す。 [根拠](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
-- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 221 Star — Apple Silicon 上の教育向け JEV 風ビジョン推論。共有コンテキストで候補を直接採点。 [根拠](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
-- **[Jeff](https://github.com/logan-markewich/jeff)** · 205 Star — GliFormer を利用したセルフホスト型の JEV 互換サービス。 [根拠](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
-- **[Laya Node Runtime](https://github.com/receptron/laya)** · 195 Star — オープンな Laya JEV 互換判断モデル向け Node.js / TypeScript Runtime。 [根拠](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
-- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 168 Star — DGX Spark 上で DiffusionGemma NVFP4 構造化意思決定を動かすコンテナレシピ。 [根拠](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
-- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 125 Star — ローカル言語モデルを JEV 互換の構造化判断エンジンへ変換するアダプタ。 [根拠](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
-- **[Reflex](https://github.com/kshetrajna12/reflex)** · 110 Star — Qwen 上で JEV/System One インターフェースを再現する小型オープン判断モデル。 [根拠](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
-- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 91 Star — カスタム微調整に対応するオープンな JEV 形式の実装。 [根拠](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
-- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 67 Star — 型付き判断リクエスト向けのオープンソース JEV 互換実装。 [根拠](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[LocalJev](https://github.com/githubnext/localjev)** · 731 Star — 小規模なローカル モデルにわたるベンチマークを備えたローカル JEV 互換サーバー。 [根拠](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
+- **[Splash](https://github.com/incoai/splash)** · 638 Star — 意思決定系モデル向けの Apple Silicon ローカル推論エンジン。 [根拠](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
+- **[Von](https://github.com/wfzyx/von)** · 498 Star — JEV 互換インターフェースを持つローカルのオープン System One 判断モデル。 [根拠](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
+- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 348 Star — TypeSafe 互換の System One インターフェースを公開する独立ローカル判断モデルです。 [根拠](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
+- **[Decider](https://github.com/Mapika/decider)** · 316 Star — Qwen3.5-2B 微調整による、1 パスの型付き意思決定と校正確率。 [根拠](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
+- **[Laya Node Runtime](https://github.com/receptron/laya)** · 295 Star — オープンな Laya JEV 互換判断モデル向け Node.js / TypeScript Runtime。 [根拠](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
+- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 272 Star — オープンモデル上の JEV 互換 API。prefill-only の提供経路を使います。 [根拠](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
+- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 268 Star — 約 151M の校正済み非自己回帰ローカル意思決定エンジン。JEV 風の型付き回答を目指す。 [根拠](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
+- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 243 Star — Apple Silicon 上の教育向け JEV 風ビジョン推論。共有コンテキストで候補を直接採点。 [根拠](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
+- **[Jeff](https://github.com/logan-markewich/jeff)** · 219 Star — GliFormer を利用したセルフホスト型の JEV 互換サービス。 [根拠](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
+- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 176 Star — DGX Spark 上で DiffusionGemma NVFP4 構造化意思決定を動かすコンテナレシピ。 [根拠](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
+- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 142 Star — ローカル言語モデルを JEV 互換の構造化判断エンジンへ変換するアダプタ。 [根拠](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
+- **[Reflex](https://github.com/kshetrajna12/reflex)** · 122 Star — Qwen 上で JEV/System One インターフェースを再現する小型オープン判断モデル。 [根拠](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
+- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 103 Star — カスタム微調整に対応するオープンな JEV 形式の実装。 [根拠](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
+- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 101 Star — 型付き判断リクエスト向けのオープンソース JEV 互換実装。 [根拠](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[jevmlx](https://github.com/bnsd55/jevmlx)** · 59 Star — Apple Silicon 上で MLX logits から制約付きフィールドを採点し、System One HTTP を提供するローカル層。 [根拠](https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py)
+- **[Open Alternative to Jev](https://github.com/ikermoel/open-alternative-jev)** · 51 Star — 型付き質問をローカル開放モデルの一回 forward に詰める、オープンな System One 風 Decider。 [根拠](https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py)
[↑ 検索入口に戻る](#discovery)
@@ -182,63 +194,66 @@ npx skills add BeatAPI/awesome-jev
### 検索・データ (15)
-- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.4K Star — OpenViking は JEV を使い、Agent の記憶・コンテキスト検索結果を較正して再順位付けできます。 [根拠](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
-- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 24.8K Star — Agent メモリに TypeSafe リランカーを追加し、無関係な検索候補を除外できます。 [根拠](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
+- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.5K Star — OpenViking は JEV を使い、Agent の記憶・コンテキスト検索結果を較正して再順位付けできます。 [根拠](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
+- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 25.3K Star — Agent メモリに TypeSafe リランカーを追加し、無関係な検索候補を除外できます。 [根拠](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
- **[GreptimeDB · JEV SQL](https://github.com/GreptimeTeam/greptimedb)** · 6.7K Star — GreptimeDB に JEV 判断を利用する実験的 SQL 述語を追加。 [根拠](https://github.com/GreptimeTeam/greptimedb/blob/b5199bc59a199875187d808158357a33b09e5f60/src/common/function/src/scalars/jev.rs)
-- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 752 Star — ローカル優先の Agent メモリライブラリで、任意で JEV リランカーを利用できます。 [根拠](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
-- **[Kody](https://github.com/kentcdodds/kody)** · 663 Star — オプションの第 2 段階の検索: ハイブリッド プールを拡大し、Workers AI typesafe/jev を使用して候補者をスコア再ランク付けします。 [根拠](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
-- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 385 Star — JEV を使用して Web を検索し、ソースと時間範囲を選択し、返されたリンクをランク付けします。 [根拠](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
-- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 358 Star — インテント検索やエージェント支援フローを含む小規模な JEV 実験。 [根拠](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
-- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 348 Star — 抽出された税務書類のページを固定 IRS フォーム カタログに基づいて分類します。 [根拠](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
-- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 286 Star — PostgreSQL クエリに行の自然言語フィルタリング、分類、ランク付けを追加します。 [根拠](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
-- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 202 Star — JEV を使って大規模コレクションから関連項目を探す CLI ランキングツール。 [根拠](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
-- **[DocJEV](https://github.com/jerryjliu/docjev)** · 201 Star — 専用 JEV エンジンを備えた文書分類・分割ツール。 [根拠](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
-- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 124 Star — 多言語の AND、OR、NOT、確率しきい値を使用した意味ベースの grep。 [根拠](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
-- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 81 Star — JEV で次のリレーションを分類する Neo4j グラフナビゲーター。 [根拠](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
-- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 81 Star — 分類、はい/いいえの判断、スコアリングのために SQL から JEV を呼び出すためのプレアルファ版の PostgreSQL C 拡張機能。 [根拠](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
-- **[jegrep](https://github.com/can1357/jegrep)** · 75 Star — ライブ コード ツリーのセマンティック grep: 必要なものを記述し、インデックスを作成せずにファイルと元の行範囲を取得します。 [根拠](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 756 Star — ローカル優先の Agent メモリライブラリで、任意で JEV リランカーを利用できます。 [根拠](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
+- **[Kody](https://github.com/kentcdodds/kody)** · 666 Star — オプションの第 2 段階の検索: ハイブリッド プールを拡大し、Workers AI typesafe/jev を使用して候補者をスコア再ランク付けします。 [根拠](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
+- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 416 Star — JEV を使用して Web を検索し、ソースと時間範囲を選択し、返されたリンクをランク付けします。 [根拠](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
+- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 399 Star — 抽出された税務書類のページを固定 IRS フォーム カタログに基づいて分類します。 [根拠](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
+- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 370 Star — インテント検索やエージェント支援フローを含む小規模な JEV 実験。 [根拠](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
+- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 314 Star — PostgreSQL クエリに行の自然言語フィルタリング、分類、ランク付けを追加します。 [根拠](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
+- **[DocJEV](https://github.com/jerryjliu/docjev)** · 287 Star — 専用 JEV エンジンを備えた文書分類・分割ツール。 [根拠](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
+- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 209 Star — JEV を使って大規模コレクションから関連項目を探す CLI ランキングツール。 [根拠](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
+- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 130 Star — 多言語の AND、OR、NOT、確率しきい値を使用した意味ベースの grep。 [根拠](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
+- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 100 Star — JEV で次のリレーションを分類する Neo4j グラフナビゲーター。 [根拠](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
+- **[jegrep](https://github.com/can1357/jegrep)** · 83 Star — ライブ コード ツリーのセマンティック grep: 必要なものを記述し、インデックスを作成せずにファイルと元の行範囲を取得します。 [根拠](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 82 Star — 分類、はい/いいえの判断、スコアリングのために SQL から JEV を呼び出すためのプレアルファ版の PostgreSQL C 拡張機能。 [根拠](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
[↑ 検索入口に戻る](#discovery)
-### 安全性・レビュー (14)
+### 安全性・レビュー (16)
-- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.3K Star — サブスクリプションを API 化するゲートウェイに、コンテンツ審査用 TypeSafe クライアントを内蔵。 [根拠](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
+- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.4K Star — サブスクリプションを API 化するゲートウェイに、コンテンツ審査用 TypeSafe クライアントを内蔵。 [根拠](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
- **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 5.0K Star — モデルのリクエストとレスポンスを検査するための JEV Webhook ガードレールの例。 [根拠](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts)
- **[Latitude · JEV Preclassifier](https://github.com/latitude-dev/latitude-llm)** · 4.7K Star — 実行する会話チェックを選ぶための、任意で使える JEV 事前分類器。 [根拠](https://github.com/latitude-dev/latitude-llm/blob/6b2484c1d74973877fc7d353b8867c7496568d78/packages/domain/flaggers/src/use-cases/run-jev-preclassifier.ts#L1)
-- **[Agent](https://github.com/AgentiLoop/Agent)** · 622 Star — TypeSafeKit クライアントを使用した、ネイティブ macOS エージェント内のオプションの JEV コマンド リスク アドバイザ。 [根拠](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
-- **[jev-align](https://github.com/sutro-sh/jev-align)** · 271 Star — GEPA を使用して人間の例から JEV 分類器を調整するための CLI。 [根拠](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
-- **[Abide](https://github.com/coldteadotai/abide)** · 207 Star — 通常のリンターでは表現できない制約を扱う、コーディング Agent 向けルール層。 [根拠](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
-- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 177 Star — Interlinked は、任意の JEV 判断と根拠チェックをローカルのコーディング Agent 検査に追加します。 [根拠](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
-- **[Unclutter](https://github.com/kitze/unclutter)** · 176 Star — JEV を使用して広告、プロモーション、購読ポップアップを識別し、再利用可能な非表示ルールを保存するブラウザ拡張機能。 [根拠](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
-- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 134 Star — 実行前にツールのリスクにフラグを立て、出力のシークレットと障害タイプをチェックする Pi 拡張機能。 [根拠](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
-- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 132 Star — プロジェクト ルール、範囲外のアクション、繰り返される失敗、完了要求のチェックを Pi エージェントに追加します。 [根拠](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
-- **[Supercov](https://github.com/supercorp-ai/supercov)** · 94 Star — JEV で限定的な品質チェックを行うコード品質・カバレッジツール。 [根拠](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[fx · TypeSafe Permission Reviewer](https://github.com/vercel-labs/fx)** · 3.1K Star — 保留中のツール操作を審査する TypeSafe Jev 権限レビュアー付き Zig コーディング Agent。 [根拠](https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig)
+- **[Agent](https://github.com/AgentiLoop/Agent)** · 629 Star — TypeSafeKit クライアントを使用した、ネイティブ macOS エージェント内のオプションの JEV コマンド リスク アドバイザ。 [根拠](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
+- **[jev-align](https://github.com/sutro-sh/jev-align)** · 276 Star — GEPA を使用して人間の例から JEV 分類器を調整するための CLI。 [根拠](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
+- **[Abide](https://github.com/coldteadotai/abide)** · 210 Star — 通常のリンターでは表現できない制約を扱う、コーディング Agent 向けルール層。 [根拠](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
+- **[Unclutter](https://github.com/kitze/unclutter)** · 203 Star — JEV を使用して広告、プロモーション、購読ポップアップを識別し、再利用可能な非表示ルールを保存するブラウザ拡張機能。 [根拠](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
+- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 178 Star — Interlinked は、任意の JEV 判断と根拠チェックをローカルのコーディング Agent 検査に追加します。 [根拠](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
+- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 140 Star — 実行前にツールのリスクにフラグを立て、出力のシークレットと障害タイプをチェックする Pi 拡張機能。 [根拠](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
+- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 136 Star — プロジェクト ルール、範囲外のアクション、繰り返される失敗、完了要求のチェックを Pi エージェントに追加します。 [根拠](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
+- **[Supercov](https://github.com/supercorp-ai/supercov)** · 106 Star — JEV で限定的な品質チェックを行うコード品質・カバレッジツール。 [根拠](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 74 Star — 依存なしの JEV クライアントでリポジトリの検出候補を一括評価する意味的コードリンターです。 [根拠](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
- **[Captaincore](https://github.com/CaptainCore/captaincore)** · 71 Star — WordPress ツールキットCaptainCore の JEV コマンドは、構造化された質問に答え、マルウェア スキャナーの検出結果に優先順位を付けてレビューします。 [根拠](https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go)
-- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 70 Star — 依存なしの JEV クライアントでリポジトリの検出候補を一括評価する意味的コードリンターです。 [根拠](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
-- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 55 Star — JEV で意味的な lint ルールを評価する Oxlint プラグイン。 [根拠](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 58 Star — JEV で意味的な lint ルールを評価する Oxlint プラグイン。 [根拠](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Canny · Jev Hooks](https://github.com/qkal/Canny)** · 50 Star — 追記専用台帳を保ち、任意で Jev に Noul 助言チェックを頼む Claude Code / Codex hooks。 [根拠](https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts)
[↑ 検索入口に戻る](#discovery)
-### Agent ワークフロー (14)
+### Agent ワークフロー (15)
-- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 30.9K Star — サブエージェント モデルと推論強度の必要性を分類する Claude Code mod。 [根拠](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
+- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 31.2K Star — サブエージェント モデルと推論強度の必要性を分類する Claude Code mod。 [根拠](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
- **[Openwork](https://github.com/different-ai/openwork)** · 23.7K Star — CI と Skill ワークフローで JEV レビューを実行できる、オープンソースの cowork 風エージェント作業空間。 [根拠](https://github.com/different-ai/openwork/blob/80c74d2160034fa1738fc6be78dd24ddfeccfc59/.github/scripts/jev-test-coverage-review.mjs)
- **[Eve](https://github.com/vercel/eve)** · 5.3K Star — Skill・評価・ツール付きワークフローをまとめる Vercel のオープンなエージェント枠組み。 [根拠](https://github.com/vercel/eve/blob/d004e6d47e9d25d0380c24b5a47b65a18f8b2784/README.md)
- **[RNSkill · JEV Office Gate](https://github.com/Pluviobyte/rnskill)** · 1.6K Star — Agent Skill 集に、限定されたオフィス文書チェック用 JEV ゲートを収録。 [根拠](https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py)
- **[NewsJack](https://github.com/elvisun/newsjack)** · 1.3K Star — ライブ ニュース フィードをスクリーニングしてタイムリーなブランド チャンスを見つけるオープンソースの PR ワークフロー。 [根拠](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer)
-- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 683 Star — コーディング Agent Runtime が JEV レーンでルーティング・圧縮・記憶・ツール結果を処理。 [根拠](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
-- **[Smithers](https://github.com/smithersai/smithers)** · 418 Star — TypeScript 設定でワークフローを定義し、JEV 分類パスを持つエージェント枠組み。 [根拠](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
-- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 173 Star — いつコンテキストを圧縮してトークンを節約するかを助言するコーディング Agent Skill。 [根拠](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
-- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 134 Star — 長い Bash 出力をモデルに渡す前に TypeSafe JEV で刈り込む Claude Code プラグイン。 [根拠](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
-- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 115 Star — JEV 決定を Claude Code および Codex ツール フローに挿入するためのゲートウェイ。 [根拠](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
-- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 109 Star — JEV で現在のセッション文脈に対して Agent Skills を順位付けする Rust CLI。 [根拠](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
+- **[Agent Beacon · Jev Evaluator](https://github.com/asymptote-labs/agent-beacon)** · 1.1K Star — 複数 Harness 向け Agent メモリ。Jev で編集済みセッション投影を採点してから教訓を昇格できます。 [根拠](https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go)
+- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 684 Star — コーディング Agent Runtime が JEV レーンでルーティング・圧縮・記憶・ツール結果を処理。 [根拠](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
+- **[Smithers](https://github.com/smithersai/smithers)** · 420 Star — TypeScript 設定でワークフローを定義し、JEV 分類パスを持つエージェント枠組み。 [根拠](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
+- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 180 Star — いつコンテキストを圧縮してトークンを節約するかを助言するコーディング Agent Skill。 [根拠](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
+- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 179 Star — JEV 決定を Claude Code および Codex ツール フローに挿入するためのゲートウェイ。 [根拠](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
+- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 140 Star — 長い Bash 出力をモデルに渡す前に TypeSafe JEV で刈り込む Claude Code プラグイン。 [根拠](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
+- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 115 Star — DeepSeek Harness と互換コーディング Agent ホスト向けの構造化 JEV 判断プラグイン。 [根拠](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
+- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 114 Star — JEV で現在のセッション文脈に対して Agent Skills を順位付けする Rust CLI。 [根拠](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
- **[Bluenoise](https://github.com/rokcso/bluenoise)** · 90 Star — デフォルトでローカル ルールを使用する X/Twitter フィルタリング拡張機能。オプションで JEV が一致しない返信をチェックします。 [根拠](https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts)
-- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 78 Star — DeepSeek Harness と互換コーディング Agent ホスト向けの構造化 JEV 判断プラグイン。 [根拠](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
-- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 62 Star — 履歴を保持・短縮・削除するか JEV に判断させる文脈クリーナー。 [根拠](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
+- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 66 Star — 履歴を保持・短縮・削除するか JEV に判断させる文脈クリーナー。 [根拠](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
[↑ 検索入口に戻る](#discovery)
@@ -246,15 +261,15 @@ npx skills add BeatAPI/awesome-jev
### UI・自動化 (9)
-- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.0K Star — 事前定義されたコンポーネントとプロパティから選択する JEV UI 構成の実験。 [根拠](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
-- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 1.6K Star — 表示中の会話を読み、候補返信を提案するが自動送信しない Android 会話コパイロット。 [根拠](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
+- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.1K Star — 事前定義されたコンポーネントとプロパティから選択する JEV UI 構成の実験。 [根拠](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
+- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 4.1K Star — 表示中の会話を読み、候補返信を提案するが自動送信しない Android 会話コパイロット。 [根拠](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
- **[Aiavatarkit](https://github.com/uezo/aiavatarkit)** · 678 Star — オプションの AIAvatarKit コンポーネントは、JEV を使用して音声トランスクリプトからターン終了を判断します。 [根拠](https://github.com/uezo/aiavatarkit/blob/38b617b8b9269939734e70ef503d7ea6976acdbd/aiavatar/sts/vad/turn_end_gates/jev.py#L172)
-- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 642 Star — JEV または Gemini Live によって駆動されるローカル macOS コンピューター使用コンパニオン。 [根拠](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
-- **[Notra](https://github.com/usenotra/notra)** · 207 Star — 作業成果を公開コンテンツに変え、Agent Skill を接続するプロダクト。 [根拠](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
+- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 654 Star — JEV または Gemini Live によって駆動されるローカル macOS コンピューター使用コンパニオン。 [根拠](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
+- **[Notra](https://github.com/usenotra/notra)** · 213 Star — 作業成果を公開コンテンツに変え、Agent Skill を接続するプロダクト。 [根拠](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
- **[OpenWhisper](https://github.com/Knuckles92/OpenWhisper)** · 187 Star — オプションの JEV を備えたディクテーションと会議メモのアプリは、トピックの変更、メモを取る指示、機密テキストをチェックします。 [根拠](https://github.com/Knuckles92/OpenWhisper/blob/9e83653df183096104769e302a3c907cb277c551/README.md)
-- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 84 Star — JEV がコード提供の返答・ツール候補だけから選ぶツール利用チャット。 [根拠](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 91 Star — JEV がコード提供の返答・ツール候補だけから選ぶツール利用チャット。 [根拠](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 91 Star — YouTube 拡張機能は、JEV を使用してライブ音声とトランスクリプトからスポンサー付きセグメントを検出し、プロモーション ブロックを自動的にスキップします。 [根拠](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
- **[Jevmeter](https://github.com/ChetasLua/jevmeter)** · 81 Star — 選択したルーブリックに対してトランスクリプト文を評価するよう JEV に依頼することで、スコア メーター付きの編集済みビデオを作成します。 [根拠](https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py)
-- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 81 Star — YouTube 拡張機能は、JEV を使用してライブ音声とトランスクリプトからスポンサー付きセグメントを検出し、プロモーション ブロックを自動的にスキップします。 [根拠](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
[↑ 検索入口に戻る](#discovery)
@@ -262,61 +277,63 @@ npx skills add BeatAPI/awesome-jev
### 開発者ツール (35)
-- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.0K Star — 保持されたテキストをそのまま保持しながら、Claude Code ツール履歴を圧縮します。 [根拠](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
-- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.6K Star — TypeSafe の System One / JEV API 向け公式 Agent Skill 集。 [根拠](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
-- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.0K Star — 公開 JEV プロジェクト、連携、議論をまとめた大規模コミュニティ目录。 [根拠](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
-- **[Celesto](https://github.com/CelestoAI/celesto)** · 958 Star — Celesto PR レビューの例では、サンドボックス チェックを準備し、候補となる結果について一般的なモデルと JEV を比較します。 [根拠](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
-- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 810 Star — ターミナルからコンテンツ生成するための Agent Skill 付き CLI。 [根拠](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
-- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 772 Star — 根拠付きの JEV ユースケース・パターン・プロンプト・スターター集。 [根拠](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
-- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 628 Star — JEV と TypeSafe System One 向けツールのキュレーション目录。 [根拠](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
-- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 502 Star — Git diff またはコードベースを段階的にレビューし、ローカル ダッシュボードにレビュー リードを表示します。 [根拠](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
-- **[Foreman](https://github.com/thruwire/foreman)** · 472 Star — ワーカーの差分、ログ、テストを読み取り、スタック/オフトラック/検証について JEV Nouls に質問し、Python ポリシーを適用する独立したスーパーバイザー ループ。 [根拠](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
-- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 431 Star — TypeSafe・System One・JEV の公式／コミュニティ資源をまとめたキュレーション一覧。 [根拠](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
-- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 431 Star — TypeSafe JEV のプロジェクト・SDK・デモ・評価を根拠付きでまとめたガイド。 [根拠](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
-- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 398 Star — 標準ライブラリだけで動く JEV クライアントを含む Agent スキルとワークフロー集です。 [根拠](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
-- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 334 Star — 根拠付き JEV エコシステムレーダー。平易な発見 UI と GitHub 同期付き。 [根拠](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
-- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 316 Star — JEV / TypeSafe System One で作られたプロジェクトの根拠付き一覧。 [根拠](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
-- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 281 Star — OrchestKit はオプションで JEV を使用してコーディング セッションを分類し、信頼度がしきい値を満たしたときにその色を設定できます。 [根拠](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
-- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 257 Star — JEV アプリ、ライブラリ、System One 資料のキュレーション一覧。 [根拠](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
-- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 245 Star — 証拠チェック、コンテンツ スクリーニング、検索、再ランキング、分類、比較、抽出のための 8 つの MCP ツール。 [根拠](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
-- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 227 Star — Claude Code、Claude Desktop、Codex、Pi から JEV へ型付き質問を送れる MCP サーバー。 [根拠](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
-- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 196 Star — ローカルの MCP コード品質レビュアーが、構造化スコアをコーディング Agent に返します。 [根拠](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
-- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 187 Star — 根拠付き JEV ディレクトリと、再利用可能な JEV 専用 GitHub Review ワークフロー。 [根拠](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
-- **[Perch](https://github.com/lakeday-org/perch)** · 167 Star — 発見を発火すべきかを JEV / System One に問うセマンティック lint。 [根拠](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
-- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 159 Star — System One の論文、オープン再現、独立評価を集めたギャラリー。 [根拠](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
-- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 148 Star — コミュニティ運営の多言語 JEV オープンソース・カタログ。 [根拠](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.3K Star — 保持されたテキストをそのまま保持しながら、Claude Code ツール履歴を圧縮します。 [根拠](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
+- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.9K Star — TypeSafe の System One / JEV API 向け公式 Agent Skill 集。 [根拠](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
+- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.3K Star — 公開 JEV プロジェクト、連携、議論をまとめた大規模コミュニティ目录。 [根拠](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
+- **[Celesto](https://github.com/CelestoAI/celesto)** · 966 Star — Celesto PR レビューの例では、サンドボックス チェックを準備し、候補となる結果について一般的なモデルと JEV を比較します。 [根拠](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
+- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 811 Star — ターミナルからコンテンツ生成するための Agent Skill 付き CLI。 [根拠](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
+- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 806 Star — 根拠付きの JEV ユースケース・パターン・プロンプト・スターター集。 [根拠](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
+- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 658 Star — JEV と TypeSafe System One 向けツールのキュレーション目录。 [根拠](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
+- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 589 Star — コミュニティ運営の多言語 JEV オープンソース・カタログ。 [根拠](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 556 Star — Git diff またはコードベースを段階的にレビューし、ローカル ダッシュボードにレビュー リードを表示します。 [根拠](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
+- **[Foreman](https://github.com/thruwire/foreman)** · 517 Star — ワーカーの差分、ログ、テストを読み取り、スタック/オフトラック/検証について JEV Nouls に質問し、Python ポリシーを適用する独立したスーパーバイザー ループ。 [根拠](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
+- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 476 Star — TypeSafe・System One・JEV の公式／コミュニティ資源をまとめたキュレーション一覧。 [根拠](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
+- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 476 Star — TypeSafe JEV のプロジェクト・SDK・デモ・評価を根拠付きでまとめたガイド。 [根拠](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
+- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 437 Star — 標準ライブラリだけで動く JEV クライアントを含む Agent スキルとワークフロー集です。 [根拠](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
+- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 398 Star — 根拠付き JEV エコシステムレーダー。平易な発見 UI と GitHub 同期付き。 [根拠](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
+- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 354 Star — JEV / TypeSafe System One で作られたプロジェクトの根拠付き一覧。 [根拠](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
+- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 303 Star — JEV アプリ、ライブラリ、System One 資料のキュレーション一覧。 [根拠](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
+- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 290 Star — 証拠チェック、コンテンツ スクリーニング、検索、再ランキング、分類、比較、抽出のための 8 つの MCP ツール。 [根拠](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
+- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 283 Star — OrchestKit はオプションで JEV を使用してコーディング セッションを分類し、信頼度がしきい値を満たしたときにその色を設定できます。 [根拠](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
+- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 262 Star — Claude Code、Claude Desktop、Codex、Pi から JEV へ型付き質問を送れる MCP サーバー。 [根拠](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
+- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 208 Star — System One の論文、オープン再現、独立評価を集めたギャラリー。 [根拠](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
+- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 206 Star — ローカルの MCP コード品質レビュアーが、構造化スコアをコーディング Agent に返します。 [根拠](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
+- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 198 Star — 根拠付き JEV ディレクトリと、再利用可能な JEV 専用 GitHub Review ワークフロー。 [根拠](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
+- **[Perch](https://github.com/lakeday-org/perch)** · 170 Star — 発見を発火すべきかを JEV / System One に問うセマンティック lint。 [根拠](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
+- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 149 Star — awesomejev.com の元カタログで、リポジトリ、連携、デモ、研究を収録。 [根拠](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
+- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 130 Star — 型付きで信頼度を扱う判断を中心にした TypeSafe JEV 資料一覧。 [根拠](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Building with JEV Skill](https://github.com/dbreunig/building-with-jev-skill)** · 128 Star — JEV / System One を呼ぶプログラムの作成・改善を助ける Agent Skill。 [根拠](https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md)
-- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 127 Star — awesomejev.com の元カタログで、リポジトリ、連携、デモ、研究を収録。 [根拠](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
-- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 125 Star — 型付きで信頼度を扱う判断を中心にした TypeSafe JEV 資料一覧。 [根拠](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Awesome JEV (Promethe-us)](https://github.com/Promethe-us/awesome-jev)** · 122 Star — 出典を確認した JEV の資料、プロジェクト、研究、運用情報をまとめた二言語コレクションです。 [根拠](https://github.com/Promethe-us/awesome-jev/blob/636623538eda3c363b8c396880ccdc8aba70d424/README.md)
-- **[Taskuary](https://github.com/ldbumble/taskuary)** · 116 Star — タスク状態に関するユーザー定義の条件をチェックするための Taskuary のオプションの JEV 判定モジュール。 [根拠](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
-- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 111 Star — 専用アダプターを備えたコーディング Agent 向けの限定 JEV ワークフロー。 [根拠](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
-- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 100 Star — JEV のユースケース、プロジェクト、SDK、ツール、学習資料をまとめたガイドです。 [根拠](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
-- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 96 Star — JEV によってランク付けされた Fish スタイルの Zsh 履歴提案ツール。コンテキストに基づいてローカル履歴から候補コマンドを順序付けします。 [根拠](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
-- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 84 Star — 公開済み JEV デモ、リポジトリ、報告値、モデル制約を出典付きで整理したカタログです。 [根拠](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
-- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 73 Star — 実行可能な型付き判断例を含む JEV 資料一覧。 [根拠](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
-- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 71 Star — JEV、オープン判断モデル、分類器、リランカーを比較する再現可能なベンチマークです。 [根拠](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
-- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 59 Star — 厳選プロジェクト、実践チュートリアル、価格、独立した注意点を含む中国語 JEV ガイドです。 [根拠](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
+- **[Taskuary](https://github.com/ldbumble/taskuary)** · 118 Star — タスク状態に関するユーザー定義の条件をチェックするための Taskuary のオプションの JEV 判定モジュール。 [根拠](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
+- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 114 Star — 公開済み JEV デモ、リポジトリ、報告値、モデル制約を出典付きで整理したカタログです。 [根拠](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
+- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 113 Star — JEV のユースケース、プロジェクト、SDK、ツール、学習資料をまとめたガイドです。 [根拠](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
+- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 113 Star — 専用アダプターを備えたコーディング Agent 向けの限定 JEV ワークフロー。 [根拠](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
+- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 104 Star — JEV によってランク付けされた Fish スタイルの Zsh 履歴提案ツール。コンテキストに基づいてローカル履歴から候補コマンドを順序付けします。 [根拠](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
+- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 83 Star — JEV、オープン判断モデル、分類器、リランカーを比較する再現可能なベンチマークです。 [根拠](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
+- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 78 Star — 実行可能な型付き判断例を含む JEV 資料一覧。 [根拠](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
+- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 60 Star — 厳選プロジェクト、実践チュートリアル、価格、独立した注意点を含む中国語 JEV ガイドです。 [根拠](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
[↑ 検索入口に戻る](#discovery)
-### 業務特化ツール (13)
+### 業務特化ツール (15)
- **[AI Hedge Fund · JEV Adapter](https://github.com/virattt/ai-hedge-fund)** · 63.7K Star — 構造化された戦略判断のためのオプションの JEV アダプターを備えた教育用ヘッジファンド プロトタイプ。 [根拠](https://github.com/virattt/ai-hedge-fund/blob/154a8b2f46dca0f40764d814e4e747b0ad71f4c4/hedge_fund/llm/client.py)
- **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · 12.0K Star — 選択されたライブエントリーの前に JEV 決定ゲートを備えたオープンソースの取引 OS。 [根拠](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py)
-- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 1.9K Star — モナドブロックごとに 1 つの JEV 取引決定を要求できるマーケットメイク実験。 [根拠](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
-- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 485 Star — プランナーが目標を定め、JEV が合法なゲーム操作を選ぶ Minecraft Agent。 [根拠](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
-- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 338 Star — JEV 構造化エミュレータ RAM とスクリーンショットの代わりにテレメトリを提供する実験的な NES マリオ コントローラ。 [根拠](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
-- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 170 Star — JEV、ローカルモデル、汎用モデルの判断ポリシーを比較する身体化 Agent 評価基盤です。 [根拠](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
-- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 155 Star — JEV がローカルで生成されたパスと速度の中から選択するブラウザー運転シミュレーター。 [根拠](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
-- **[Dasheng](https://github.com/wquguru/dasheng)** · 119 Star — ストリーミング ASR と単語ごとの JEV 判断を組み合わせた音読練習アプリ。 [根拠](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
-- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 119 Star — JEV 戦術的なアドバイスのためにカメラ バッファーからシーンの特徴を導出する MuJoCo ドローン実験。 [根拠](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
-- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 103 Star — 標準 LLM ツールの選択と 100 個のモック ツールにわたる JEV ルーティングを比較するエージェント ツール評価ハーネス。 [根拠](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
+- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 2.1K Star — モナドブロックごとに 1 つの JEV 取引決定を要求できるマーケットメイク実験。 [根拠](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
+- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 524 Star — プランナーが目標を定め、JEV が合法なゲーム操作を選ぶ Minecraft Agent。 [根拠](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
+- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 355 Star — JEV 構造化エミュレータ RAM とスクリーンショットの代わりにテレメトリを提供する実験的な NES マリオ コントローラ。 [根拠](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
+- **[quackd · Jev Stepper](https://github.com/rokbenko/quackd)** · 229 Star — 許可済みの離散ロボット呼び出しから選ぶ任意 System One ステッパー付きロボット制御 CLI。 [根拠](https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py)
+- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 191 Star — JEV、ローカルモデル、汎用モデルの判断ポリシーを比較する身体化 Agent 評価基盤です。 [根拠](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
+- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 169 Star — JEV がローカルで生成されたパスと速度の中から選択するブラウザー運転シミュレーター。 [根拠](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
+- **[Dasheng](https://github.com/wquguru/dasheng)** · 135 Star — ストリーミング ASR と単語ごとの JEV 判断を組み合わせた音読練習アプリ。 [根拠](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
+- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 134 Star — JEV 戦術的なアドバイスのためにカメラ バッファーからシーンの特徴を導出する MuJoCo ドローン実験。 [根拠](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
+- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 110 Star — JEV を限定判断レイヤーに使う Hyperliquid ライブ取引 Agent。 [根拠](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
+- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 105 Star — 標準 LLM ツールの選択と 100 個のモック ツールにわたる JEV ルーティングを比較するエージェント ツール評価ハーネス。 [根拠](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
- **[Formanator](https://github.com/timrogers/formanator)** · 99 Star — 任意の JEV レシート分類を備えた福利厚生申請 CLI / MCP クライアント。 [根拠](https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs)
-- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 95 Star — JEV を限定判断レイヤーに使う Hyperliquid ライブ取引 Agent。 [根拠](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
-- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 69 Star — リバランス判断サービスで JEV を使う自律流動性 Agent。 [根拠](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 79 Star — リバランス判断サービスで JEV を使う自律流動性 Agent。 [根拠](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[HA-Jev](https://github.com/AboveColin/HA-Jev)** · 50 Star — 家の状態への型付き Jev 質問をセンサーと自動化アクションにする Home Assistant 連携。 [根拠](https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py)
[↑ 検索入口に戻る](#discovery)
diff --git a/README.md b/README.md
index c872682..2fa4de8 100644
--- a/README.md
+++ b/README.md
@@ -1,14 +1,22 @@
+
+
+
+
+
Awesome JEV
Find JEV projects for your use case — see what they do and which implementation to learn from.
+Want to use JEV in your own app? Get started with the free JEV API on BeatAPI →
+
Find a project · Featured project gallery · Search with your agent · All projects · Search online
English · 简体中文 · 日本語
-183 projects · 10 project types · Updated 2026-09-22
50+ GitHub stars per repository · Source-reviewed
+192 projects · 10 project types · Updated 2026-09-23
50+ GitHub stars per repository · Source-reviewed
@@ -28,13 +36,13 @@ Choose what you want to build. Each guide points to relevant projects and the sp
| | |
| --- | --- |
| [](https://github.com/browser-use/jev-ultrafast) | [](https://github.com/BerriAI/litellm) |
-| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · Browser & computer use · 16.3K stars
Chooses a browser action and matching DOM target in one JEV request.
[Original case · 2.9M views](https://x.com/gregpr07/status/2100411066966749359) · [Fixed-commit evidence](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · Routing & optimization · 59.4K stars
Classifies request complexity before backend model routing.
[Fixed-commit evidence](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
+| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · Browser & computer use · 18.2K stars
Chooses a browser action and matching DOM target in one JEV request.
[Original case · 2.9M views](https://x.com/gregpr07/status/2100411066966749359) · [Fixed-commit evidence](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · Routing & optimization · 59.4K stars
Classifies request complexity before backend model routing.
[Fixed-commit evidence](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
| [](https://github.com/langchain-ai/langchain) | [](https://github.com/vercel/ai) |
-| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK integration · 146.8K stars
Exposes binary, classification, and ordinal JEV decisions inside Python workflows.
[Fixed-commit evidence](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK integration · 26.9K stars
Maps choice, score, and yes/no questions onto a unified evaluate interface.
[Fixed-commit evidence](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
+| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK integration · 146.9K stars
Exposes binary, classification, and ordinal JEV decisions inside Python workflows.
[Fixed-commit evidence](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK integration · 26.9K stars
Maps choice, score, and yes/no questions onto a unified evaluate interface.
[Fixed-commit evidence](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
| [](https://github.com/NandhaKishorM/laya) | [](https://github.com/can1357/jegrep) |
-| **[Laya](https://github.com/NandhaKishorM/laya)** · Open model · 11.5K stars
Multilingual System 1 typed decisions in one forward pass, with a checkpoint router.
[Hugging Face model](https://huggingface.co/convaiinnovations/laya) · [Fixed-commit evidence](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)| **[jegrep](https://github.com/can1357/jegrep)** · Search & data
Searches live code trees semantically without building an index.
[Fixed-commit evidence](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs) |
+| **[Laya](https://github.com/NandhaKishorM/laya)** · Open model · 17.1K stars
Multilingual System 1 typed decisions in one forward pass, with a checkpoint router.
[Hugging Face model](https://huggingface.co/convaiinnovations/laya) · [Fixed-commit evidence](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)| **[jegrep](https://github.com/can1357/jegrep)** · Search & data
Searches live code trees semantically without building an index.
[Fixed-commit evidence](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs) |
| [](https://github.com/trycua/cua) | [](https://github.com/agentgateway/agentgateway) |
-| **[Cua · JEV Use](https://github.com/trycua/cua)** · Browser & computer use · 25.8K stars
Selects an action ID from DOM or visual-region descriptions.
[Fixed-commit evidence](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · Safety & review · 5.0K stars
Scores jailbreak, harmful-content, and secret-leakage checks at the gateway.
[Fixed-commit evidence](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
+| **[Cua · JEV Use](https://github.com/trycua/cua)** · Browser & computer use · 26.0K stars
Selects an action ID from DOM or visual-region descriptions.
[Fixed-commit evidence](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · Safety & review · 5.0K stars
Scores jailbreak, harmful-content, and secret-leakage checks at the gateway.
[Fixed-commit evidence](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
| [](https://github.com/elvisun/newsjack) | [](https://github.com/OpenByteInc/QuantDinger) |
| **[NewsJack](https://github.com/elvisun/newsjack)** · Agent workflow · 1.3K stars
Filters hundreds of live news items before an agent handles the selected opportunities.
[Original case · 583K views](https://x.com/elvissun/status/2100951347080421409) · [Fixed-commit evidence](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer) | **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · Domain tool · 12.0K stars
Adds an evidence and risk gate before selected live-trading entries.
[Fixed-commit evidence](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py) |
@@ -60,49 +68,50 @@ npx skills add BeatAPI/awesome-jev
-## All 183 projects
+## All 192 projects
Browse the full catalogue by project type. Use the scenario guides above when you have a specific task in mind.
-[Browser & computer use · 16](#browser-computer-use) · [SDK integrations · 30](#sdk-integrations) · [Routing & optimization · 16](#routing-optimization) · [Open models · 21](#open-models) · [Search & data · 15](#search-data)
-[Safety & review · 14](#safety-review) · [Agent workflows · 14](#agent-workflows) · [Interfaces & automation · 9](#interfaces) · [Developer tools · 35](#developer-tools) · [Domain tools · 13](#domain-tools)
+[Browser & computer use · 17](#browser-computer-use) · [SDK integrations · 31](#sdk-integrations) · [Routing & optimization · 16](#routing-optimization) · [Open models · 23](#open-models) · [Search & data · 15](#search-data)
+[Safety & review · 16](#safety-review) · [Agent workflows · 15](#agent-workflows) · [Interfaces & automation · 9](#interfaces) · [Developer tools · 35](#developer-tools) · [Domain tools · 15](#domain-tools)
-### Browser & computer use (16)
+### Browser & computer use (17)
-- **[Cua · JEV Use](https://github.com/trycua/cua)** · 25.8K stars — A computer-use example pairing Cua Driver observation and execution with bounded JEV choices. [Source](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
-- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 16.3K 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](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
-- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.4K stars — An optional JEV skill that chooses native desktop controls and actions from accessibility data. [Source](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
-- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 765 stars — Builds candidate actions from OCR and UI state for Jev to control macOS, calling a text model only when writing is needed. [Source](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
-- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 549 stars — Computer-use helpers that call JEV to decide the next UI action. [Source](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
+- **[Cua · JEV Use](https://github.com/trycua/cua)** · 26.0K stars — A computer-use example pairing Cua Driver observation and execution with bounded JEV choices. [Source](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
+- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 18.2K 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](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
+- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.5K stars — An optional JEV skill that chooses native desktop controls and actions from accessibility data. [Source](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
+- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 843 stars — Builds candidate actions from OCR and UI state for Jev to control macOS, calling a text model only when writing is needed. [Source](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
+- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 581 stars — Computer-use helpers that call JEV to decide the next UI action. [Source](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
- **[Omg.dev](https://github.com/BennyKok/omg.dev)** · 535 stars — An omg.dev mobile testing script can use Jev to read the accessibility tree and choose the next interaction. [Source](https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts)
-- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 334 stars — A Skill for Codex browser workflows: Jev chooses navigation, clicks, and scrolling while Codex handles text input and final checks. [Source](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
-- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 329 stars — Controls an Android phone through Mobilerun, with a web studio and CLI showing Jev decisions. [Source](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
-- **[Third Hand](https://github.com/shhivv/third-hand)** · 283 stars — A Swift client stack that calls JEV while driving local computer-use style helpers. [Source](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
-- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 231 stars — Drives a browser from a task and URL, returning the final page, screenshot and action trace. [Source](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
-- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 216 stars — Controls a Playwright browser by sending incremental speech transcripts to Jev. [Source](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
-- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 127 stars — A fast computer-use action layer with a TypeSafe policy implementation. [Source](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
-- **[JEV Use](https://github.com/savka777/jev-use)** · 86 stars — A macOS computer-use harness that drives Accessibility controls without a vision model. [Source](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
-- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 76 stars — A browser-agent benchmark harness with frozen JEV experiments across accessibility and WebMCP interfaces. [Source](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
-- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 68 stars — An experimental Chrome extension that asks Jev whether candidate DOM elements are ads, then highlights or removes them. [Source](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
-- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 67 stars — A browser automation library, CLI, and MCP server pairing LLM plans with JEV actions. [Source](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 387 stars — A Skill for Codex browser workflows: Jev chooses navigation, clicks, and scrolling while Codex handles text input and final checks. [Source](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
+- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 356 stars — Controls an Android phone through Mobilerun, with a web studio and CLI showing Jev decisions. [Source](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
+- **[Third Hand](https://github.com/shhivv/third-hand)** · 288 stars — A Swift client stack that calls JEV while driving local computer-use style helpers. [Source](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
+- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 247 stars — Controls a Playwright browser by sending incremental speech transcripts to Jev. [Source](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
+- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 240 stars — Drives a browser from a task and URL, returning the final page, screenshot and action trace. [Source](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
+- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 128 stars — A fast computer-use action layer with a TypeSafe policy implementation. [Source](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
+- **[fastbrowse · Jev](https://github.com/agent-labs-dev/fastbrowse)** · 97 stars — A browser agent that indexes page controls and lets Jev pick the next action while an LLM plans and reads. [Source](https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py)
+- **[JEV Use](https://github.com/savka777/jev-use)** · 92 stars — A macOS computer-use harness that drives Accessibility controls without a vision model. [Source](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
+- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 79 stars — A browser-agent benchmark harness with frozen JEV experiments across accessibility and WebMCP interfaces. [Source](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
+- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 76 stars — A browser automation library, CLI, and MCP server pairing LLM plans with JEV actions. [Source](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 71 stars — An experimental Chrome extension that asks Jev whether candidate DOM elements are ads, then highlights or removes them. [Source](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
[↑ Back to discovery](#discovery)
-### SDK integrations (30)
+### SDK integrations (31)
-- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.8K stars — An optional JEV classifier integration for Python LangChain workflows. [Source](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
+- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.9K stars — An optional JEV classifier integration for Python LangChain workflows. [Source](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
- **[Composio · TypeSafe Provider](https://github.com/ComposioHQ/composio)** · 30.3K stars — A TypeSafe provider that uses JEV to choose among tools and bounded argument options. [Source](https://github.com/ComposioHQ/composio/blob/4b5920bf7aa55c8a44657b060d4bd25ce7b13a9a/ts/packages/providers/typesafe/src/decide.ts)
- **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · 26.9K stars — The AI SDK TypeSafe provider exposes JEV through the shared evaluate interface. [Source](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts)
- **[Pydantic AI · TypeSafe](https://github.com/pydantic/pydantic-ai)** · 20.1K stars — A TypeSafe model integration that maps supported structured outputs to JEV questions. [Source](https://github.com/pydantic/pydantic-ai/blob/c4898abb54dc25ae6f6aef208a4c0661b30a455e/pydantic_ai_slim/pydantic_ai/models/typesafe.py)
- **[Eliza · TypeSafe Client](https://github.com/elizaOS/eliza)** · 19.4K stars — An optional TypeSafe HTTP adapter in the Eliza agent operating system. [Source](https://github.com/elizaOS/eliza/blob/ebc808e3a67fb941e29153d89fc896524d32fe3c/packages/agent/src/services/typesafe/client.ts)
- **[LangChain.js · TypeSafe](https://github.com/langchain-ai/langchainjs)** · 18.2K stars — A TypeSafeClassifier integration for JavaScript LangChain workflows. [Source](https://github.com/langchain-ai/langchainjs/blob/206d8b992bcf90ce7d46f2158f1ad85fc1d826c0/libs/providers/langchain-typesafe/src/classifier.ts)
- **[Rig · TypeSafe Crate](https://github.com/0xPlaygrounds/rig)** · 8.7K stars — An experimental Rust crate for expressing JEV questions and answers with Rust types. [Source](https://github.com/0xPlaygrounds/rig/blob/2d16c1b25f6749b3a2cd841beddf767106495069/crates/rig-typesafeai/src/wire.rs#L40)
-- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.2K stars — Bifrost exposes TypeSafe as a native provider through its shared decision operation. [Source](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
-- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 6.9K stars — A self-hosted AI gateway with a first-class JEV channel and native Decisions routes. [Source](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
+- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.3K stars — Bifrost exposes TypeSafe as a native provider through its shared decision operation. [Source](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
+- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 7.0K stars — A self-hosted AI gateway with a first-class JEV channel and native Decisions routes. [Source](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
- **[Open Connector · TypeSafe](https://github.com/oomol-lab/open-connector)** · 5.9K stars — Open Connector exposes TypeSafe evaluation through its provider runtime. [Source](https://github.com/oomol-lab/open-connector/blob/4e6d8533ae436cba4781b6fb69d1bf31e5fb5ffe/src/providers/typesafe_ai/runtime.ts)
- **[Kiln · JEV Adapter](https://github.com/Kiln-AI/Kiln)** · 5.1K stars — Kiln maps compatible JSON schemas to JEV questions for structured single-turn tasks. [Source](https://github.com/Kiln-AI/Kiln/blob/5992e2e51ec6782fce33f277145954ba3ae8467e/libs/core/kiln_ai/adapters/model_adapters/jev_adapter.py)
- **[Ax · TypeSafe Client](https://github.com/ax-llm/ax)** · 2.9K stars — A TypeSafe integration for boolean, finite-class, and native JEV signatures. [Source](https://github.com/ax-llm/ax/blob/5c43344f9ef3016db576fa2c3b59d48ef21b4d71/src/ax/ai/typesafe/client.ts#L1)
@@ -110,20 +119,21 @@ Browse the full catalogue by project type. Use the scenario guides above when yo
- **[Vellum Assistant · JEV](https://github.com/vellum-ai/vellum-assistant)** · 1.3K stars — An optional JEV provider that submits conversation state and explicit questions. [Source](https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts)
- **[Laravel AI · TypeSafe](https://github.com/laravel/ai)** · 1.2K stars — Laravel AI includes a native TypeSafe classification provider backed by JEV. [Source](https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php)
- **[Req LLM](https://github.com/agentjido/req_llm)** · 581 stars — A TypeSafe provider for calling Jev through ReqLLM’s evaluate interface in Elixir. [Source](https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md)
-- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 458 stars — Adapter turning open LLM endpoints into Jev-compatible classification services without training a separate classifier head. [Source](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
-- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 406 stars — A zero-shot classification service with JEV as its primary typed-decision backend. [Source](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
+- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 485 stars — Adapter turning open LLM endpoints into Jev-compatible classification services without training a separate classifier head. [Source](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
+- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 411 stars — A zero-shot classification service with JEV as its primary typed-decision backend. [Source](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
- **[Instructor Php](https://github.com/cognesy/instructor-php)** · 327 stars — A TypeSafe Decision driver within Instructor PHP’s Polyglot module. [Source](https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md)
-- **[Openjev](https://github.com/razorback16/openjev)** · 286 stars — An independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model. [Source](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
-- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 248 stars — A dedicated TypeSafe module in a Scala client that supports multiple AI providers. [Source](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
-- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 244 stars — Pi’s programmable runtime includes an optional Jev loop for observing state, making decisions and running bounded actions. [Source](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
-- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 243 stars — A drop-in TypeSafeClient replacement that backs System One calls with ordinary LLM APIs. [Source](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
-- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 214 stars — The JavaScript and TypeScript SDK published by TypeSafe, with typed Jev requests and answers. [Source](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
-- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 188 stars — Official TypeSafe Python SDK with synchronous and asynchronous clients for Jev System One, plus question and answer types. [Source](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[Openjev](https://github.com/razorback16/openjev)** · 319 stars — An independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model. [Source](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
+- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 266 stars — A drop-in TypeSafeClient replacement that backs System One calls with ordinary LLM APIs. [Source](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
+- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 253 stars — Pi’s programmable runtime includes an optional Jev loop for observing state, making decisions and running bounded actions. [Source](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
+- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 249 stars — A dedicated TypeSafe module in a Scala client that supports multiple AI providers. [Source](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
+- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 226 stars — The JavaScript and TypeScript SDK published by TypeSafe, with typed Jev requests and answers. [Source](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
+- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 205 stars — Official TypeSafe Python SDK with synchronous and asynchronous clients for Jev System One, plus question and answer types. [Source](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[AI SDK Python · Jev Evaluate](https://github.com/vercel-labs/ai-python)** · 185 stars — Vercel's AI SDK for Python with an experimental evaluate path that asks typed questions through typesafe-ai/jev. [Source](https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py)
- **[Runline](https://github.com/Michaelliv/runline)** · 163 stars — A TypeSafe plugin exposing Jev decisions as callable actions in Runline Agent JavaScript. [Source](https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md)
-- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 137 stars — NeuroLink ships a TypeSafe provider and classifier-routing support for JEV. [Source](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
+- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 138 stars — NeuroLink ships a TypeSafe provider and classifier-routing support for JEV. [Source](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
- **[AI](https://github.com/hackclub/ai)** · 133 stars — A Jev forwarding endpoint in the Hack Club AI proxy, using its authentication, limits and usage logging. [Source](https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md)
- **[Effect Agent](https://github.com/danieljvdm/effect-agent)** · 121 stars — An Effect Agent TypeSafe decision provider for typed question sets and optional model selection. [Source](https://github.com/danieljvdm/effect-agent/blob/88005e497e9b627eeb16d670f278903c57601da9/README.md)
-- **[Advocaat](https://github.com/pithings/advocaat)** · 89 stars — A small TypeScript client for asking Jev multiple typed questions about the same data. [Source](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
+- **[Advocaat](https://github.com/pithings/advocaat)** · 90 stars — A small TypeScript client for asking Jev multiple typed questions about the same data. [Source](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
[↑ Back to discovery](#discovery)
@@ -132,49 +142,51 @@ Browse the full catalogue by project type. Use the scenario guides above when yo
### Routing & optimization (16)
- **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · 59.4K stars — LiteLLM can use JEV inside its complexity-based model router. [Source](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70)
-- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.4K stars — A coding agent with an optional TypeSafe judgment provider for bounded workflow decisions. [Source](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
-- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.2K stars — An optional automatic model router that classifies a message before model selection. [Source](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
-- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 6.9K stars — An agent crew dispatcher that can match task briefs to rules with JEV. [Source](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
+- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.8K stars — A coding agent with an optional TypeSafe judgment provider for bounded workflow decisions. [Source](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
+- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.3K stars — An optional automatic model router that classifies a message before model selection. [Source](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
+- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 7.0K stars — An agent crew dispatcher that can match task briefs to rules with JEV. [Source](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
- **[Agent Router · TypeSafe](https://github.com/theagentrouter/agent-router)** · 2.1K stars — Agent Router provides a native TypeSafe System One translator in its Envoy-based gateway. [Source](https://github.com/theagentrouter/agent-router/blob/7d7c07ffdb14241362def2f7d3f76cbd06d518cc/internal/translator/typesafe_systemone.go)
- **[LLMGateway · System One](https://github.com/theopenco/llmgateway)** · 1.7K stars — LLMGateway implements a native System One route with typed schemas and provider mappings. [Source](https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts)
-- **[Atomic](https://github.com/bastani-inc/atomic)** · 812 stars — An optional Jev decision backend in the Atomic coding Agent for bounded structured choices such as routing. [Source](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Atomic](https://github.com/bastani-inc/atomic)** · 814 stars — An optional Jev decision backend in the Atomic coding Agent for bounded structured choices such as routing. [Source](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 600 stars — JEV-powered routing, memory, compaction, skill selection, and computer/browser use for Hermes agents. [Source](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
- **[Vexjoy Agent](https://github.com/notque/vexjoy-agent)** · 423 stars — An optional Jev routing path that matches VexJoy requests to specialist Agents, skills and workflows. [Source](https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md)
-- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 401 stars — JEV-powered routing, memory, compaction, skill selection, and computer/browser use for Hermes agents. [Source](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
-- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 331 stars — An optional Jev dispatch classifier for choosing among WrongStack specialist Agents. [Source](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
-- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 316 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](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
-- **[Skillbox](https://github.com/kitze/skillbox)** · 223 stars — A self-hosted, versioned Agent skills library with optional Jev recommendations. [Source](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
-- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 184 stars — Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode. [Source](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
-- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 145 stars — A lightweight JEV router for models, tools, and subagents. [Source](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
-- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 74 stars — A JEV decision layer for Grok Bot with usage gates and skill templates. [Source](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
-- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 63 stars — A CLI that routes tasks to coding agents, models, and reasoning effort with JEV. [Source](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
+- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 349 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](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
+- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 332 stars — An optional Jev dispatch classifier for choosing among WrongStack specialist Agents. [Source](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
+- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 229 stars — Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode. [Source](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
+- **[Skillbox](https://github.com/kitze/skillbox)** · 229 stars — A self-hosted, versioned Agent skills library with optional Jev recommendations. [Source](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
+- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 173 stars — A lightweight JEV router for models, tools, and subagents. [Source](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
+- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 77 stars — A JEV decision layer for Grok Bot with usage gates and skill templates. [Source](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
+- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 68 stars — A CLI that routes tasks to coding agents, models, and reasoning effort with JEV. [Source](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
[↑ Back to discovery](#discovery)
-### Open models (21)
+### Open models (23)
-- **[Laya](https://github.com/NandhaKishorM/laya)** · 11.5K stars — Multilingual non-autoregressive System 1 decision engine: typed choice/score/noul in one forward pass, with a router across checkpoints. [Source](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
-- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.4K stars — An independent open-model implementation of the semantic-if interface pattern. [Source](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
-- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 2.6K stars — A tiny trainable Jev-like decision model based on Qwen2.5-0.5B. [Source](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
-- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 1.9K stars — A 0.6B open replica of the JEV interface with parallel decisions and a training pipeline. [Source](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
-- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.5K stars — An open toolkit for local typed decisions, data curation, and evaluation. [Source](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
+- **[Laya](https://github.com/NandhaKishorM/laya)** · 17.1K stars — Multilingual non-autoregressive System 1 decision engine: typed choice/score/noul in one forward pass, with a router across checkpoints. [Source](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
+- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 4.5K stars — A tiny trainable Jev-like decision model based on Qwen2.5-0.5B. [Source](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
+- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.9K stars — An independent open-model implementation of the semantic-if interface pattern. [Source](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
+- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 2.0K stars — A 0.6B open replica of the JEV interface with parallel decisions and a training pipeline. [Source](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
+- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.6K stars — An open toolkit for local typed decisions, data curation, and evaluation. [Source](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
- **[Jevlike](https://github.com/vinnylarouge/jevlike)** · 1.2K stars — An independent starter model that scores a changing list of text or visual options in one pass. [Source](https://github.com/vinnylarouge/jevlike/blob/94f5fd1b0b11d52bbdfdf4e0ee6aa96b568f8452/README.md)
-- **[LocalJev](https://github.com/githubnext/localjev)** · 704 stars — A local Jev-compatible server with benchmarks across small local models. [Source](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
-- **[Splash](https://github.com/incoai/splash)** · 578 stars — A local Apple Silicon inference engine oriented around decision-style models. [Source](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
-- **[Von](https://github.com/wfzyx/von)** · 374 stars — An open, local System One decision model with a JEV-compatible interface. [Source](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
-- **[Decider](https://github.com/Mapika/decider)** · 279 stars — One-pass typed decisions with calibrated probabilities, fine-tuned from Qwen3.5-2B. [Source](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
-- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 273 stars — An independent local decision model exposing a TypeSafe-compatible System One interface. [Source](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
-- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 258 stars — A JEV-compatible API endpoint built on open models with a prefill-only serving path. [Source](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
-- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 246 stars — A calibrated ~151M non-autoregressive local decision engine aiming at JEV-style typed answers. [Source](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
-- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 221 stars — An educational JEV-like visual inference experiment on Apple Silicon with shared context scoring. [Source](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
-- **[Jeff](https://github.com/logan-markewich/jeff)** · 205 stars — A self-hosted Jev-compatible service powered by GliFormer. [Source](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
-- **[Laya Node Runtime](https://github.com/receptron/laya)** · 195 stars — A Node.js and TypeScript runtime for the open Laya JEV-compatible decision model. [Source](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
-- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 168 stars — A DGX Spark container recipe for running DiffusionGemma NVFP4 structured decisions. [Source](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
-- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 125 stars — An adapter that turns local language models into JEV-compatible structured decision engines. [Source](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
-- **[Reflex](https://github.com/kshetrajna12/reflex)** · 110 stars — A small open decision model that recreates the JEV/System One interface on Qwen. [Source](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
-- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 91 stars — An open JEV-style implementation with custom fine-tuning support. [Source](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
-- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 67 stars — An open-source JEV-compatible implementation for typed decision requests. [Source](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[LocalJev](https://github.com/githubnext/localjev)** · 731 stars — A local Jev-compatible server with benchmarks across small local models. [Source](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
+- **[Splash](https://github.com/incoai/splash)** · 638 stars — A local Apple Silicon inference engine oriented around decision-style models. [Source](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
+- **[Von](https://github.com/wfzyx/von)** · 498 stars — An open, local System One decision model with a JEV-compatible interface. [Source](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
+- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 348 stars — An independent local decision model exposing a TypeSafe-compatible System One interface. [Source](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
+- **[Decider](https://github.com/Mapika/decider)** · 316 stars — One-pass typed decisions with calibrated probabilities, fine-tuned from Qwen3.5-2B. [Source](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
+- **[Laya Node Runtime](https://github.com/receptron/laya)** · 295 stars — A Node.js and TypeScript runtime for the open Laya JEV-compatible decision model. [Source](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
+- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 272 stars — A JEV-compatible API endpoint built on open models with a prefill-only serving path. [Source](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
+- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 268 stars — A calibrated ~151M non-autoregressive local decision engine aiming at JEV-style typed answers. [Source](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
+- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 243 stars — An educational JEV-like visual inference experiment on Apple Silicon with shared context scoring. [Source](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
+- **[Jeff](https://github.com/logan-markewich/jeff)** · 219 stars — A self-hosted Jev-compatible service powered by GliFormer. [Source](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
+- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 176 stars — A DGX Spark container recipe for running DiffusionGemma NVFP4 structured decisions. [Source](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
+- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 142 stars — An adapter that turns local language models into JEV-compatible structured decision engines. [Source](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
+- **[Reflex](https://github.com/kshetrajna12/reflex)** · 122 stars — A small open decision model that recreates the JEV/System One interface on Qwen. [Source](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
+- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 103 stars — An open JEV-style implementation with custom fine-tuning support. [Source](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
+- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 101 stars — An open-source JEV-compatible implementation for typed decision requests. [Source](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[jevmlx](https://github.com/bnsd55/jevmlx)** · 59 stars — A local Apple Silicon decision layer that scores constrained fields from MLX logits and speaks System One HTTP. [Source](https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py)
+- **[Open Alternative to Jev](https://github.com/ikermoel/open-alternative-jev)** · 51 stars — An open System One-style decider that packs typed questions into one forward pass over local open models. [Source](https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py)
[↑ Back to discovery](#discovery)
@@ -182,63 +194,66 @@ Browse the full catalogue by project type. Use the scenario guides above when yo
### Search & data (15)
-- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.4K stars — OpenViking can use JEV as a calibrated reranker for agent memory and context retrieval. [Source](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
-- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 24.8K stars — An agent-memory system with a TypeSafe reranker that can prune irrelevant recall candidates. [Source](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
+- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.5K stars — OpenViking can use JEV as a calibrated reranker for agent memory and context retrieval. [Source](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
+- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 25.3K stars — An agent-memory system with a TypeSafe reranker that can prune irrelevant recall candidates. [Source](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
- **[GreptimeDB · JEV SQL](https://github.com/GreptimeTeam/greptimedb)** · 6.7K stars — GreptimeDB adds an experimental SQL predicate backed by JEV judgments. [Source](https://github.com/GreptimeTeam/greptimedb/blob/b5199bc59a199875187d808158357a33b09e5f60/src/common/function/src/scalars/jev.rs)
-- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 752 stars — A local-first agent-memory library with an opt-in JEV reranker for recalled memories. [Source](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
-- **[Kody](https://github.com/kentcdodds/kody)** · 663 stars — Optional second-stage search: widen the hybrid pool, then Score-rerank candidates with Workers AI typesafe/jev. [Source](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
-- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 385 stars — Searches the web with Jev choosing sources and time ranges, then ranking returned links. [Source](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
-- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 358 stars — Small Jev experiments including intent search and agent-assist flows. [Source](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
-- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 348 stars — Classifies extracted tax-document pages against a fixed IRS form catalogue. [Source](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
-- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 286 stars — Adds natural-language filtering, classification, and ranking of rows to PostgreSQL queries. [Source](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
-- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 202 stars — A command-line ranking tool that uses JEV to find relevant items in large collections. [Source](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
-- **[DocJEV](https://github.com/jerryjliu/docjev)** · 201 stars — A document classifier and splitter with a dedicated JEV engine. [Source](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
-- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 124 stars — Meaning-based grep with multilingual AND, OR, NOT, and probability thresholds. [Source](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
-- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 81 stars — A Neo4j graph navigator that uses JEV to classify the next relationship. [Source](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
-- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 81 stars — A pre-alpha PostgreSQL C extension for calling Jev from SQL for classification, yes/no judgments, and scoring. [Source](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
-- **[jegrep](https://github.com/can1357/jegrep)** · 75 stars — Semantic grep for live code trees: describe what you need, get files and original line ranges without building an index. [Source](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 756 stars — A local-first agent-memory library with an opt-in JEV reranker for recalled memories. [Source](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
+- **[Kody](https://github.com/kentcdodds/kody)** · 666 stars — Optional second-stage search: widen the hybrid pool, then Score-rerank candidates with Workers AI typesafe/jev. [Source](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
+- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 416 stars — Searches the web with Jev choosing sources and time ranges, then ranking returned links. [Source](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
+- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 399 stars — Classifies extracted tax-document pages against a fixed IRS form catalogue. [Source](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
+- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 370 stars — Small Jev experiments including intent search and agent-assist flows. [Source](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
+- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 314 stars — Adds natural-language filtering, classification, and ranking of rows to PostgreSQL queries. [Source](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
+- **[DocJEV](https://github.com/jerryjliu/docjev)** · 287 stars — A document classifier and splitter with a dedicated JEV engine. [Source](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
+- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 209 stars — A command-line ranking tool that uses JEV to find relevant items in large collections. [Source](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
+- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 130 stars — Meaning-based grep with multilingual AND, OR, NOT, and probability thresholds. [Source](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
+- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 100 stars — A Neo4j graph navigator that uses JEV to classify the next relationship. [Source](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
+- **[jegrep](https://github.com/can1357/jegrep)** · 83 stars — Semantic grep for live code trees: describe what you need, get files and original line ranges without building an index. [Source](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 82 stars — A pre-alpha PostgreSQL C extension for calling Jev from SQL for classification, yes/no judgments, and scoring. [Source](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
[↑ Back to discovery](#discovery)
-### Safety & review (14)
+### Safety & review (16)
-- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.3K stars — A subscription-to-API gateway with a native TypeSafe client for content moderation. [Source](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
+- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.4K stars — A subscription-to-API gateway with a native TypeSafe client for content moderation. [Source](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
- **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 5.0K stars — A JEV webhook guardrail example for inspecting model requests and responses. [Source](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts)
- **[Latitude · JEV Preclassifier](https://github.com/latitude-dev/latitude-llm)** · 4.7K stars — An optional JEV preclassifier for selecting which conversation checks should run. [Source](https://github.com/latitude-dev/latitude-llm/blob/6b2484c1d74973877fc7d353b8867c7496568d78/packages/domain/flaggers/src/use-cases/run-jev-preclassifier.ts#L1)
-- **[Agent](https://github.com/AgentiLoop/Agent)** · 622 stars — An optional Jev command-risk advisor inside a native macOS Agent, with a TypeSafeKit client. [Source](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
-- **[jev-align](https://github.com/sutro-sh/jev-align)** · 271 stars — A CLI for calibrating Jev classifiers from human examples with GEPA. [Source](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
-- **[Abide](https://github.com/coldteadotai/abide)** · 207 stars — A coding-agent rule layer for constraints that ordinary linters cannot express. [Source](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
-- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 177 stars — Interlinked adds optional Jev judgments and evidence checks to local coding-agent checks. [Source](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
-- **[Unclutter](https://github.com/kitze/unclutter)** · 176 stars — A browser extension using Jev to identify ads, promotions, and subscription popups and save reusable hiding rules. [Source](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
-- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 134 stars — A Pi extension that flags tool risks before execution and checks output for secrets and failure types. [Source](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
-- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 132 stars — Adds checks for project rules, out-of-scope actions, repeated failures, and completion claims to Pi Agents. [Source](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
-- **[Supercov](https://github.com/supercorp-ai/supercov)** · 94 stars — A code-quality and coverage tool that uses JEV for bounded quality checks. [Source](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[fx · TypeSafe Permission Reviewer](https://github.com/vercel-labs/fx)** · 3.1K stars — A Zig coding agent with an optional TypeSafe Jev permission reviewer for pending tool actions. [Source](https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig)
+- **[Agent](https://github.com/AgentiLoop/Agent)** · 629 stars — An optional Jev command-risk advisor inside a native macOS Agent, with a TypeSafeKit client. [Source](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
+- **[jev-align](https://github.com/sutro-sh/jev-align)** · 276 stars — A CLI for calibrating Jev classifiers from human examples with GEPA. [Source](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
+- **[Abide](https://github.com/coldteadotai/abide)** · 210 stars — A coding-agent rule layer for constraints that ordinary linters cannot express. [Source](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
+- **[Unclutter](https://github.com/kitze/unclutter)** · 203 stars — A browser extension using Jev to identify ads, promotions, and subscription popups and save reusable hiding rules. [Source](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
+- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 178 stars — Interlinked adds optional Jev judgments and evidence checks to local coding-agent checks. [Source](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
+- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 140 stars — A Pi extension that flags tool risks before execution and checks output for secrets and failure types. [Source](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
+- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 136 stars — Adds checks for project rules, out-of-scope actions, repeated failures, and completion claims to Pi Agents. [Source](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
+- **[Supercov](https://github.com/supercorp-ai/supercov)** · 106 stars — A code-quality and coverage tool that uses JEV for bounded quality checks. [Source](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 74 stars — A semantic code linter that batches repository findings through a zero-dependency JEV client. [Source](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
- **[Captaincore](https://github.com/CaptainCore/captaincore)** · 71 stars — Jev commands in the WordPress toolkit CaptainCore answer structured questions and prioritize malware scanner findings for review. [Source](https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go)
-- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 70 stars — A semantic code linter that batches repository findings through a zero-dependency JEV client. [Source](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
-- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 55 stars — An Oxlint plugin that evaluates semantic lint rules with JEV. [Source](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 58 stars — An Oxlint plugin that evaluates semantic lint rules with JEV. [Source](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Canny · Jev Hooks](https://github.com/qkal/Canny)** · 50 stars — Claude Code and Codex hooks that keep an append-only ledger and optionally ask Jev Noul questions for advisory checks. [Source](https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts)
[↑ Back to discovery](#discovery)
-### Agent workflows (14)
+### Agent workflows (15)
-- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 30.9K stars — A Claude Code mod that classifies subagent model and reasoning-effort needs. [Source](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
+- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 31.2K stars — A Claude Code mod that classifies subagent model and reasoning-effort needs. [Source](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
- **[Openwork](https://github.com/different-ai/openwork)** · 23.7K stars — An open-source cowork-style agent workspace that can run JEV-backed reviews inside CI and skill workflows. [Source](https://github.com/different-ai/openwork/blob/80c74d2160034fa1738fc6be78dd24ddfeccfc59/.github/scripts/jev-test-coverage-review.mjs)
- **[Eve](https://github.com/vercel/eve)** · 5.3K stars — Vercel’s open agent framework for packaging skills, evals, and tool-using workflows. [Source](https://github.com/vercel/eve/blob/d004e6d47e9d25d0380c24b5a47b65a18f8b2784/README.md)
- **[RNSkill · JEV Office Gate](https://github.com/Pluviobyte/rnskill)** · 1.6K stars — An agent-skill collection containing a JEV gate for bounded office-document checks. [Source](https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py)
- **[NewsJack](https://github.com/elvisun/newsjack)** · 1.3K stars — An open-source PR workflow that screens a live news feed for timely brand opportunities. [Source](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer)
-- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 683 stars — A coding-agent runtime with JEV lanes for routing, compaction, memory, and tool-result handling. [Source](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
-- **[Smithers](https://github.com/smithersai/smithers)** · 418 stars — An agentic TypeScript workflow framework with a first-class JEV classify path. [Source](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
-- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 173 stars — A coding-agent skill/plugin that advises when to compact context to save tokens. [Source](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
-- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 134 stars — A Claude Code plugin that trims long Bash output with TypeSafe JEV before the model sees it. [Source](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
-- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 115 stars — A gateway for inserting Jev decisions into Claude Code and Codex tool flows. [Source](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
-- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 109 stars — A Rust CLI that ranks agent skills against live session context with JEV. [Source](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
+- **[Agent Beacon · Jev Evaluator](https://github.com/asymptote-labs/agent-beacon)** · 1.1K stars — Cross-harness agent memory that can ask Jev to score bounded, redacted session projections before promoting lessons. [Source](https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go)
+- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 684 stars — A coding-agent runtime with JEV lanes for routing, compaction, memory, and tool-result handling. [Source](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
+- **[Smithers](https://github.com/smithersai/smithers)** · 420 stars — An agentic TypeScript workflow framework with a first-class JEV classify path. [Source](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
+- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 180 stars — A coding-agent skill/plugin that advises when to compact context to save tokens. [Source](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
+- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 179 stars — A gateway for inserting Jev decisions into Claude Code and Codex tool flows. [Source](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
+- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 140 stars — A Claude Code plugin that trims long Bash output with TypeSafe JEV before the model sees it. [Source](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
+- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 115 stars — A structured JEV decision plugin for DeepSeek Harness and compatible coding-agent hosts. [Source](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
+- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 114 stars — A Rust CLI that ranks agent skills against live session context with JEV. [Source](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
- **[Bluenoise](https://github.com/rokcso/bluenoise)** · 90 stars — An X/Twitter filtering extension using local rules by default, with optional Jev checks for unmatched replies. [Source](https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts)
-- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 78 stars — A structured JEV decision plugin for DeepSeek Harness and compatible coding-agent hosts. [Source](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
-- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 62 stars — A context cleaner that asks JEV which history to retain, truncate, or drop. [Source](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
+- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 66 stars — A context cleaner that asks JEV which history to retain, truncate, or drop. [Source](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
[↑ Back to discovery](#discovery)
@@ -246,15 +261,15 @@ Browse the full catalogue by project type. Use the scenario guides above when yo
### Interfaces & automation (9)
-- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.0K stars — A JEV UI-composition experiment that selects from predefined components and properties. [Source](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
-- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 1.6K stars — An Android conversation copilot that reads visible chat context and proposes replies without auto-sending. [Source](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
+- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.1K stars — A JEV UI-composition experiment that selects from predefined components and properties. [Source](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
+- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 4.1K stars — An Android conversation copilot that reads visible chat context and proposes replies without auto-sending. [Source](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
- **[Aiavatarkit](https://github.com/uezo/aiavatarkit)** · 678 stars — An optional AIAvatarKit component uses Jev to judge turn endings from speech transcripts. [Source](https://github.com/uezo/aiavatarkit/blob/38b617b8b9269939734e70ef503d7ea6976acdbd/aiavatar/sts/vad/turn_end_gates/jev.py#L172)
-- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 642 stars — A local macOS computer-use companion driven by Jev or Gemini Live. [Source](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
-- **[Notra](https://github.com/usenotra/notra)** · 207 stars — A product that turns work artifacts into publishable content with agent skill hooks. [Source](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
+- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 654 stars — A local macOS computer-use companion driven by Jev or Gemini Live. [Source](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
+- **[Notra](https://github.com/usenotra/notra)** · 213 stars — A product that turns work artifacts into publishable content with agent skill hooks. [Source](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
- **[OpenWhisper](https://github.com/Knuckles92/OpenWhisper)** · 187 stars — A dictation and meeting-notes app with optional Jev checks for topic changes, note-taker instructions and sensitive text. [Source](https://github.com/Knuckles92/OpenWhisper/blob/9e83653df183096104769e302a3c907cb277c551/README.md)
-- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 84 stars — A tool-using chat interface where JEV chooses only from code-supplied reply and tool options. [Source](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 91 stars — A tool-using chat interface where JEV chooses only from code-supplied reply and tool options. [Source](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 91 stars — YouTube extension detecting sponsored segments from live audio and transcripts using Jev, skipping promotional blocks automatically. [Source](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
- **[Jevmeter](https://github.com/ChetasLua/jevmeter)** · 81 stars — Creates edited videos with score meters by asking Jev to rate transcript sentences against selected rubrics. [Source](https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py)
-- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 81 stars — YouTube extension detecting sponsored segments from live audio and transcripts using Jev, skipping promotional blocks automatically. [Source](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
[↑ Back to discovery](#discovery)
@@ -262,61 +277,63 @@ Browse the full catalogue by project type. Use the scenario guides above when yo
### Developer tools (35)
-- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.0K stars — Compacts Claude Code tool history while keeping retained text verbatim. [Source](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
-- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.6K stars — Official agent skills for building against TypeSafe’s System One / JEV API. [Source](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
-- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.0K stars — A large community directory of public JEV projects, integrations, and discussions. [Source](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
-- **[Celesto](https://github.com/CelestoAI/celesto)** · 958 stars — A Celesto PR-review example prepares sandbox checks and compares a general model with Jev on candidate findings. [Source](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
-- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 810 stars — A terminal generator CLI that ships with an agent skill for creating content from the command line. [Source](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
-- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 772 stars — An evidence-backed playbook of JEV use cases, patterns, prompts, and starter code. [Source](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
-- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 628 stars — A curated directory focused on tools built for JEV and TypeSafe System One. [Source](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
-- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 502 stars — Reviews a Git diff or a codebase in stages and displays review leads in a local dashboard. [Source](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
-- **[Foreman](https://github.com/thruwire/foreman)** · 472 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](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
-- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 431 stars — A curated list of official and community resources for TypeSafe, System One, and JEV. [Source](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
-- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 431 stars — A source-backed field guide to TypeSafe JEV projects, SDKs, demos, and evaluations. [Source](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
-- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 398 stars — A collection of agent skills and workflows with a standard-library JEV decision client. [Source](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
-- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 334 stars — A source-backed JEV ecosystem radar with plain-language discovery and GitHub sync. [Source](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
-- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 316 stars — A source-backed list of projects built with JEV / TypeSafe System One. [Source](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
-- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 281 stars — OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold. [Source](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
-- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 257 stars — A curated list of JEV applications, libraries, and System One resources. [Source](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
-- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 245 stars — Eight MCP tools for evidence checks, content screening, search, reranking, classification, comparison and extraction. [Source](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
-- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 227 stars — An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions. [Source](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
-- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 196 stars — A local MCP code-quality reviewer returning structured scores to coding Agents. [Source](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
-- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 187 stars — A source-backed JEV directory plus a reusable JEV-only GitHub review workflow. [Source](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
-- **[Perch](https://github.com/lakeday-org/perch)** · 167 stars — Semantic code linting that asks JEV / System One whether findings should fire. [Source](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
-- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 159 stars — A gallery of System One papers, open reproductions, and independent evaluations. [Source](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
-- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 148 stars — A multilingual, community-maintained catalogue of open-source JEV projects. [Source](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.3K stars — Compacts Claude Code tool history while keeping retained text verbatim. [Source](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
+- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.9K stars — Official agent skills for building against TypeSafe’s System One / JEV API. [Source](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
+- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.3K stars — A large community directory of public JEV projects, integrations, and discussions. [Source](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
+- **[Celesto](https://github.com/CelestoAI/celesto)** · 966 stars — A Celesto PR-review example prepares sandbox checks and compares a general model with Jev on candidate findings. [Source](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
+- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 811 stars — A terminal generator CLI that ships with an agent skill for creating content from the command line. [Source](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
+- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 806 stars — An evidence-backed playbook of JEV use cases, patterns, prompts, and starter code. [Source](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
+- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 658 stars — A curated directory focused on tools built for JEV and TypeSafe System One. [Source](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
+- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 589 stars — A multilingual, community-maintained catalogue of open-source JEV projects. [Source](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 556 stars — Reviews a Git diff or a codebase in stages and displays review leads in a local dashboard. [Source](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
+- **[Foreman](https://github.com/thruwire/foreman)** · 517 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](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
+- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 476 stars — A curated list of official and community resources for TypeSafe, System One, and JEV. [Source](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
+- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 476 stars — A source-backed field guide to TypeSafe JEV projects, SDKs, demos, and evaluations. [Source](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
+- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 437 stars — A collection of agent skills and workflows with a standard-library JEV decision client. [Source](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
+- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 398 stars — A source-backed JEV ecosystem radar with plain-language discovery and GitHub sync. [Source](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
+- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 354 stars — A source-backed list of projects built with JEV / TypeSafe System One. [Source](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
+- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 303 stars — A curated list of JEV applications, libraries, and System One resources. [Source](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
+- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 290 stars — Eight MCP tools for evidence checks, content screening, search, reranking, classification, comparison and extraction. [Source](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
+- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 283 stars — OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold. [Source](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
+- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 262 stars — An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions. [Source](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
+- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 208 stars — A gallery of System One papers, open reproductions, and independent evaluations. [Source](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
+- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 206 stars — A local MCP code-quality reviewer returning structured scores to coding Agents. [Source](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
+- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 198 stars — A source-backed JEV directory plus a reusable JEV-only GitHub review workflow. [Source](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
+- **[Perch](https://github.com/lakeday-org/perch)** · 170 stars — Semantic code linting that asks JEV / System One whether findings should fire. [Source](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
+- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 149 stars — The source catalogue behind awesomejev.com, covering repos, integrations, demos, and research. [Source](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
+- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 130 stars — A TypeSafe JEV resource list centered on typed, confidence-aware decisions. [Source](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Building with JEV Skill](https://github.com/dbreunig/building-with-jev-skill)** · 128 stars — An agent skill for writing and improving programs that call JEV / System One. [Source](https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md)
-- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 127 stars — The source catalogue behind awesomejev.com, covering repos, integrations, demos, and research. [Source](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
-- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 125 stars — A TypeSafe JEV resource list centered on typed, confidence-aware decisions. [Source](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Awesome JEV (Promethe-us)](https://github.com/Promethe-us/awesome-jev)** · 122 stars — A source-backed bilingual collection of JEV resources, projects, research, and operational updates. [Source](https://github.com/Promethe-us/awesome-jev/blob/636623538eda3c363b8c396880ccdc8aba70d424/README.md)
-- **[Taskuary](https://github.com/ldbumble/taskuary)** · 116 stars — An optional Jev judgment module in Taskuary for checking user-defined conditions on task state. [Source](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
-- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 111 stars — Bounded JEV workflows for coding agents with a dedicated adapter. [Source](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
-- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 100 stars — A curated guide to JEV use cases, projects, SDKs, tools, and learning resources. [Source](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
-- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 96 stars — Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context. [Source](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
-- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 84 stars — A source-linked catalogue of shipped JEV demos, repositories, reported metrics, and model limits. [Source](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
-- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 73 stars — A curated JEV resource list with runnable typed-decision examples. [Source](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
-- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 71 stars — A reproducible benchmark comparing JEV with open decision models, classifiers, and rerankers. [Source](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
-- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 59 stars — A Chinese JEV ecosystem guide with curated projects, practical tutorials, pricing, and independent caveats. [Source](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
+- **[Taskuary](https://github.com/ldbumble/taskuary)** · 118 stars — An optional Jev judgment module in Taskuary for checking user-defined conditions on task state. [Source](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
+- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 114 stars — A source-linked catalogue of shipped JEV demos, repositories, reported metrics, and model limits. [Source](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
+- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 113 stars — A curated guide to JEV use cases, projects, SDKs, tools, and learning resources. [Source](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
+- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 113 stars — Bounded JEV workflows for coding agents with a dedicated adapter. [Source](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
+- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 104 stars — Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context. [Source](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
+- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 83 stars — A reproducible benchmark comparing JEV with open decision models, classifiers, and rerankers. [Source](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
+- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 78 stars — A curated JEV resource list with runnable typed-decision examples. [Source](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
+- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 60 stars — A Chinese JEV ecosystem guide with curated projects, practical tutorials, pricing, and independent caveats. [Source](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
[↑ Back to discovery](#discovery)
-### Domain tools (13)
+### Domain tools (15)
- **[AI Hedge Fund · JEV Adapter](https://github.com/virattt/ai-hedge-fund)** · 63.7K stars — An educational hedge-fund prototype with an optional JEV adapter for structured strategy judgments. [Source](https://github.com/virattt/ai-hedge-fund/blob/154a8b2f46dca0f40764d814e4e747b0ad71f4c4/hedge_fund/llm/client.py)
- **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · 12.0K stars — An open-source trading OS with a JEV decision gate before selected live entries. [Source](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py)
-- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 1.9K stars — A market-making experiment that can request one JEV trade decision per Monad block. [Source](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
-- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 485 stars — A Minecraft agent where a planner sets objectives and JEV selects legal game actions. [Source](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
-- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 338 stars — An experimental NES Mario controller that gives Jev structured emulator RAM and telemetry instead of screenshots. [Source](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
-- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 170 stars — An embodied-agent evaluation stack comparing JEV with local and general-model decision policies. [Source](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
-- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 155 stars — A browser driving simulator where Jev chooses among locally generated paths and speeds. [Source](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
-- **[Dasheng](https://github.com/wquguru/dasheng)** · 119 stars — A reading-practice app that combines streaming ASR with per-word JEV judgments. [Source](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
-- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 119 stars — A MuJoCo drone experiment deriving scene features from camera buffers for Jev tactical advice. [Source](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
-- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 103 stars — Agent tool evaluation harness comparing standard LLM tool selection against Jev routing across 100 mocked tools. [Source](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
+- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 2.1K stars — A market-making experiment that can request one JEV trade decision per Monad block. [Source](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
+- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 524 stars — A Minecraft agent where a planner sets objectives and JEV selects legal game actions. [Source](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
+- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 355 stars — An experimental NES Mario controller that gives Jev structured emulator RAM and telemetry instead of screenshots. [Source](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
+- **[quackd · Jev Stepper](https://github.com/rokbenko/quackd)** · 229 stars — A robot orchestration CLI with an optional System One stepper that chooses among permitted discrete robot calls. [Source](https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py)
+- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 191 stars — An embodied-agent evaluation stack comparing JEV with local and general-model decision policies. [Source](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
+- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 169 stars — A browser driving simulator where Jev chooses among locally generated paths and speeds. [Source](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
+- **[Dasheng](https://github.com/wquguru/dasheng)** · 135 stars — A reading-practice app that combines streaming ASR with per-word JEV judgments. [Source](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
+- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 134 stars — A MuJoCo drone experiment deriving scene features from camera buffers for Jev tactical advice. [Source](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
+- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 110 stars — A live Hyperliquid trading agent with JEV as a bounded decision layer. [Source](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
+- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 105 stars — Agent tool evaluation harness comparing standard LLM tool selection against Jev routing across 100 mocked tools. [Source](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
- **[Formanator](https://github.com/timrogers/formanator)** · 99 stars — A CLI and MCP client for benefit claims with optional JEV receipt classification. [Source](https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs)
-- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 95 stars — A live Hyperliquid trading agent with JEV as a bounded decision layer. [Source](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
-- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 69 stars — An autonomous liquidity agent that uses JEV inside its rebalancing decision service. [Source](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 79 stars — An autonomous liquidity agent that uses JEV inside its rebalancing decision service. [Source](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[HA-Jev](https://github.com/AboveColin/HA-Jev)** · 50 stars — A Home Assistant integration that turns typed Jev questions about house state into sensors and automation actions. [Source](https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py)
[↑ Back to discovery](#discovery)
diff --git a/README.zh-CN.md b/README.zh-CN.md
index 0b87ee3..79a4d59 100644
--- a/README.zh-CN.md
+++ b/README.zh-CN.md
@@ -1,14 +1,22 @@
+
+
+
+
+
Awesome JEV
从你的使用场景出发，找到值得参考的 JEV 项目，看懂它解决什么问题、哪部分值得借鉴。
+想在自己的项目中接入 JEV？ 在 BeatAPI 免费接入 JEV API →
+
快速查找 · 精选项目 Gallery · 用 Agent 搜索 · 全部项目 · 在线搜索
English · 简体中文 · 日本語
-183 个项目 · 10 种项目类型 · 更新于 2026-09-22
每个仓库 50+ Star · 已核对源码
+192 个项目 · 10 种项目类型 · 更新于 2026-09-23
每个仓库 50+ Star · 已核对源码
@@ -28,13 +36,13 @@
| | |
| --- | --- |
| [](https://github.com/browser-use/jev-ultrafast) | [](https://github.com/BerriAI/litellm) |
-| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 浏览器与电脑操作 · 16.3K Star
一次 JEV 请求选择浏览器动作和对应 DOM 目标。
[原始案例 · 294 万浏览](https://x.com/gregpr07/status/2100411066966749359) · [固定版本源码](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · 路由与优化 · 59.4K Star
在路由后端模型前判断请求复杂度。
[固定版本源码](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
+| **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 浏览器与电脑操作 · 18.2K Star
一次 JEV 请求选择浏览器动作和对应 DOM 目标。
[原始案例 · 294 万浏览](https://x.com/gregpr07/status/2100411066966749359) · [固定版本源码](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py) | **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · 路由与优化 · 59.4K Star
在路由后端模型前判断请求复杂度。
[固定版本源码](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70) |
| [](https://github.com/langchain-ai/langchain) | [](https://github.com/vercel/ai) |
-| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK 集成 · 146.8K Star
在 Python 工作流中提供二元、分类和有序评分判断。
[固定版本源码](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK 集成 · 26.9K Star
把 choice、score 和 yes/no 问题映射到统一 evaluate 接口。
[固定版本源码](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
+| **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · SDK 集成 · 146.9K Star
在 Python 工作流中提供二元、分类和有序评分判断。
[固定版本源码](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py) | **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · SDK 集成 · 26.9K Star
把 choice、score 和 yes/no 问题映射到统一 evaluate 接口。
[固定版本源码](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts) |
| [](https://github.com/NandhaKishorM/laya) | [](https://github.com/can1357/jegrep) |
-| **[Laya](https://github.com/NandhaKishorM/laya)** · 开放模型 · 11.5K Star
多语言 System 1 类型化判断，单次前向完成，并带检查点路由。
[Hugging Face 模型](https://huggingface.co/convaiinnovations/laya) · [固定版本源码](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)| **[jegrep](https://github.com/can1357/jegrep)** · 搜索与数据
无需建立索引，直接对实时代码树做语义检索。
[固定版本源码](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs) |
+| **[Laya](https://github.com/NandhaKishorM/laya)** · 开放模型 · 17.1K Star
多语言 System 1 类型化判断，单次前向完成，并带检查点路由。
[Hugging Face 模型](https://huggingface.co/convaiinnovations/laya) · [固定版本源码](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)| **[jegrep](https://github.com/can1357/jegrep)** · 搜索与数据
无需建立索引，直接对实时代码树做语义检索。
[固定版本源码](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs) |
| [](https://github.com/trycua/cua) | [](https://github.com/agentgateway/agentgateway) |
-| **[Cua · JEV Use](https://github.com/trycua/cua)** · 浏览器与电脑操作 · 25.8K Star
根据 DOM 或视觉区域描述选择下一步动作 ID。
[固定版本源码](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 安全与审查 · 5.0K Star
在网关层判断越狱、有害内容与密钥泄露风险。
[固定版本源码](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
+| **[Cua · JEV Use](https://github.com/trycua/cua)** · 浏览器与电脑操作 · 26.0K Star
根据 DOM 或视觉区域描述选择下一步动作 ID。
[固定版本源码](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11) | **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 安全与审查 · 5.0K Star
在网关层判断越狱、有害内容与密钥泄露风险。
[固定版本源码](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts) |
| [](https://github.com/elvisun/newsjack) | [](https://github.com/OpenByteInc/QuantDinger) |
| **[NewsJack](https://github.com/elvisun/newsjack)** · Agent 工作流 · 1.3K Star
先筛选数百条实时新闻，再让 Agent 继续处理少量入选机会。
[原始案例 · 58.3 万浏览](https://x.com/elvissun/status/2100951347080421409) · [固定版本源码](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer) | **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · 垂直工具 · 12.0K Star
在部分真实交易入场前增加证据质量与风险判断闸门。
[固定版本源码](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py) |
@@ -60,49 +68,50 @@ npx skills add BeatAPI/awesome-jev
-## 全部 183 个项目
+## 全部 192 个项目
下面按项目类型浏览完整目录。如果已经有具体需求，可以先看上方的场景指南。
-[浏览器与电脑操作 · 16](#browser-computer-use) · [SDK 与框架集成 · 30](#sdk-integrations) · [路由与优化 · 16](#routing-optimization) · [开放模型 · 21](#open-models) · [搜索与数据 · 15](#search-data)
-[安全与审查 · 14](#safety-review) · [Agent 工作流 · 14](#agent-workflows) · [界面与自动化 · 9](#interfaces) · [开发者工具 · 35](#developer-tools) · [垂直工具 · 13](#domain-tools)
+[浏览器与电脑操作 · 17](#browser-computer-use) · [SDK 与框架集成 · 31](#sdk-integrations) · [路由与优化 · 16](#routing-optimization) · [开放模型 · 23](#open-models) · [搜索与数据 · 15](#search-data)
+[安全与审查 · 16](#safety-review) · [Agent 工作流 · 15](#agent-workflows) · [界面与自动化 · 9](#interfaces) · [开发者工具 · 35](#developer-tools) · [垂直工具 · 15](#domain-tools)
-### 浏览器与电脑操作 (16)
+### 浏览器与电脑操作 (17)
-- **[Cua · JEV Use](https://github.com/trycua/cua)** · 25.8K Star — 把 Cua Driver 的观察与执行能力和 JEV 的有限动作选择结合起来。 [源码证据](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
-- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 16.3K Star — 用 JEV 一次选出浏览器动作与对应 DOM 元素，只在需要输入文本时再调用文本模型。 [源码证据](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
-- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.4K Star — 根据系统无障碍数据选择原生桌面控件与动作的可选 JEV Skill。 [源码证据](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
-- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 765 Star — 用 OCR 和界面状态构造候选动作，让 Jev 决定如何操作 macOS，需要写文字时再调用文本模型。 [源码证据](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
-- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 549 Star — 用 JEV 决定下一步界面动作的 Computer Use 辅助工具。 [源码证据](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
+- **[Cua · JEV Use](https://github.com/trycua/cua)** · 26.0K Star — 把 Cua Driver 的观察与执行能力和 JEV 的有限动作选择结合起来。 [源码证据](https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11)
+- **[Jev Ultrafast](https://github.com/browser-use/jev-ultrafast)** · 18.2K Star — 用 JEV 一次选出浏览器动作与对应 DOM 元素，只在需要输入文本时再调用文本模型。 [源码证据](https://github.com/browser-use/jev-ultrafast/blob/1231850a0bf1a0c0341fe408ef1668dbbfdfac46/jev_ultrafast/model.py)
+- **[Agent Desktop · JEV Skill](https://github.com/lahfir/agent-desktop)** · 1.5K Star — 根据系统无障碍数据选择原生桌面控件与动作的可选 JEV Skill。 [源码证据](https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs)
+- **[TypeSafe Computer Use](https://github.com/awlevin/typesafe-computer-use)** · 843 Star — 用 OCR 和界面状态构造候选动作，让 Jev 决定如何操作 macOS，需要写文字时再调用文本模型。 [源码证据](https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py)
+- **[Jev CU](https://github.com/Sac-Y/Jev-cu)** · 581 Star — 用 JEV 决定下一步界面动作的 Computer Use 辅助工具。 [源码证据](https://github.com/Sac-Y/Jev-cu/blob/38fb31de7dfe6209bbe6e04057c00c6e885ba577/scripts/jev-decide.mjs)
- **[Omg.dev](https://github.com/BennyKok/omg.dev)** · 535 Star — omg.dev 的移动端测试脚本可让 Jev 读取可访问性树并选择下一步交互。 [源码证据](https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts)
-- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 334 Star — 给 Codex 浏览器工作流加一个 Skill：Jev 选导航、点击和滚动，Codex 保留文字输入与最终核验。 [源码证据](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
-- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 329 Star — 通过 Mobilerun 控制 Android 手机，网页面板与命令行可查看 Jev 的操作过程。 [源码证据](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
-- **[Third Hand](https://github.com/shhivv/third-hand)** · 283 Star — Swift 客户端栈：在驱动本地 computer-use 辅助逻辑时调用 JEV。 [源码证据](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
-- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 231 Star — 给定任务与网址后驱动浏览器，返回最终页面、截图和逐步操作记录。 [源码证据](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
-- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 216 Star — 用语音控制 Playwright 浏览器，将逐步转写的口令交给 Jev 判断。 [源码证据](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
-- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 127 Star — 带 TypeSafe Policy 实现的高速 Computer Use 动作层。 [源码证据](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
-- **[JEV Use](https://github.com/savka777/jev-use)** · 86 Star — 无需视觉模型、通过无障碍控件操作 macOS 的 Computer Use 工具。 [源码证据](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
-- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 76 Star — 浏览器 Agent 基准测试工具，包含基于无障碍树与 WebMCP 接口的冻结 JEV 实验。 [源码证据](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
-- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 68 Star — 一个实验性 Chrome 扩展，让 Jev 判断候选 DOM 元素是否是广告，再高亮或移除。 [源码证据](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
-- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 67 Star — 把 LLM 规划与 JEV 动作结合的浏览器自动化库、CLI 与 MCP Server。 [源码证据](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[JEV Browser Use](https://github.com/wy-coliney/jev-browser-use)** · 387 Star — 给 Codex 浏览器工作流加一个 Skill：Jev 选导航、点击和滚动，Codex 保留文字输入与最终核验。 [源码证据](https://github.com/wy-coliney/jev-browser-use/blob/f14b60e0ae1ee90cd73eb6650e30a666a84c021a/skills/jev-browser-use/bridge.mjs)
+- **[Mobile JEV](https://github.com/droidrun/mobile-jev)** · 356 Star — 通过 Mobilerun 控制 Android 手机，网页面板与命令行可查看 Jev 的操作过程。 [源码证据](https://github.com/droidrun/mobile-jev/blob/395fc222beac4f059f9a0beb337d114a2b066e99/scripts/mobile-agent/policy.mjs#L224)
+- **[Third Hand](https://github.com/shhivv/third-hand)** · 288 Star — Swift 客户端栈：在驱动本地 computer-use 辅助逻辑时调用 JEV。 [源码证据](https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift)
+- **[JEV Voice Browser](https://github.com/moritzkremb/jev-voice-browser)** · 247 Star — 用语音控制 Playwright 浏览器，将逐步转写的口令交给 Jev 判断。 [源码证据](https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123)
+- **[JEV Browser](https://github.com/jkudish/jev-browser)** · 240 Star — 给定任务与网址后驱动浏览器，返回最终页面、截图和逐步操作记录。 [源码证据](https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1)
+- **[Arc CUA · TypeSafe Policy](https://github.com/shhivv/arc-cua)** · 128 Star — 带 TypeSafe Policy 实现的高速 Computer Use 动作层。 [源码证据](https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py)
+- **[fastbrowse · Jev](https://github.com/agent-labs-dev/fastbrowse)** · 97 Star — 浏览器 Agent：索引页面控件，由 Jev 选择下一步动作，LLM 负责规划与阅读。 [源码证据](https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py)
+- **[JEV Use](https://github.com/savka777/jev-use)** · 92 Star — 无需视觉模型、通过无障碍控件操作 macOS 的 Computer Use 工具。 [源码证据](https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md)
+- **[WindTunnel · JEV Benchmarks](https://github.com/nekuda-ai/WindTunnel)** · 79 Star — 浏览器 Agent 基准测试工具，包含基于无障碍树与 WebMCP 接口的冻结 JEV 实验。 [源码证据](https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs)
+- **[JEV Browser](https://github.com/Ying-Kai-Liao/jev-browser)** · 76 Star — 把 LLM 规划与 JEV 动作结合的浏览器自动化库、CLI 与 MCP Server。 [源码证据](https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs)
+- **[TypeSafe Adblock](https://github.com/realZachi/typesafe-adblock)** · 71 Star — 一个实验性 Chrome 扩展，让 Jev 判断候选 DOM 元素是否是广告，再高亮或移除。 [源码证据](https://github.com/realZachi/typesafe-adblock/blob/7e067d243d87b7fe4d511653c0ddcd77b9beee18/src/typesafe.js)
[↑ 返回检索入口](#discovery)
-### SDK 与框架集成 (30)
+### SDK 与框架集成 (31)
-- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.8K Star — 面向 Python LangChain 工作流的可选 JEV 分类器集成。 [源码证据](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
+- **[LangChain · TypeSafe](https://github.com/langchain-ai/langchain)** · 146.9K Star — 面向 Python LangChain 工作流的可选 JEV 分类器集成。 [源码证据](https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py)
- **[Composio · TypeSafe Provider](https://github.com/ComposioHQ/composio)** · 30.3K Star — 用 JEV 在工具与有限参数选项之间做选择的 TypeSafe Provider。 [源码证据](https://github.com/ComposioHQ/composio/blob/4b5920bf7aa55c8a44657b060d4bd25ce7b13a9a/ts/packages/providers/typesafe/src/decide.ts)
- **[Vercel AI SDK · TypeSafe](https://github.com/vercel/ai)** · 26.9K Star — AI SDK 的 TypeSafe Provider 通过统一 evaluate 接口调用 JEV。 [源码证据](https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts)
- **[Pydantic AI · TypeSafe](https://github.com/pydantic/pydantic-ai)** · 20.1K Star — 把受支持的结构化输出映射为 JEV 问题的 TypeSafe 模型集成。 [源码证据](https://github.com/pydantic/pydantic-ai/blob/c4898abb54dc25ae6f6aef208a4c0661b30a455e/pydantic_ai_slim/pydantic_ai/models/typesafe.py)
- **[Eliza · TypeSafe Client](https://github.com/elizaOS/eliza)** · 19.4K Star — Eliza Agent 操作系统中的可选 TypeSafe HTTP 适配器。 [源码证据](https://github.com/elizaOS/eliza/blob/ebc808e3a67fb941e29153d89fc896524d32fe3c/packages/agent/src/services/typesafe/client.ts)
- **[LangChain.js · TypeSafe](https://github.com/langchain-ai/langchainjs)** · 18.2K Star — 面向 JavaScript LangChain 工作流的 TypeSafeClassifier 集成。 [源码证据](https://github.com/langchain-ai/langchainjs/blob/206d8b992bcf90ce7d46f2158f1ad85fc1d826c0/libs/providers/langchain-typesafe/src/classifier.ts)
- **[Rig · TypeSafe Crate](https://github.com/0xPlaygrounds/rig)** · 8.7K Star — 用 Rust 类型表达 JEV 问题与答案的实验性 crate。 [源码证据](https://github.com/0xPlaygrounds/rig/blob/2d16c1b25f6749b3a2cd841beddf767106495069/crates/rig-typesafeai/src/wire.rs#L40)
-- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.2K Star — Bifrost 通过统一 Decision 操作把 TypeSafe 接成原生 Provider。 [源码证据](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
-- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 6.9K Star — 自托管 AI 网关内置一等 JEV 渠道与原生 Decisions 路由。 [源码证据](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
+- **[Bifrost · TypeSafe Provider](https://github.com/maximhq/bifrost)** · 8.3K Star — Bifrost 通过统一 Decision 操作把 TypeSafe 接成原生 Provider。 [源码证据](https://github.com/maximhq/bifrost/blob/40c3f7ee3a1a4277c5b269b850ca7725d419907f/core/providers/typesafe/typesafe.go)
+- **[GPT-Load · JEV Provider](https://github.com/tbphp/gpt-load)** · 7.0K Star — 自托管 AI 网关内置一等 JEV 渠道与原生 Decisions 路由。 [源码证据](https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go)
- **[Open Connector · TypeSafe](https://github.com/oomol-lab/open-connector)** · 5.9K Star — Open Connector 通过 Provider Runtime 暴露 TypeSafe 评估能力。 [源码证据](https://github.com/oomol-lab/open-connector/blob/4e6d8533ae436cba4781b6fb69d1bf31e5fb5ffe/src/providers/typesafe_ai/runtime.ts)
- **[Kiln · JEV Adapter](https://github.com/Kiln-AI/Kiln)** · 5.1K Star — Kiln 把兼容的 JSON Schema 映射为 JEV 问题，用于结构化单轮任务。 [源码证据](https://github.com/Kiln-AI/Kiln/blob/5992e2e51ec6782fce33f277145954ba3ae8467e/libs/core/kiln_ai/adapters/model_adapters/jev_adapter.py)
- **[Ax · TypeSafe Client](https://github.com/ax-llm/ax)** · 2.9K Star — 支持布尔、有限分类以及原生 JEV 请求的 TypeSafe 集成。 [源码证据](https://github.com/ax-llm/ax/blob/5c43344f9ef3016db576fa2c3b59d48ef21b4d71/src/ax/ai/typesafe/client.ts#L1)
@@ -110,20 +119,21 @@ npx skills add BeatAPI/awesome-jev
- **[Vellum Assistant · JEV](https://github.com/vellum-ai/vellum-assistant)** · 1.3K Star — 提交会话状态与显式问题的可选 JEV Provider。 [源码证据](https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts)
- **[Laravel AI · TypeSafe](https://github.com/laravel/ai)** · 1.2K Star — Laravel AI 内置由 JEV 驱动的 TypeSafe 分类 Provider。 [源码证据](https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php)
- **[Req LLM](https://github.com/agentjido/req_llm)** · 581 Star — ReqLLM 的 TypeSafe provider，让 Elixir 应用通过 evaluate 接口调用 Jev。 [源码证据](https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md)
-- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 458 Star — 将任意开源大语言模型转化为分类器与 Jev 兼容端点的适配服务，无需额外训练专用分类头。 [源码证据](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
-- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 406 Star — 以 JEV 为主要类型化决策后端的零样本分类服务。 [源码证据](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
+- **[Simple JEV](https://github.com/featherless-ai/simple-jev)** · 485 Star — 将任意开源大语言模型转化为分类器与 Jev 兼容端点的适配服务，无需额外训练专用分类头。 [源码证据](https://github.com/featherless-ai/simple-jev/blob/0dd5396ffce671ab7c4bfc031506d8e558cf8d23/demos/jevpilot/src/simple-jev-api.js#L1-L85)
+- **[Classifier.dev](https://github.com/mrmps/classifier-dev)** · 411 Star — 以 JEV 为主要类型化决策后端的零样本分类服务。 [源码证据](https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts)
- **[Instructor Php](https://github.com/cognesy/instructor-php)** · 327 Star — Instructor PHP 的 Polyglot 模块内置 TypeSafe Decision 驱动。 [源码证据](https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md)
-- **[Openjev](https://github.com/razorback16/openjev)** · 286 Star — 一个兼容 Jev System One 接口的独立决策服务，使用开源 DiffusionGemma 模型运行。 [源码证据](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
-- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 248 Star — 让 Scala 应用也能接入 Jev。这个多模型客户端新增了独立 TypeSafe 模块，用状态和判断题获取结构化答案。 [源码证据](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
-- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 244 Star — 给 Pi 的工具运行时加上可编程的 Jev 决策循环。先写好观察、判断和执行步骤，再让它按预算在前台或后台运行。 [源码证据](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
-- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 243 Star — 可替换 TypeSafeClient 的适配器，用普通 LLM API 承接 System One 调用。 [源码证据](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
-- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 214 Star — TypeSafe 组织发布的 JavaScript 与 TypeScript SDK，提供 Jev 请求和回答类型。 [源码证据](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
-- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 188 Star — TypeSafe 官方 Python SDK，提供 Jev System One 的同步、异步客户端及问题和回答类型。 [源码证据](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[Openjev](https://github.com/razorback16/openjev)** · 319 Star — 一个兼容 Jev System One 接口的独立决策服务，使用开源 DiffusionGemma 模型运行。 [源码证据](https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py)
+- **[System One Adapter (Python)](https://github.com/typesafe-ai/system-one-adapter-python)** · 266 Star — 可替换 TypeSafeClient 的适配器，用普通 LLM API 承接 System One 调用。 [源码证据](https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py)
+- **[Pi Fabric](https://github.com/monotykamary/pi-fabric)** · 253 Star — 给 Pi 的工具运行时加上可编程的 Jev 决策循环。先写好观察、判断和执行步骤，再让它按预算在前台或后台运行。 [源码证据](https://github.com/monotykamary/pi-fabric/blob/2e5fcdd94cc364445b115756edefcaab24f68928/README.md)
+- **[Openai Scala Client](https://github.com/cequence-io/openai-scala-client)** · 249 Star — 让 Scala 应用也能接入 Jev。这个多模型客户端新增了独立 TypeSafe 模块，用状态和判断题获取结构化答案。 [源码证据](https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md)
+- **[TypeSafe SDK Js](https://github.com/typesafe-ai/typesafe-sdk-js)** · 226 Star — TypeSafe 组织发布的 JavaScript 与 TypeScript SDK，提供 Jev 请求和回答类型。 [源码证据](https://github.com/typesafe-ai/typesafe-sdk-js/blob/66880ccded6cb642dc1809620c2b108c33730214/README.md)
+- **[TypeSafe SDK Python](https://github.com/typesafe-ai/typesafe-sdk-python)** · 205 Star — TypeSafe 官方 Python SDK，提供 Jev System One 的同步、异步客户端及问题和回答类型。 [源码证据](https://github.com/typesafe-ai/typesafe-sdk-python/blob/2ce5c65f13646cab6e6f782328194c9d85f3300a/src/typesafe_sdk/_core/client/sync/client.py#L206-L221)
+- **[AI SDK Python · Jev Evaluate](https://github.com/vercel-labs/ai-python)** · 185 Star — Vercel Python AI SDK 的实验性 evaluate 路径，可通过 typesafe-ai/jev 提问类型化问题。 [源码证据](https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py)
- **[Runline](https://github.com/Michaelliv/runline)** · 163 Star — Runline 的 TypeSafe 插件，把 Jev 判断作为 Agent JavaScript 可调用的动作。 [源码证据](https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md)
-- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 137 Star — NeuroLink 提供 TypeSafe Provider 与基于 JEV 的分类路由能力。 [源码证据](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
+- **[NeuroLink · TypeSafe](https://github.com/Juspay/neurolink)** · 138 Star — NeuroLink 提供 TypeSafe Provider 与基于 JEV 的分类路由能力。 [源码证据](https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts)
- **[AI](https://github.com/hackclub/ai)** · 133 Star — Hack Club AI 代理中的 Jev 转发接口，复用已有鉴权、限额和用量记录。 [源码证据](https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md)
- **[Effect Agent](https://github.com/danieljvdm/effect-agent)** · 121 Star — Effect Agent 的 TypeSafe 决策 provider，支持类型化问题集与可选模型选择。 [源码证据](https://github.com/danieljvdm/effect-agent/blob/88005e497e9b627eeb16d670f278903c57601da9/README.md)
-- **[Advocaat](https://github.com/pithings/advocaat)** · 89 Star — 用简短的 TypeScript 调用向 Jev 提问。把同一份数据里的多个判断一次写好，直接拿到概率、选项和分数。 [源码证据](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
+- **[Advocaat](https://github.com/pithings/advocaat)** · 90 Star — 用简短的 TypeScript 调用向 Jev 提问。把同一份数据里的多个判断一次写好，直接拿到概率、选项和分数。 [源码证据](https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md)
[↑ 返回检索入口](#discovery)
@@ -132,49 +142,51 @@ npx skills add BeatAPI/awesome-jev
### 路由与优化 (16)
- **[LiteLLM · JEV Router](https://github.com/BerriAI/litellm)** · 59.4K Star — LiteLLM 可在按复杂度路由模型的策略中使用 JEV。 [源码证据](https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70)
-- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.4K Star — 在编码 Agent 中提供可选 TypeSafe 判断器，处理有边界的工作流决策。 [源码证据](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
-- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.2K Star — 在选择模型前先对消息分类的可选自动路由器。 [源码证据](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
-- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 6.9K Star — 可用 JEV 把任务简报匹配到调度规则的 Agent 团队调度器。 [源码证据](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
+- **[Oh My Pi · JEV Judgments](https://github.com/can1357/oh-my-pi)** · 32.8K Star — 在编码 Agent 中提供可选 TypeSafe 判断器，处理有边界的工作流决策。 [源码证据](https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4)
+- **[OpenChamber · JEV Router](https://github.com/openchamber/openchamber)** · 10.3K Star — 在选择模型前先对消息分类的可选自动路由器。 [源码证据](https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js)
+- **[Firstmate · JEV Dispatch](https://github.com/kunchenguid/firstmate)** · 7.0K Star — 可用 JEV 把任务简报匹配到调度规则的 Agent 团队调度器。 [源码证据](https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh)
- **[Agent Router · TypeSafe](https://github.com/theagentrouter/agent-router)** · 2.1K Star — Agent Router 在 Envoy 网关中提供原生 TypeSafe System One 转换器。 [源码证据](https://github.com/theagentrouter/agent-router/blob/7d7c07ffdb14241362def2f7d3f76cbd06d518cc/internal/translator/typesafe_systemone.go)
- **[LLMGateway · System One](https://github.com/theopenco/llmgateway)** · 1.7K Star — LLMGateway 实现带类型化 Schema 与 Provider 映射的原生 System One 路由。 [源码证据](https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts)
-- **[Atomic](https://github.com/bastani-inc/atomic)** · 812 Star — Atomic 编程 Agent 的可选 Jev 决策后端，为路由等流程提供受限的结构化选择。 [源码证据](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Atomic](https://github.com/bastani-inc/atomic)** · 814 Star — Atomic 编程 Agent 的可选 Jev 决策后端，为路由等流程提供受限的结构化选择。 [源码证据](https://github.com/bastani-inc/atomic/blob/33ca4ccb5f39488aeba67c0212a52a31cd14059d/packages/coding-agent/src/core/structured-output/jev.ts)
+- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 600 Star — 为 Hermes Agent 提供 JEV 驱动的路由、记忆、压缩、Skill 选择与电脑/浏览器操作。 [源码证据](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
- **[Vexjoy Agent](https://github.com/notque/vexjoy-agent)** · 423 Star — 给 VexJoy 的任务分派增加一条 Jev 路线。输入需求后，判断该选哪位专长 Agent、哪项技能和哪条工作流。 [源码证据](https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md)
-- **[Hermes JEV Skills](https://github.com/kerpopule/hermes-jev-skills)** · 401 Star — 为 Hermes Agent 提供 JEV 驱动的路由、记忆、压缩、Skill 选择与电脑/浏览器操作。 [源码证据](https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py)
-- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 331 Star — 给 WrongStack 编程 Agent 增加一个可选分派助手。遇到多个相近的专长 Agent 时，用 Jev 判断谁更适合当前任务。 [源码证据](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
-- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 316 Star — Claude Code / CLI 代理：Jev 给任务复杂度打分，并在当前账号可用的模型里选一个，再由本地策略决定是否更换。 [源码证据](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
-- **[Skillbox](https://github.com/kitze/skillbox)** · 223 Star — 自建一个有版本管理的 Agent 技能库，还能选配 Jev 推荐：告诉它当前任务，从你有权限使用的技能里挑更相关的。 [源码证据](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
-- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 184 Star — 每轮先让 Jev 判断任务类型与难度，再由本地策略为 Codex 选模型、思考深度和速度档。 [源码证据](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
-- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 145 Star — 用于模型、工具与子 Agent 的轻量 JEV 路由器。 [源码证据](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
-- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 74 Star — 为 Grok Bot 提供用量闸门与 Skill 模板的 JEV 决策层。 [源码证据](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
-- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 63 Star — 用 JEV 把任务路由到编码 Agent、模型与推理强度的 CLI。 [源码证据](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
+- **[JEV Router](https://github.com/gargpratyush/jev-router)** · 349 Star — Claude Code / CLI 代理：Jev 给任务复杂度打分，并在当前账号可用的模型里选一个，再由本地策略决定是否更换。 [源码证据](https://github.com/gargpratyush/jev-router/blob/38da6b84ea01241bfc41fbddc0928d0f40a703f0/src/policy.mjs#L1-L63)
+- **[WrongStack](https://github.com/WrongStack/WrongStack)** · 332 Star — 给 WrongStack 编程 Agent 增加一个可选分派助手。遇到多个相近的专长 Agent 时，用 Jev 判断谁更适合当前任务。 [源码证据](https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md)
+- **[JEV Codex Router](https://github.com/0xNatoshi/jev-codex-router)** · 229 Star — 每轮先让 Jev 判断任务类型与难度，再由本地策略为 Codex 选模型、思考深度和速度档。 [源码证据](https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py)
+- **[Skillbox](https://github.com/kitze/skillbox)** · 229 Star — 自建一个有版本管理的 Agent 技能库，还能选配 Jev 推荐：告诉它当前任务，从你有权限使用的技能里挑更相关的。 [源码证据](https://github.com/kitze/skillbox/blob/d83ba4ecd254c8dfa6a759d1feb5141384e26a9e/README.md)
+- **[JevRouter](https://github.com/BillionsBobby/JevRouter)** · 173 Star — 用于模型、工具与子 Agent 的轻量 JEV 路由器。 [源码证据](https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts)
+- **[Grok Bot JEV](https://github.com/Bodila51/grok-bot-jev)** · 77 Star — 为 Grok Bot 提供用量闸门与 Skill 模板的 JEV 决策层。 [源码证据](https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md)
+- **[Agent Router](https://github.com/nidhi-singh02/agent-router)** · 68 Star — 用 JEV 把任务路由到编码 Agent、模型与推理强度的 CLI。 [源码证据](https://github.com/nidhi-singh02/agent-router/blob/ad7571f38ea31ffbf3c28391f9e6d6383a7c08ba/packages/router/src/semantic/typesafe-client.ts)
[↑ 返回检索入口](#discovery)
-### 开放模型 (21)
+### 开放模型 (23)
-- **[Laya](https://github.com/NandhaKishorM/laya)** · 11.5K Star — 多语言非自回归 System 1 决策引擎：单次前向完成类型化 choice/score/noul，并由路由在检查点间选型。 [源码证据](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
-- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.4K Star — 用开放模型实现“语义 if”接口模式的独立研究项目。 [源码证据](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
-- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 2.6K Star — 基于 Qwen2.5-0.5B、可在 MacBook 上训练运行的轻量决策模型。 [源码证据](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
-- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 1.9K Star — 提供并行判断、动态候选与完整训练流程的 0.6B 开放 JEV 接口复刻。 [源码证据](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
-- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.5K Star — 面向本地类型化决策、数据整理与评测的开放工具包。 [源码证据](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
+- **[Laya](https://github.com/NandhaKishorM/laya)** · 17.1K Star — 多语言非自回归 System 1 决策引擎：单次前向完成类型化 choice/score/noul，并由路由在检查点间选型。 [源码证据](https://github.com/NandhaKishorM/laya/blob/d113dca2512fb3eaca313534bc54c7162d87c1d4/README.md)
+- **[Kev 0.5B](https://github.com/jaredpalmer/kev)** · 4.5K Star — 基于 Qwen2.5-0.5B、可在 MacBook 上训练运行的轻量决策模型。 [源码证据](https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md)
+- **[SemIf](https://github.com/TheoLeeCJ/SemIf)** · 3.9K Star — 用开放模型实现“语义 if”接口模式的独立研究项目。 [源码证据](https://github.com/TheoLeeCJ/SemIf/blob/ca3ba65f142967030ecb453346e94d6f476a69df/README.md)
+- **[NanoJev](https://github.com/TianyuCodings/NanoJev)** · 2.0K Star — 提供并行判断、动态候选与完整训练流程的 0.6B 开放 JEV 接口复刻。 [源码证据](https://github.com/TianyuCodings/NanoJev/blob/71a513bb0163b5634467842b523ee0c0ed6fb1c7/README.md)
+- **[Nimble](https://github.com/bespokelabsai/nimble)** · 1.6K Star — 面向本地类型化决策、数据整理与评测的开放工具包。 [源码证据](https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md)
- **[Jevlike](https://github.com/vinnylarouge/jevlike)** · 1.2K Star — 一次为动态文本或视觉候选列表评分的独立入门模型。 [源码证据](https://github.com/vinnylarouge/jevlike/blob/94f5fd1b0b11d52bbdfdf4e0ee6aa96b568f8452/README.md)
-- **[LocalJev](https://github.com/githubnext/localjev)** · 704 Star — 基于多种本地小模型、提供 JEV 兼容接口与评测的本地服务。 [源码证据](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
-- **[Splash](https://github.com/incoai/splash)** · 578 Star — 面向决策类模型的本地 Apple Silicon 推理引擎。 [源码证据](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
-- **[Von](https://github.com/wfzyx/von)** · 374 Star — 提供 JEV 兼容接口的本地开放 System One 决策模型。 [源码证据](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
-- **[Decider](https://github.com/Mapika/decider)** · 279 Star — 一次前向的类型化决策与校准概率，基于 Qwen3.5-2B 微调。 [源码证据](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
-- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 273 Star — 独立的本地决策模型，提供兼容 TypeSafe 的 System One 接口。 [源码证据](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
-- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 258 Star — 基于开源模型的 JEV 兼容 API，采用 prefill-only 服务路径。 [源码证据](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
-- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 246 Star — 约 151M 的校准非自回归本地决策引擎，目标对齐类 JEV 类型化答案。 [源码证据](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
-- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 221 Star — 在 Apple Silicon 上的教学向类 JEV 视觉推理实验，共享上下文直接给候选打分。 [源码证据](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
-- **[Jeff](https://github.com/logan-markewich/jeff)** · 205 Star — 基于 GliFormer、可自托管的 JEV 兼容服务。 [源码证据](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
-- **[Laya Node Runtime](https://github.com/receptron/laya)** · 195 Star — 面向开放 Laya JEV 兼容决策模型的 Node.js / TypeScript Runtime。 [源码证据](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
-- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 168 Star — 在 DGX Spark 上跑 DiffusionGemma NVFP4 结构化决策的容器配方。 [源码证据](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
-- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 125 Star — 把本地语言模型转成 JEV 兼容结构化决策引擎的适配器。 [源码证据](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
-- **[Reflex](https://github.com/kshetrajna12/reflex)** · 110 Star — 在 Qwen 上复现 JEV/System One 接口的小型开放决策模型。 [源码证据](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
-- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 91 Star — 支持自定义微调的开放 JEV 风格实现。 [源码证据](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
-- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 67 Star — 面向类型化决策请求的开源 JEV 兼容实现。 [源码证据](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[LocalJev](https://github.com/githubnext/localjev)** · 731 Star — 基于多种本地小模型、提供 JEV 兼容接口与评测的本地服务。 [源码证据](https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md)
+- **[Splash](https://github.com/incoai/splash)** · 638 Star — 面向决策类模型的本地 Apple Silicon 推理引擎。 [源码证据](https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md)
+- **[Von](https://github.com/wfzyx/von)** · 498 Star — 提供 JEV 兼容接口的本地开放 System One 决策模型。 [源码证据](https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md)
+- **[Rizzo Flow](https://github.com/Rizzo-AI-Academy/rizzo-flow)** · 348 Star — 独立的本地决策模型，提供兼容 TypeSafe 的 System One 接口。 [源码证据](https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py)
+- **[Decider](https://github.com/Mapika/decider)** · 316 Star — 一次前向的类型化决策与校准概率，基于 Qwen3.5-2B 微调。 [源码证据](https://github.com/Mapika/decider/blob/c4daaac28af9fea95d627015cffa2dd5a5926ee6/decider/__init__.py)
+- **[Laya Node Runtime](https://github.com/receptron/laya)** · 295 Star — 面向开放 Laya JEV 兼容决策模型的 Node.js / TypeScript Runtime。 [源码证据](https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts)
+- **[OpenJEV SGLang](https://github.com/ekzhang/openjev-sglang)** · 272 Star — 基于开源模型的 JEV 兼容 API，采用 prefill-only 服务路径。 [源码证据](https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py)
+- **[OpenJEV Verdict 2.0](https://github.com/Heman10x-NGU/openJev-verdict-2.0)** · 268 Star — 约 151M 的校准非自回归本地决策引擎，目标对齐类 JEV 类型化答案。 [源码证据](https://github.com/Heman10x-NGU/openJev-verdict-2.0/blob/a458733c5f43fc7f30b6e4381636cbfbf8437633/openjev/__init__.py)
+- **[JEV Visual](https://github.com/hr98w/jev-visual)** · 243 Star — 在 Apple Silicon 上的教学向类 JEV 视觉推理实验，共享上下文直接给候选打分。 [源码证据](https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py)
+- **[Jeff](https://github.com/logan-markewich/jeff)** · 219 Star — 基于 GliFormer、可自托管的 JEV 兼容服务。 [源码证据](https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md)
+- **[djev-spark](https://github.com/mmastrac/djev-spark)** · 176 Star — 在 DGX Spark 上跑 DiffusionGemma NVFP4 结构化决策的容器配方。 [源码证据](https://github.com/mmastrac/djev-spark/blob/1444f3e927f83ba508e5b28a4fd4fdd9ecd0976b/README.md)
+- **[LLM2Jev](https://github.com/Yinsongxu/LLM2Jev)** · 142 Star — 把本地语言模型转成 JEV 兼容结构化决策引擎的适配器。 [源码证据](https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py)
+- **[Reflex](https://github.com/kshetrajna12/reflex)** · 122 Star — 在 Qwen 上复现 JEV/System One 接口的小型开放决策模型。 [源码证据](https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md)
+- **[Open JEV](https://github.com/daseinlabs/open-jev)** · 103 Star — 支持自定义微调的开放 JEV 风格实现。 [源码证据](https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md)
+- **[OpenJev](https://github.com/SiliconLabAI/OpenJev)** · 101 Star — 面向类型化决策请求的开源 JEV 兼容实现。 [源码证据](https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md)
+- **[jevmlx](https://github.com/bnsd55/jevmlx)** · 59 Star — 在 Apple Silicon 本地从 MLX logits 为受限字段打分，并提供 System One HTTP 接口。 [源码证据](https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py)
+- **[Open Alternative to Jev](https://github.com/ikermoel/open-alternative-jev)** · 51 Star — 开源 System One 风格决策器，把类型化问题打包进本地开放模型的一次前向。 [源码证据](https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py)
[↑ 返回检索入口](#discovery)
@@ -182,63 +194,66 @@ npx skills add BeatAPI/awesome-jev
### 搜索与数据 (15)
-- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.4K Star — OpenViking 可用 JEV 对 Agent 记忆与上下文检索结果做校准重排。 [源码证据](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
-- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 24.8K Star — 在 Agent 记忆系统中加入 TypeSafe 重排器，可剔除无关召回候选。 [源码证据](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
+- **[OpenViking · JEV Rerank](https://github.com/volcengine/OpenViking)** · 38.5K Star — OpenViking 可用 JEV 对 Agent 记忆与上下文检索结果做校准重排。 [源码证据](https://github.com/volcengine/OpenViking/blob/b8bed5a1ad3a1c524b5e1fd0fa591df51ca9b7cc/openviking/models/rerank/jev_rerank.py)
+- **[Hindsight · TypeSafe Rerank](https://github.com/vectorize-io/hindsight)** · 25.3K Star — 在 Agent 记忆系统中加入 TypeSafe 重排器，可剔除无关召回候选。 [源码证据](https://github.com/vectorize-io/hindsight/blob/680406b3dd9cca2108c7f0e204820a09b4e30906/hindsight-api-slim/hindsight_api/engine/cross_encoder.py)
- **[GreptimeDB · JEV SQL](https://github.com/GreptimeTeam/greptimedb)** · 6.7K Star — GreptimeDB 增加由 JEV 判断驱动的实验性 SQL 谓词。 [源码证据](https://github.com/GreptimeTeam/greptimedb/blob/b5199bc59a199875187d808158357a33b09e5f60/src/common/function/src/scalars/jev.rs)
-- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 752 Star — 本地优先的 Agent Memory 类库，可选用 JEV 对召回记忆重新排序。 [源码证据](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
-- **[Kody](https://github.com/kentcdodds/kody)** · 663 Star — 可选的二段检索：先扩大混合召回，再用 Cloudflare Workers AI 上的 `typesafe/jev` Score 重排。 [源码证据](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
-- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 385 Star — 用自然语言搜网页：Jev 选择搜索来源和时间范围，再给返回的链接排序。 [源码证据](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
-- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 358 Star — 包含意图搜索与 Agent Assist 流程的一组 JEV 实验。 [源码证据](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
-- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 348 Star — 把税务文档页面归入预先定义的 IRS 表格目录。 [源码证据](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
-- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 286 Star — 在 PostgreSQL 查询中用自然语言给数据行筛选、分类和排序。 [源码证据](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
-- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 202 Star — 使用 JEV 在大型集合中寻找相关条目的命令行排序工具。 [源码证据](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
-- **[DocJEV](https://github.com/jerryjliu/docjev)** · 201 Star — 带专用 JEV Engine 的文档分类与切分工具。 [源码证据](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
-- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 124 Star — 支持跨语言 AND、OR、NOT 与概率阈值的“按意思搜索”grep。 [源码证据](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
-- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 81 Star — 使用 JEV 判断下一条关系的 Neo4j 图导航器。 [源码证据](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
-- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 81 Star — 一个预览阶段的 PostgreSQL C 扩展，让 SQL 直接调用 Jev 做分类、是非判断和评分。 [源码证据](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
-- **[jegrep](https://github.com/can1357/jegrep)** · 75 Star — 面向实时代码树的语义 grep：用自然语言描述目标，无需建索引即可返回文件与原始行号范围。 [源码证据](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Hippo Memory · JEV Reranker](https://github.com/kitfunso/hippo-memory)** · 756 Star — 本地优先的 Agent Memory 类库，可选用 JEV 对召回记忆重新排序。 [源码证据](https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts)
+- **[Kody](https://github.com/kentcdodds/kody)** · 666 Star — 可选的二段检索：先扩大混合召回，再用 Cloudflare Workers AI 上的 `typesafe/jev` Score 重排。 [源码证据](https://github.com/kentcdodds/kody/blob/ed9275186dc817530dd2289e62c208ac3809391f/packages/worker/src/mcp/tools/search-jev-rerank.ts)
+- **[JEV Search](https://github.com/superagents-lab/jev-search)** · 416 Star — 用自然语言搜网页：Jev 选择搜索来源和时间范围，再给返回的链接排序。 [源码证据](https://github.com/superagents-lab/jev-search/blob/522868762f0637b20bf533f136e930cceb83b9f3/src/lib/typesafe.ts#L48)
+- **[Tax Document Classifier](https://github.com/kyotofin/tax-doc-classifier)** · 399 Star — 把税务文档页面归入预先定义的 IRS 表格目录。 [源码证据](https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts)
+- **[Jev Experiments](https://github.com/dabit3/jev-experiments)** · 370 Star — 包含意图搜索与 Agent Assist 流程的一组 JEV 实验。 [源码证据](https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist)
+- **[Pg JEV](https://github.com/realZachi/pg-jev)** · 314 Star — 在 PostgreSQL 查询中用自然语言给数据行筛选、分类和排序。 [源码证据](https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql)
+- **[DocJEV](https://github.com/jerryjliu/docjev)** · 287 Star — 带专用 JEV Engine 的文档分类与切分工具。 [源码证据](https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py)
+- **[SiftRank · JEV](https://github.com/noperator/siftrank)** · 209 Star — 使用 JEV 在大型集合中寻找相关条目的命令行排序工具。 [源码证据](https://github.com/noperator/siftrank/blob/03e7afe3289a204ea3dcc51613cea91877a651de/pkg/siftrank/jev_provider.go)
+- **[jev-semgrep](https://github.com/uehaj/jev-semgrep)** · 130 Star — 支持跨语言 AND、OR、NOT 与概率阈值的“按意思搜索”grep。 [源码证据](https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md)
+- **[Neo4JEV](https://github.com/jexp/neo4jev)** · 100 Star — 使用 JEV 判断下一条关系的 Neo4j 图导航器。 [源码证据](https://github.com/jexp/neo4jev/blob/d157bbe496eb91813475156942bef1c6badfb342/src/neo4jev/navigator.py)
+- **[jegrep](https://github.com/can1357/jegrep)** · 83 Star — 面向实时代码树的语义 grep：用自然语言描述目标，无需建索引即可返回文件与原始行号范围。 [源码证据](https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs)
+- **[Pg TypeSafe](https://github.com/giuliosmall/pg_typesafe)** · 82 Star — 一个预览阶段的 PostgreSQL C 扩展，让 SQL 直接调用 Jev 做分类、是非判断和评分。 [源码证据](https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c)
[↑ 返回检索入口](#discovery)
-### 安全与审查 (14)
+### 安全与审查 (16)
-- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.3K Star — 在订阅转 API 网关中内置 TypeSafe 客户端，用于内容审核。 [源码证据](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
+- **[Sub2API · JEV Moderation](https://github.com/Wei-Shaw/sub2api)** · 42.4K Star — 在订阅转 API 网关中内置 TypeSafe 客户端，用于内容审核。 [源码证据](https://github.com/Wei-Shaw/sub2api/blob/1c0a69c0ceddb2fd21581c17ab09f6c500b89ba1/backend/internal/pkg/typesafe/client.go)
- **[Agentgateway · JEV Guardrail](https://github.com/agentgateway/agentgateway)** · 5.0K Star — 用 JEV Webhook 检查模型请求与响应的 Guardrail 示例。 [源码证据](https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts)
- **[Latitude · JEV Preclassifier](https://github.com/latitude-dev/latitude-llm)** · 4.7K Star — 选择一段会话应执行哪些检查的可选 JEV 预分类器。 [源码证据](https://github.com/latitude-dev/latitude-llm/blob/6b2484c1d74973877fc7d353b8867c7496568d78/packages/domain/flaggers/src/use-cases/run-jev-preclassifier.ts#L1)
-- **[Agent](https://github.com/AgentiLoop/Agent)** · 622 Star — 原生 macOS Agent 内的可选 Jev 命令风险顾问，配有 TypeSafeKit 客户端。 [源码证据](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
-- **[jev-align](https://github.com/sutro-sh/jev-align)** · 271 Star — 用人工正反例和 GEPA 校准 JEV 分类标准的 CLI。 [源码证据](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
-- **[Abide](https://github.com/coldteadotai/abide)** · 207 Star — 为普通 Linter 难以表达的项目规则增加 Agent 级约束。 [源码证据](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
-- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 177 Star — Interlinked 在编程 Agent 的本地检查之外，提供可选 Jev 判断与证据检查。 [源码证据](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
-- **[Unclutter](https://github.com/kitze/unclutter)** · 176 Star — 用 Jev 帮浏览器扩展识别网页中的广告、促销与订阅弹窗，并保存可复用的隐藏规则。 [源码证据](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
-- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 134 Star — Pi 编码 Agent 扩展：执行前提示工具风险，执行后检查秘密泄露与失败类型。 [源码证据](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
-- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 132 Star — 给 Pi Agent 加项目规则、越界操作、重复失败和完成声明的检查。 [源码证据](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
-- **[Supercov](https://github.com/supercorp-ai/supercov)** · 94 Star — 使用 JEV 做有限质量检查的代码质量与覆盖率工具。 [源码证据](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[fx · TypeSafe Permission Reviewer](https://github.com/vercel-labs/fx)** · 3.1K Star — Zig 编码 Agent，可选 TypeSafe Jev 权限审查器，用于审核待执行工具动作。 [源码证据](https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig)
+- **[Agent](https://github.com/AgentiLoop/Agent)** · 629 Star — 原生 macOS Agent 内的可选 Jev 命令风险顾问，配有 TypeSafeKit 客户端。 [源码证据](https://github.com/AgentiLoop/Agent/blob/078f87ceca1d1190cc73706ac8ec16442e766a27/README.md)
+- **[jev-align](https://github.com/sutro-sh/jev-align)** · 276 Star — 用人工正反例和 GEPA 校准 JEV 分类标准的 CLI。 [源码证据](https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md)
+- **[Abide](https://github.com/coldteadotai/abide)** · 210 Star — 为普通 Linter 难以表达的项目规则增加 Agent 级约束。 [源码证据](https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md)
+- **[Unclutter](https://github.com/kitze/unclutter)** · 203 Star — 用 Jev 帮浏览器扩展识别网页中的广告、促销与订阅弹窗，并保存可复用的隐藏规则。 [源码证据](https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts)
+- **[Interlinked CLI](https://github.com/QuentinCody/interlinked-cli)** · 178 Star — Interlinked 在编程 Agent 的本地检查之外，提供可选 Jev 判断与证据检查。 [源码证据](https://github.com/QuentinCody/interlinked-cli/blob/207330d8131c5203ecc74e9fd4c24ba416463718/src/harness/jev/client.ts#L14)
+- **[Pi JEV](https://github.com/y0usaf/pi-jev)** · 140 Star — Pi 编码 Agent 扩展：执行前提示工具风险，执行后检查秘密泄露与失败类型。 [源码证据](https://github.com/y0usaf/pi-jev/blob/b3478fd4ca1ac8ffcb703f6dc8d6069b555f531e/README.md)
+- **[Pi Warden](https://github.com/DevMortimer/pi-warden)** · 136 Star — 给 Pi Agent 加项目规则、越界操作、重复失败和完成声明的检查。 [源码证据](https://github.com/DevMortimer/pi-warden/blob/e6c801679464b1a9624225114eb5fb443c27d823/src/guard.ts)
+- **[Supercov](https://github.com/supercorp-ai/supercov)** · 106 Star — 使用 JEV 做有限质量检查的代码质量与覆盖率工具。 [源码证据](https://github.com/supercorp-ai/supercov/blob/55f5ce93a239829c224b89e6749991310be91ea4/crates/supercov-cli/src/quality.rs)
+- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 74 Star — 语义代码 Linter，通过零依赖 JEV 客户端批量判断仓库问题。 [源码证据](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
- **[Captaincore](https://github.com/CaptainCore/captaincore)** · 71 Star — WordPress 运维工具 CaptainCore 的 Jev 命令，可询问结构化问题，并给恶意代码扫描结果排复核优先级。 [源码证据](https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go)
-- **[JEV Lint](https://github.com/mizchi/jev-lint)** · 70 Star — 语义代码 Linter，通过零依赖 JEV 客户端批量判断仓库问题。 [源码证据](https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts)
-- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 55 Star — 用 JEV 评估语义 Lint 规则的 Oxlint 插件。 [源码证据](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Oxlint Plugin JEV](https://github.com/wobsoriano/oxlint-plugin-jev)** · 58 Star — 用 JEV 评估语义 Lint 规则的 Oxlint 插件。 [源码证据](https://github.com/wobsoriano/oxlint-plugin-jev/blob/18c5bc9097d88344382a98a78a67698c9c7ecf01/src/jev.ts)
+- **[Canny · Jev Hooks](https://github.com/qkal/Canny)** · 50 Star — Claude Code / Codex Hooks：维护只追加账本，并可选向 Jev 询问 Noul 做建议性检查。 [源码证据](https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts)
[↑ 返回检索入口](#discovery)
-### Agent 工作流 (14)
+### Agent 工作流 (15)
-- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 30.9K Star — 为 Claude Code 判断子 Agent 模型与推理强度的 Mod。 [源码证据](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
+- **[Jev Model Router](https://github.com/davila7/claude-code-templates)** · 31.2K Star — 为 Claude Code 判断子 Agent 模型与推理强度的 Mod。 [源码证据](https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts)
- **[Openwork](https://github.com/different-ai/openwork)** · 23.7K Star — 开源的 cowork 风格 Agent 工作区，可在 CI 与 Skill 流程里跑 JEV 驱动的审查。 [源码证据](https://github.com/different-ai/openwork/blob/80c74d2160034fa1738fc6be78dd24ddfeccfc59/.github/scripts/jev-test-coverage-review.mjs)
- **[Eve](https://github.com/vercel/eve)** · 5.3K Star — Vercel 开源的 Agent 框架，用于打包 Skill、评测与带工具的工作流。 [源码证据](https://github.com/vercel/eve/blob/d004e6d47e9d25d0380c24b5a47b65a18f8b2784/README.md)
- **[RNSkill · JEV Office Gate](https://github.com/Pluviobyte/rnskill)** · 1.6K Star — Agent Skill 合集中包含一个用于有限办公文档检查的 JEV Gate。 [源码证据](https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py)
- **[NewsJack](https://github.com/elvisun/newsjack)** · 1.3K Star — 从实时新闻流中筛选品牌可跟进话题的开源 PR 工作流。 [源码证据](https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer)
-- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 683 Star — 编码 Agent Runtime 使用 JEV 分支处理路由、压缩、记忆与工具结果。 [源码证据](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
-- **[Smithers](https://github.com/smithersai/smithers)** · 418 Star — 以 TypeScript 配置定义工作流的 Agent 框架，内置 JEV 分类路径。 [源码证据](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
-- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 173 Star — 编码 Agent 的 Skill/插件：建议何时压缩上下文以节省 token。 [源码证据](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
-- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 134 Star — Claude Code 插件：在模型看到长 Bash 输出前，用 TypeSafe JEV 裁剪。 [源码证据](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
-- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 115 Star — 把 JEV 判断接入 Claude Code 与 Codex 工具流程的网关。 [源码证据](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
-- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 109 Star — 用 JEV 根据实时会话上下文排序 Agent Skills 的 Rust CLI。 [源码证据](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
+- **[Agent Beacon · Jev Evaluator](https://github.com/asymptote-labs/agent-beacon)** · 1.1K Star — 跨 Harness 的 Agent 记忆系统，可用 Jev 对脱敏会话投影打分后再提升经验。 [源码证据](https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go)
+- **[Distill · JEV Runtime](https://github.com/samuelfaj/distill)** · 684 Star — 编码 Agent Runtime 使用 JEV 分支处理路由、压缩、记忆与工具结果。 [源码证据](https://github.com/samuelfaj/distill/blob/d5f031153cb817f023138a9a0b3fce5516f727c4/crates/codegen/distill-shell/src/jev.rs)
+- **[Smithers](https://github.com/smithersai/smithers)** · 420 Star — 以 TypeScript 配置定义工作流的 Agent 框架，内置 JEV 分类路径。 [源码证据](https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts)
+- **[Compact Adviser](https://github.com/kunchenguid/compact-adviser)** · 180 Star — 编码 Agent 的 Skill/插件：建议何时压缩上下文以节省 token。 [源码证据](https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md)
+- **[jev-gateway](https://github.com/vinilana/jev-gateway)** · 179 Star — 把 JEV 判断接入 Claude Code 与 Codex 工具流程的网关。 [源码证据](https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md)
+- **[JEV Pruner](https://github.com/tamaratran/jev-pruner)** · 140 Star — Claude Code 插件：在模型看到长 Bash 输出前，用 TypeSafe JEV 裁剪。 [源码证据](https://github.com/tamaratran/jev-pruner/blob/47d017c34eab7690b95f075ce6f4839247c5dc0a/src/jev.ts)
+- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 115 Star — 面向 DeepSeek Harness 及兼容编码 Agent Host 的结构化 JEV 决策插件。 [源码证据](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
+- **[Skillranker](https://github.com/Dicklesworthstone/skillranker)** · 114 Star — 用 JEV 根据实时会话上下文排序 Agent Skills 的 Rust CLI。 [源码证据](https://github.com/Dicklesworthstone/skillranker/blob/16743e9f9dee04850d01015d8f5c863a4ed4d0ff/README.md)
- **[Bluenoise](https://github.com/rokcso/bluenoise)** · 90 Star — 为 X/Twitter 过滤帖子与回复的浏览器扩展；默认用本地规则，可选择用 Jev 检查未匹配的回复。 [源码证据](https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts)
-- **[JEV DSH Decision](https://github.com/Devin-AXIS/jev-dsh-decision)** · 78 Star — 面向 DeepSeek Harness 及兼容编码 Agent Host 的结构化 JEV 决策插件。 [源码证据](https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs)
-- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 62 Star — 让 JEV 判断哪些历史应保留、截断或丢弃的上下文清理器。 [源码证据](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
+- **[Save Token JEV Clean](https://github.com/IAmUnbounded/save-token-jev-clean)** · 66 Star — 让 JEV 判断哪些历史应保留、截断或丢弃的上下文清理器。 [源码证据](https://github.com/IAmUnbounded/save-token-jev-clean/blob/a7007354a8d3747f06ff82130561edb2822a17df/src/client.ts)
[↑ 返回检索入口](#discovery)
@@ -246,15 +261,15 @@ npx skills add BeatAPI/awesome-jev
### 界面与自动化 (9)
-- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.0K Star — 用 JEV 从预定义组件与属性中进行选择的 UI 组合实验。 [源码证据](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
-- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 1.6K Star — 读取可见聊天上下文、给出候选回复但不自动发送的 Android 对话副驾。 [源码证据](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
+- **[json-render · JEV Compose](https://github.com/vercel-labs/json-render)** · 18.1K Star — 用 JEV 从预定义组件与属性中进行选择的 UI 组合实验。 [源码证据](https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts)
+- **[JEV Chat Jarvis](https://github.com/jev-chat/jev-chat-jarvis)** · 4.1K Star — 读取可见聊天上下文、给出候选回复但不自动发送的 Android 对话副驾。 [源码证据](https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt)
- **[Aiavatarkit](https://github.com/uezo/aiavatarkit)** · 678 Star — AIAvatarKit 的可选 Jev 组件根据转写内容判断用户是否结束发言。 [源码证据](https://github.com/uezo/aiavatarkit/blob/38b617b8b9269939734e70ef503d7ea6976acdbd/aiavatar/sts/vad/turn_end_gates/jev.py#L172)
-- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 642 Star — 由 JEV 或 Gemini Live 驱动的本地 macOS 电脑操作伴侣。 [源码证据](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
-- **[Notra](https://github.com/usenotra/notra)** · 207 Star — 把工作产物转成可发布内容的产品，并挂接 Agent Skill。 [源码证据](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
+- **[TipTour macOS](https://github.com/milind-soni/tiptour-macos)** · 654 Star — 由 JEV 或 Gemini Live 驱动的本地 macOS 电脑操作伴侣。 [源码证据](https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md)
+- **[Notra](https://github.com/usenotra/notra)** · 213 Star — 把工作产物转成可发布内容的产品，并挂接 Agent Skill。 [源码证据](https://github.com/usenotra/notra/blob/f792a620ef1066b34665ec84c6f55fc9ca5d9954/README.md)
- **[OpenWhisper](https://github.com/Knuckles92/OpenWhisper)** · 187 Star — 语音听写与会议记录应用，可选用 Jev 检查话题变化、面向记录助手的指令和敏感文本。 [源码证据](https://github.com/Knuckles92/OpenWhisper/blob/9e83653df183096104769e302a3c907cb277c551/README.md)
-- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 84 Star — 由 JEV 仅从代码提供的回复与工具选项中选择的工具型聊天界面。 [源码证据](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[JEV Chat](https://github.com/w3cj/jev-chat)** · 91 Star — 由 JEV 仅从代码提供的回复与工具选项中选择的工具型聊天界面。 [源码证据](https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts)
+- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 91 Star — 结合实时音频与字幕由 Jev 驱动的 YouTube 视频赞助广告片段检测与自动跳过扩展。 [源码证据](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
- **[Jevmeter](https://github.com/ChetasLua/jevmeter)** · 81 Star — 把视频转成带评分仪表的视频成片：Jev 按选定规则给字幕句子评分，再由渲染器叠加显示。 [源码证据](https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py)
-- **[Youtube Sponsor Detection](https://github.com/trungdq88/youtube-sponsor-detection)** · 81 Star — 结合实时音频与字幕由 Jev 驱动的 YouTube 视频赞助广告片段检测与自动跳过扩展。 [源码证据](https://github.com/trungdq88/youtube-sponsor-detection/blob/de01f0568d043035889a296a61ce21e0accc8b16/extension/lib/jev.js#L1-L541)
[↑ 返回检索入口](#discovery)
@@ -262,61 +277,63 @@ npx skills add BeatAPI/awesome-jev
### 开发者工具 (35)
-- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.0K Star — 压缩 Claude Code 工具历史，同时原样保留仍有价值的内容。 [源码证据](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
-- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.6K Star — 面向 TypeSafe System One / JEV API 的官方 Agent Skill 集合。 [源码证据](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
-- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.0K Star — 收录公开 JEV 项目、集成与讨论的大型社区目录。 [源码证据](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
-- **[Celesto](https://github.com/CelestoAI/celesto)** · 958 Star — Celesto 的 PR 审查示例在沙盒中准备检查，再比较普通模型与 Jev 对候选问题的判断。 [源码证据](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
-- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 810 Star — 终端生成 CLI，并附带可从命令行创作内容的 Agent Skill。 [源码证据](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
-- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 772 Star — 有证据支撑的 JEV 用例、模式、提示词与起步代码手册。 [源码证据](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
-- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 628 Star — 聚焦 JEV 与 TypeSafe System One 工具的精选目录。 [源码证据](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
-- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 502 Star — 分阶段检查 Git diff 或整个代码库，在本地面板里展示可复核的审查线索。 [源码证据](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
-- **[Foreman](https://github.com/thruwire/foreman)** · 472 Star — 独立监督环读取编码工人的 diff、日志和测试，用 Jev Noul 判断卡住、跑偏、该验证，再由 Python 策略干预。 [源码证据](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
-- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 431 Star — TypeSafe、System One 与 JEV 的官方与社区资源精选列表。 [源码证据](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
-- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 431 Star — 带源码依据的 TypeSafe JEV 项目、SDK、演示与评测指南。 [源码证据](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
-- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 398 Star — 一套 Agent Skills 与工作流，内含仅依赖标准库的 JEV 决策客户端。 [源码证据](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
-- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 334 Star — 有源码依据的 JEV 生态雷达，含白话发现页与 GitHub 同步。 [源码证据](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
-- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 316 Star — 基于源码核对的 JEV / TypeSafe System One 项目精选列表。 [源码证据](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
-- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 281 Star — OrchestKit 可选用 Jev 给编程会话分类，符合阈值时用结果决定会话颜色。 [源码证据](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
-- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 257 Star — JEV 应用、类库与 System One 资源精选清单。 [源码证据](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
-- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 245 Star — 给 Agent 提供核对引用、筛查内容、查找、重排、分类、比较和提取等八个 MCP 判断工具。 [源码证据](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
-- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 227 Star — 让 Claude Code、Claude Desktop、Codex 和 Pi 通过 MCP 或扩展向 Jev 提问，获取结构化判断。 [源码证据](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
-- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 196 Star — 供编码 Agent 使用的本地 MCP 代码质量检查器，返回多个维度的结构化评分。 [源码证据](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
-- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 187 Star — 有源码依据的 JEV 目录，并附带可复用的纯 JEV GitHub Review 工作流。 [源码证据](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
-- **[Perch](https://github.com/lakeday-org/perch)** · 167 Star — 语义代码检查：用 JEV / System One 判断某条发现是否该触发。 [源码证据](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
-- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 159 Star — 汇总 System One 论文、开放复现与独立评测的 Gallery。 [源码证据](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
-- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 148 Star — 社区维护的多语言 JEV 开源项目目录。 [源码证据](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Fast Jev Compaction](https://github.com/tamaratran/fast-jev-compaction)** · 6.3K Star — 压缩 Claude Code 工具历史，同时原样保留仍有价值的内容。 [源码证据](https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts)
+- **[TypeSafe Skills](https://github.com/typesafe-ai/skills)** · 1.9K Star — 面向 TypeSafe System One / JEV API 的官方 Agent Skill 集合。 [源码证据](https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md)
+- **[Awesome JEV (yibie)](https://github.com/yibie/awesome-jev)** · 1.3K Star — 收录公开 JEV 项目、集成与讨论的大型社区目录。 [源码证据](https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md)
+- **[Celesto](https://github.com/CelestoAI/celesto)** · 966 Star — Celesto 的 PR 审查示例在沙盒中准备检查，再比较普通模型与 Jev 对候选问题的判断。 [源码证据](https://github.com/CelestoAI/celesto/blob/fff7bb567752baee4ce195fcb78b5ee5e4da889f/examples/pr-review-jev/models.py#L96)
+- **[AI CLI](https://github.com/vercel-labs/ai-cli)** · 811 Star — 终端生成 CLI，并附带可从命令行创作内容的 Agent Skill。 [源码证据](https://github.com/vercel-labs/ai-cli/blob/6a0ed5d04ea60ee536029d469499b82898d1b214/skills/ai-cli/SKILL.md)
+- **[Awesome JEV by TypeSafe](https://github.com/Anil-matcha/awesome-jev-by-typesafe)** · 806 Star — 有证据支撑的 JEV 用例、模式、提示词与起步代码手册。 [源码证据](https://github.com/Anil-matcha/awesome-jev-by-typesafe/blob/0f4a1eadcdd70f4fc1cf8eadcf72794dfe416095/docs/jev-use-case-playbook.md)
+- **[Awesome JEV Tools](https://github.com/v-modal/awesome-jev-tools)** · 658 Star — 聚焦 JEV 与 TypeSafe System One 工具的精选目录。 [源码证据](https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md)
+- **[Awesome JEV (HeyJunPenn)](https://github.com/heyjunpenn/awesome-jev)** · 589 Star — 社区维护的多语言 JEV 开源项目目录。 [源码证据](https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md)
+- **[Jev Review](https://github.com/devagrawal09/jev-review)** · 556 Star — 分阶段检查 Git diff 或整个代码库，在本地面板里展示可复核的审查线索。 [源码证据](https://github.com/devagrawal09/jev-review/blob/31f89602797fb7bea007f8a480bf368bf564954e/src/review/judgments.ts#L38)
+- **[Foreman](https://github.com/thruwire/foreman)** · 517 Star — 独立监督环读取编码工人的 diff、日志和测试，用 Jev Noul 判断卡住、跑偏、该验证，再由 Python 策略干预。 [源码证据](https://github.com/thruwire/foreman/blob/209182dac7a3467033fd093ab4ca47d21279984a/src/foreman/foreman/jev.py)
+- **[Awesome TypeSafe](https://github.com/AbdelStark/awesome-typesafe)** · 476 Star — TypeSafe、System One 与 JEV 的官方与社区资源精选列表。 [源码证据](https://github.com/AbdelStark/awesome-typesafe/blob/d44dc0a1e26d41bc7f543d39b4b62ae5cb9f13c8/README.md)
+- **[Awesome TypeSafe JEV](https://github.com/AbdelStark/awesome-typesafe-jev)** · 476 Star — 带源码依据的 TypeSafe JEV 项目、SDK、演示与评测指南。 [源码证据](https://github.com/AbdelStark/awesome-typesafe-jev/blob/9384099af36cd439b7f0bd4311560ac02eeabe7c/README.md)
+- **[Awesome Jev Skills](https://github.com/wuyoscar/jev-skill)** · 437 Star — 一套 Agent Skills 与工作流，内含仅依赖标准库的 JEV 决策客户端。 [源码证据](https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py)
+- **[Awesome JEV Projects](https://github.com/logicrw/awesome-jev-projects)** · 398 Star — 有源码依据的 JEV 生态雷达，含白话发现页与 GitHub 同步。 [源码证据](https://github.com/logicrw/awesome-jev-projects/blob/d1b1b8d6c68c1754425778606d20d2cb14b09895/SKILL.md)
+- **[Awesome JEV (cobanov)](https://github.com/cobanov/awesome-jev)** · 354 Star — 基于源码核对的 JEV / TypeSafe System One 项目精选列表。 [源码证据](https://github.com/cobanov/awesome-jev/blob/2c8c34a7a89499a0cec3ca7de3a23fec3ea9dc97/README.md)
+- **[Awesome JEV (AnotiaWang)](https://github.com/AnotiaWang/awesome-jev)** · 303 Star — JEV 应用、类库与 System One 资源精选清单。 [源码证据](https://github.com/AnotiaWang/awesome-jev/blob/04e65b59a936b2fed0d40eaec71a0a11e33bb422/README.md)
+- **[JEV MCP](https://github.com/jkudish/jev-mcp)** · 290 Star — 给 Agent 提供核对引用、筛查内容、查找、重排、分类、比较和提取等八个 MCP 判断工具。 [源码证据](https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127)
+- **[Orchestkit](https://github.com/yonatangross/orchestkit)** · 283 Star — OrchestKit 可选用 Jev 给编程会话分类，符合阈值时用结果决定会话颜色。 [源码证据](https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts)
+- **[TypeSafe MCP](https://github.com/itsmostafa/typesafe-mcp)** · 262 Star — 让 Claude Code、Claude Desktop、Codex 和 Pi 通过 MCP 或扩展向 Jev 提问，获取结构化判断。 [源码证据](https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md)
+- **[Awesome JEV Gallery](https://github.com/OmniJev/awesome-jev-gallery)** · 208 Star — 汇总 System One 论文、开放复现与独立评测的 Gallery。 [源码证据](https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md)
+- **[JEV Review](https://github.com/NiazMorshed2007/jev-review)** · 206 Star — 供编码 Agent 使用的本地 MCP 代码质量检查器，返回多个维度的结构化评分。 [源码证据](https://github.com/NiazMorshed2007/jev-review/blob/57690af54ef7d862c2483342c1e61c14dffcf727/README.md)
+- **[Awesome JEV (fatwang2)](https://github.com/fatwang2/awesome-jev)** · 198 Star — 有源码依据的 JEV 目录，并附带可复用的纯 JEV GitHub Review 工作流。 [源码证据](https://github.com/fatwang2/awesome-jev/blob/f036095ba8203001d40be955f3486cd3e0d0a81b/.github/workflows/jev-review.yml)
+- **[Perch](https://github.com/lakeday-org/perch)** · 170 Star — 语义代码检查：用 JEV / System One 判断某条发现是否该触发。 [源码证据](https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js)
+- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 149 Star — awesomejev.com 的源数据目录，覆盖仓库、集成、演示与研究。 [源码证据](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
+- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 130 Star — 围绕类型化、置信度感知决策整理的 TypeSafe JEV 资源列表。 [源码证据](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Building with JEV Skill](https://github.com/dbreunig/building-with-jev-skill)** · 128 Star — 帮助编写与改进调用 JEV / System One 程序的 Agent Skill。 [源码证据](https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md)
-- **[Awesome Jev](https://github.com/hellogumbo/awesome-jev)** · 127 Star — awesomejev.com 的源数据目录，覆盖仓库、集成、演示与研究。 [源码证据](https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json)
-- **[Awesome JEV TypeSafe](https://github.com/valentynkit/awesome-jev-typesafe)** · 125 Star — 围绕类型化、置信度感知决策整理的 TypeSafe JEV 资源列表。 [源码证据](https://github.com/valentynkit/awesome-jev-typesafe/blob/7eec7360d502826b51d1c8862c9fd2d76eecf6e4/readme.md)
- **[Awesome JEV (Promethe-us)](https://github.com/Promethe-us/awesome-jev)** · 122 Star — 带来源核验的双语 JEV 资源、项目、研究与动态合集。 [源码证据](https://github.com/Promethe-us/awesome-jev/blob/636623538eda3c363b8c396880ccdc8aba70d424/README.md)
-- **[Taskuary](https://github.com/ldbumble/taskuary)** · 116 Star — Taskuary 的可选 Jev 判断模块，对任务运行状态检查用户定义的条件。 [源码证据](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
-- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 111 Star — 为编码 Agent 提供带专用适配器的有限 JEV 工作流。 [源码证据](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
-- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 100 Star — 涵盖 JEV 用例、项目、SDK、工具与学习资料的精选指南。 [源码证据](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
-- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 96 Star — 类似于 Fish 终端样式的 Zsh 历史命令建议工具，利用 Jev 对已有历史记录根据当前上下文进行智能打分排序。 [源码证据](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
-- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 84 Star — 带原始来源的 JEV 已发布 Demo、仓库、自报指标与模型限制目录。 [源码证据](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
-- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 73 Star — 带可运行类型化决策示例的 JEV 精选资源列表。 [源码证据](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
-- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 71 Star — 用于比较 JEV、开放决策模型、分类器与重排模型的可复现基准。 [源码证据](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
-- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 59 Star — 中文 JEV 生态指南，包含精选项目、实践教程、价格信息与独立限制说明。 [源码证据](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
+- **[Taskuary](https://github.com/ldbumble/taskuary)** · 118 Star — Taskuary 的可选 Jev 判断模块，对任务运行状态检查用户定义的条件。 [源码证据](https://github.com/ldbumble/taskuary/blob/4ad29d7b292a7899767338cfcc83b2dde8f43330/README.md)
+- **[Awesome JEV Use Cases](https://github.com/walidboulanouar/awesome-jev-use-cases)** · 114 Star — 带原始来源的 JEV 已发布 Demo、仓库、自报指标与模型限制目录。 [源码证据](https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md)
+- **[Awesome JEV (kraayenjon)](https://github.com/kraayenjon/awesome-jev)** · 113 Star — 涵盖 JEV 用例、项目、SDK、工具与学习资料的精选指南。 [源码证据](https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md)
+- **[Stanley Code](https://github.com/devagrawal09/stanley-code)** · 113 Star — 为编码 Agent 提供带专用适配器的有限 JEV 工作流。 [源码证据](https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts)
+- **[JEV Shell History](https://github.com/mrnugget/jev-shell-history)** · 104 Star — 类似于 Fish 终端样式的 Zsh 历史命令建议工具，利用 Jev 对已有历史记录根据当前上下文进行智能打分排序。 [源码证据](https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196)
+- **[JevBench](https://github.com/fstandhartinger/jevbench)** · 83 Star — 用于比较 JEV、开放决策模型、分类器与重排模型的可复现基准。 [源码证据](https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py)
+- **[Awesome JEV (AppitStudio)](https://github.com/AppitStudio/awesome-jev)** · 78 Star — 带可运行类型化决策示例的 JEV 精选资源列表。 [源码证据](https://github.com/AppitStudio/awesome-jev/blob/5d654f24aecf992617bbc45e76f32a36eba62f65/README.md)
+- **[Awesome JEV ZH](https://github.com/yzfly/awesome-jev-zh)** · 60 Star — 中文 JEV 生态指南，包含精选项目、实践教程、价格信息与独立限制说明。 [源码证据](https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md)
[↑ 返回检索入口](#discovery)
-### 垂直工具 (13)
+### 垂直工具 (15)
- **[AI Hedge Fund · JEV Adapter](https://github.com/virattt/ai-hedge-fund)** · 63.7K Star — 在教育型 AI 对冲基金原型中加入可选 JEV 适配器，用于结构化策略判断。 [源码证据](https://github.com/virattt/ai-hedge-fund/blob/154a8b2f46dca0f40764d814e4e747b0ad71f4c4/hedge_fund/llm/client.py)
- **[QuantDinger · JEV Gate](https://github.com/OpenByteInc/QuantDinger)** · 12.0K Star — 在部分真实交易入场前增加 JEV 决策闸门的开源交易系统。 [源码证据](https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py)
-- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 1.9K Star — 可在 Monad 每个区块请求一次 JEV 交易判断的做市实验。 [源码证据](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
-- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 485 Star — 由规划模型设定目标、JEV 选择合法游戏动作的 Minecraft Agent。 [源码证据](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
-- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 338 Star — 从 NES 模拟器 RAM 和状态数据中提取环境，让 Jev 选择超级马力欧的手柄按键。 [源码证据](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
-- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 170 Star — 面向具身 Agent 的评测栈，用于比较 JEV、本地模型与通用模型决策策略。 [源码证据](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
-- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 155 Star — 在浏览器里开一辆小车，让 Jev 从提前算好的路线和速度里选下一步。 [源码证据](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
-- **[Dasheng](https://github.com/wquguru/dasheng)** · 119 Star — 把流式 ASR 与 JEV 逐词判断结合的朗读练习应用。 [源码证据](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
-- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 119 Star — MuJoCo 无人机仿真实验：从相机缓冲区提取场景，Jev 提供战术动作建议。 [源码证据](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
-- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 103 Star — 智能体工具选择基准测试平台，在包含 100 个模拟工具的个人助理环境下对比大模型直接选工具与 Jev 路由的效率。 [源码证据](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
+- **[JEV Trader](https://github.com/jarrodwatts/jev-trader)** · 2.1K Star — 可在 Monad 每个区块请求一次 JEV 交易判断的做市实验。 [源码证据](https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts)
+- **[Astra + JEV Minecraft Agent](https://github.com/rmalde/minecraft-agent)** · 524 Star — 由规划模型设定目标、JEV 选择合法游戏动作的 Minecraft Agent。 [源码证据](https://github.com/rmalde/minecraft-agent/blob/78b40ed59514e5e2abde33a05ce398ecb2c39e05/models.mjs)
+- **[TypeSafe Mario](https://github.com/fhshaik/typesafe-mario)** · 355 Star — 从 NES 模拟器 RAM 和状态数据中提取环境，让 Jev 选择超级马力欧的手柄按键。 [源码证据](https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md)
+- **[quackd · Jev Stepper](https://github.com/rokbenko/quackd)** · 229 Star — 机器人编排 CLI，可选 System One 步进器在已许可的离散机器人调用中做选择。 [源码证据](https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py)
+- **[Embodied JEV](https://github.com/FBddcz/embodied-jev)** · 191 Star — 面向具身 Agent 的评测栈，用于比较 JEV、本地模型与通用模型决策策略。 [源码证据](https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py)
+- **[Jevpilot](https://github.com/standardagents/jevpilot)** · 169 Star — 在浏览器里开一辆小车，让 Jev 从提前算好的路线和速度里选下一步。 [源码证据](https://github.com/standardagents/jevpilot/blob/e1beeb13b9a928fb76f167f86af584f4ce9cf180/README.md)
+- **[Dasheng](https://github.com/wquguru/dasheng)** · 135 Star — 把流式 ASR 与 JEV 逐词判断结合的朗读练习应用。 [源码证据](https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js)
+- **[JEV Drone](https://github.com/RomanSlack/jev-drone)** · 134 Star — MuJoCo 无人机仿真实验：从相机缓冲区提取场景，Jev 提供战术动作建议。 [源码证据](https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184)
+- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 110 Star — 以 JEV 作为有限决策层的 Hyperliquid 实盘交易 Agent。 [源码证据](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
+- **[JEV Eval Agent](https://github.com/vinilana/jev-eval-agent)** · 105 Star — 智能体工具选择基准测试平台，在包含 100 个模拟工具的个人助理环境下对比大模型直接选工具与 Jev 路由的效率。 [源码证据](https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154)
- **[Formanator](https://github.com/timrogers/formanator)** · 99 Star — 支持可选 JEV 收据分类的福利报销 CLI 与 MCP 客户端。 [源码证据](https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs)
-- **[JEV Trade](https://github.com/aowang-ai/jev-trade)** · 95 Star — 以 JEV 作为有限决策层的 Hyperliquid 实盘交易 Agent。 [源码证据](https://github.com/aowang-ai/jev-trade/blob/df2c9656324a8a75996eb0612de7adcfe3ce6f89/src/model.ts)
-- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 69 Star — 在再平衡决策服务中使用 JEV 的自动化流动性 Agent。 [源码证据](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[Prism Liquidity Agent](https://github.com/irfndi/prism-liquidity-agent)** · 79 Star — 在再平衡决策服务中使用 JEV 的自动化流动性 Agent。 [源码证据](https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts)
+- **[HA-Jev](https://github.com/AboveColin/HA-Jev)** · 50 Star — Home Assistant 集成：把关于家庭状态的类型化 Jev 问题变成传感器与自动化动作。 [源码证据](https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py)
[↑ 返回检索入口](#discovery)
diff --git a/data/projects.json b/data/projects.json
index 808a59a..ca41488 100644
--- a/data/projects.json
+++ b/data/projects.json
@@ -1,10 +1,10 @@
{
"schemaVersion": 3,
- "capturedAt": "2026-09-22",
+ "capturedAt": "2026-09-23",
"source": {
"name": "BeatAPI Awesome JEV curation",
- "batch": "2026-09-22-jevbest-star50-source-review",
- "policy": "Project-level deduplication, including repository redirects. Public JEV-related GitHub repositories at or above 50 stars are included only after fixed-commit source review. Discovery includes jevbest.com and direct upstream repository inspection. Runtime behavior is not independently verified."
+ "batch": "2026-09-23-star50-cross-search-and-source-review",
+ "policy": "Project-level deduplication. Public JEV-related GitHub repositories at or above 50 stars are included only after fixed-commit source review. Discovery includes GitHub repository search, competitor catalogues (AbdelStark/awesome-typesafe-jev, heyjunpenn/awesome-jev), and direct source inspection. Runtime behavior is not independently verified."
},
"projects": [
{
@@ -16,7 +16,7 @@
"evidenceUrl": "https://github.com/langchain-ai/langchain/blob/eba445b7563d1709427bd8072892975a6ea59fdc/libs/partners/typesafe/langchain_typesafe/classifier.py",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 146831,
+ "starsAtCapture": 146901,
"summary": {
"en": "An optional JEV classifier integration for Python LangChain workflows.",
"zh": "面向 Python LangChain 工作流的可选 JEV 分类器集成。",
@@ -44,7 +44,7 @@
"evidenceUrl": "https://github.com/virattt/ai-hedge-fund/blob/154a8b2f46dca0f40764d814e4e747b0ad71f4c4/hedge_fund/llm/client.py",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 63657,
+ "starsAtCapture": 63672,
"summary": {
"en": "An educational hedge-fund prototype with an optional JEV adapter for structured strategy judgments.",
"zh": "在教育型 AI 对冲基金原型中加入可选 JEV 适配器，用于结构化策略判断。",
@@ -72,7 +72,7 @@
"evidenceUrl": "https://github.com/BerriAI/litellm/blob/56116079c8022da0e8f7ff9ccb017ad5aca5aed2/litellm/router_strategy/complexity_router/jev_classifier.py#L70",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 59361,
+ "starsAtCapture": 59429,
"summary": {
"en": "LiteLLM can use JEV inside its complexity-based model router.",
"zh": "LiteLLM 可在按复杂度路由模型的策略中使用 JEV。",
@@ -96,7 +96,7 @@
"name": "Sub2API · JEV Moderation",
"owner": "Wei-Shaw",
"category": "safety-review",
- "starsAtCapture": 42323,
+ "starsAtCapture": 42435,
"language": "Go",
"license": "LGPL-3.0",
"summary": {
@@ -124,7 +124,7 @@
"name": "OpenViking · JEV Rerank",
"owner": "volcengine",
"category": "search-data",
- "starsAtCapture": 38371,
+ "starsAtCapture": 38481,
"language": "Python",
"license": "AGPL-3.0",
"summary": {
@@ -156,7 +156,7 @@
"evidenceUrl": "https://github.com/can1357/oh-my-pi/blob/78b753124d11f8dd3ae73e2524125890ff7c977e/packages/ai/src/judgment/typesafe.ts#L4",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 32384,
+ "starsAtCapture": 32810,
"summary": {
"en": "A coding agent with an optional TypeSafe judgment provider for bounded workflow decisions.",
"zh": "在编码 Agent 中提供可选 TypeSafe 判断器，处理有边界的工作流决策。",
@@ -184,7 +184,7 @@
"evidenceUrl": "https://github.com/davila7/claude-code-templates/blob/61bfcd1586bf1076f6d3cfa0436317c912811e6c/cli-tool/components/mods/productivity/jev-model-router/hooks/jev-model-router.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 30897,
+ "starsAtCapture": 31179,
"summary": {
"en": "A Claude Code mod that classifies subagent model and reasoning-effort needs.",
"zh": "为 Claude Code 判断子 Agent 模型与推理强度的 Mod。",
@@ -214,7 +214,7 @@
"evidenceUrl": "https://github.com/ComposioHQ/composio/blob/4b5920bf7aa55c8a44657b060d4bd25ce7b13a9a/ts/packages/providers/typesafe/src/decide.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 30279,
+ "starsAtCapture": 30289,
"summary": {
"en": "A TypeSafe provider that uses JEV to choose among tools and bounded argument options.",
"zh": "用 JEV 在工具与有限参数选项之间做选择的 TypeSafe Provider。",
@@ -242,7 +242,7 @@
"evidenceUrl": "https://github.com/vercel/ai/blob/73ec7015edd4f04ca9144ce93a8a037a731e5db8/packages/typesafe-ai/src/typesafe-ai-evaluation-model.ts",
"language": "TypeScript",
"license": "Mixed",
- "starsAtCapture": 26887,
+ "starsAtCapture": 26900,
"summary": {
"en": "The AI SDK TypeSafe provider exposes JEV through the shared evaluate interface.",
"zh": "AI SDK 的 TypeSafe Provider 通过统一 evaluate 接口调用 JEV。",
@@ -270,7 +270,7 @@
"evidenceUrl": "https://github.com/trycua/cua/blob/83f142c4290a0f7d9ed545ae8532858c6e4f8145/libs/cua-driver/examples/jev-use/python/jev_adapter.py#L11",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 25786,
+ "starsAtCapture": 26009,
"summary": {
"en": "A computer-use example pairing Cua Driver observation and execution with bounded JEV choices.",
"zh": "把 Cua Driver 的观察与执行能力和 JEV 的有限动作选择结合起来。",
@@ -294,7 +294,7 @@
"name": "Hindsight · TypeSafe Rerank",
"owner": "vectorize-io",
"category": "search-data",
- "starsAtCapture": 24801,
+ "starsAtCapture": 25263,
"language": "Python",
"license": "MIT",
"summary": {
@@ -331,7 +331,7 @@
},
"language": "TypeScript",
"license": "Not declared",
- "starsAtCapture": 23687,
+ "starsAtCapture": 23700,
"summary": {
"en": "An open-source cowork-style agent workspace that can run JEV-backed reviews inside CI and skill workflows.",
"zh": "开源的 cowork 风格 Agent 工作区，可在 CI 与 Skill 流程里跑 JEV 驱动的审查。",
@@ -354,7 +354,7 @@
"evidenceUrl": "https://github.com/pydantic/pydantic-ai/blob/c4898abb54dc25ae6f6aef208a4c0661b30a455e/pydantic_ai_slim/pydantic_ai/models/typesafe.py",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 20109,
+ "starsAtCapture": 20124,
"summary": {
"en": "A TypeSafe model integration that maps supported structured outputs to JEV questions.",
"zh": "把受支持的结构化输出映射为 JEV 问题的 TypeSafe 模型集成。",
@@ -382,7 +382,7 @@
"evidenceUrl": "https://github.com/elizaOS/eliza/blob/ebc808e3a67fb941e29153d89fc896524d32fe3c/packages/agent/src/services/typesafe/client.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 19406,
+ "starsAtCapture": 19410,
"summary": {
"en": "An optional TypeSafe HTTP adapter in the Eliza agent operating system.",
"zh": "Eliza Agent 操作系统中的可选 TypeSafe HTTP 适配器。",
@@ -410,7 +410,7 @@
"evidenceUrl": "https://github.com/langchain-ai/langchainjs/blob/206d8b992bcf90ce7d46f2158f1ad85fc1d826c0/libs/providers/langchain-typesafe/src/classifier.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 18214,
+ "starsAtCapture": 18220,
"summary": {
"en": "A TypeSafeClassifier integration for JavaScript LangChain workflows.",
"zh": "面向 JavaScript LangChain 工作流的 TypeSafeClassifier 集成。",
@@ -429,34 +429,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "json-render-jev",
- "name": "json-render · JEV Compose",
- "owner": "vercel-labs",
- "category": "interfaces",
- "repoUrl": "https://github.com/vercel-labs/json-render",
- "evidenceUrl": "https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts",
- "language": "TypeScript",
- "license": "Apache-2.0",
- "starsAtCapture": 17976,
- "summary": {
- "en": "A JEV UI-composition experiment that selects from predefined components and properties.",
- "zh": "用 JEV 从预定义组件与属性中进行选择的 UI 组合实验。",
- "ja": "事前定義されたコンポーネントとプロパティから選択する JEV UI 構成の実験。"
- },
- "decision": {
- "en": "Evaluates component configurations before local code assembles the UI specification.",
- "zh": "先判断组件配置，再由本地代码组装 UI 规范。",
- "ja": "ローカル コードが UI 仕様をアセンブルする前に、コンポーネント構成を評価します。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "jev-ultrafast",
"name": "Jev Ultrafast",
@@ -473,7 +445,7 @@
},
"language": "Python",
"license": "MIT",
- "starsAtCapture": 16335,
+ "starsAtCapture": 18230,
"summary": {
"en": "A browser agent that uses JEV to choose an action and matching DOM element, calling a text model only when input text is needed.",
"zh": "用 JEV 一次选出浏览器动作与对应 DOM 元素，只在需要输入文本时再调用文本模型。",
@@ -488,29 +460,29 @@
"runtimeVerified": false
},
{
- "id": "quantdinger-jev",
- "name": "QuantDinger · JEV Gate",
- "owner": "OpenByteInc",
- "category": "domain-tools",
- "repoUrl": "https://github.com/OpenByteInc/QuantDinger",
- "evidenceUrl": "https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py",
- "language": "Python",
+ "id": "json-render-jev",
+ "name": "json-render · JEV Compose",
+ "owner": "vercel-labs",
+ "category": "interfaces",
+ "repoUrl": "https://github.com/vercel-labs/json-render",
+ "evidenceUrl": "https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts",
+ "language": "TypeScript",
"license": "Apache-2.0",
- "starsAtCapture": 11959,
+ "starsAtCapture": 18112,
"summary": {
- "en": "An open-source trading OS with a JEV decision gate before selected live entries.",
- "zh": "在部分真实交易入场前增加 JEV 决策闸门的开源交易系统。",
- "ja": "選択されたライブエントリーの前に JEV 決定ゲートを備えたオープンソースの取引 OS。"
+ "en": "A JEV UI-composition experiment that selects from predefined components and properties.",
+ "zh": "用 JEV 从预定义组件与属性中进行选择的 UI 组合实验。",
+ "ja": "事前定義されたコンポーネントとプロパティから選択する JEV UI 構成の実験。"
},
"decision": {
- "en": "Checks evidence quality, signal consistency, exposure, and budget before an order reaches the exchange.",
- "zh": "在订单进入交易所前检查证据质量、信号一致性、敞口与预算。",
- "ja": "注文が取引所に届く前に、証拠の品質、信号の一貫性、露出、予算をチェックします。"
+ "en": "Evaluates component configurations before local code assembles the UI specification.",
+ "zh": "先判断组件配置，再由本地代码组装 UI 规范。",
+ "ja": "ローカル コードが UI 仕様をアセンブルする前に、コンポーネント構成を評価します。"
},
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py"
+ "url": "https://github.com/vercel-labs/json-render/blob/3ad381881194e7011ad3ccd6d668033495a06c29/apps/web/lib/jev/compose.ts"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -525,7 +497,7 @@
"modelUrl": "https://huggingface.co/convaiinnovations/laya",
"language": "Python",
"license": "Apache-2.0",
- "starsAtCapture": 11547,
+ "starsAtCapture": 17126,
"summary": {
"en": "Multilingual non-autoregressive System 1 decision engine: typed choice/score/noul in one forward pass, with a router across checkpoints.",
"zh": "多语言非自回归 System 1 决策引擎：单次前向完成类型化 choice/score/noul，并由路由在检查点间选型。",
@@ -544,6 +516,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "quantdinger-jev",
+ "name": "QuantDinger · JEV Gate",
+ "owner": "OpenByteInc",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/OpenByteInc/QuantDinger",
+ "evidenceUrl": "https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py",
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 12024,
+ "summary": {
+ "en": "An open-source trading OS with a JEV decision gate before selected live entries.",
+ "zh": "在部分真实交易入场前增加 JEV 决策闸门的开源交易系统。",
+ "ja": "選択されたライブエントリーの前に JEV 決定ゲートを備えたオープンソースの取引 OS。"
+ },
+ "decision": {
+ "en": "Checks evidence quality, signal consistency, exposure, and budget before an order reaches the exchange.",
+ "zh": "在订单进入交易所前检查证据质量、信号一致性、敞口与预算。",
+ "ja": "注文が取引所に届く前に、証拠の品質、信号の一貫性、露出、予算をチェックします。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/OpenByteInc/QuantDinger/blob/12c04eb2cdb8a9d08dc84502f5261ec3f1c56bf7/backend_api_python/app/services/ai_decision_filter.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "openchamber-jev-router",
"name": "OpenChamber · JEV Router",
@@ -553,7 +553,7 @@
"evidenceUrl": "https://github.com/openchamber/openchamber/blob/614d7f76e581a132a86575c03d3fa9aad5e624b6/packages/web/server/lib/routing/jev.js",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 10237,
+ "starsAtCapture": 10320,
"summary": {
"en": "An optional automatic model router that classifies a message before model selection.",
"zh": "在选择模型前先对消息分类的可选自动路由器。",
@@ -581,7 +581,7 @@
"evidenceUrl": "https://github.com/0xPlaygrounds/rig/blob/2d16c1b25f6749b3a2cd841beddf767106495069/crates/rig-typesafeai/src/wire.rs#L40",
"language": "Rust",
"license": "MIT",
- "starsAtCapture": 8694,
+ "starsAtCapture": 8703,
"summary": {
"en": "An experimental Rust crate for expressing JEV questions and answers with Rust types.",
"zh": "用 Rust 类型表达 JEV 问题与答案的实验性 crate。",
@@ -605,7 +605,7 @@
"name": "Bifrost · TypeSafe Provider",
"owner": "maximhq",
"category": "sdk-integrations",
- "starsAtCapture": 8227,
+ "starsAtCapture": 8250,
"language": "Go",
"license": "Apache-2.0",
"summary": {
@@ -628,34 +628,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "gpt-load-jev-provider",
- "name": "GPT-Load · JEV Provider",
- "owner": "tbphp",
- "category": "sdk-integrations",
- "starsAtCapture": 6942,
- "language": "Go",
- "license": "MIT",
- "summary": {
- "en": "A self-hosted AI gateway with a first-class JEV channel and native Decisions routes.",
- "zh": "自托管 AI 网关内置一等 JEV 渠道与原生 Decisions 路由。",
- "ja": "セルフホスト AI ゲートウェイに、JEV 専用チャネルとネイティブ Decisions ルートを搭載。"
- },
- "decision": {
- "en": "Registers the fixed TypeSafe endpoint and routes decision creation, model listing, and probes through the native protocol.",
- "zh": "注册固定 TypeSafe 端点，并通过原生协议处理决策创建、模型列表与探测。",
- "ja": "固定 TypeSafe エンドポイントを登録し、判断作成・モデル一覧・プローブをネイティブプロトコルで処理します。"
- },
- "repoUrl": "https://github.com/tbphp/gpt-load",
- "evidenceUrl": "https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "firstmate-jev",
"name": "Firstmate · JEV Dispatch",
@@ -665,7 +637,7 @@
"evidenceUrl": "https://github.com/kunchenguid/firstmate/blob/4812db801628040b609dc25a2a8a91ed5efac662/bin/fm-dispatch-resolve.sh",
"language": "Shell",
"license": "MIT",
- "starsAtCapture": 6930,
+ "starsAtCapture": 7018,
"summary": {
"en": "An agent crew dispatcher that can match task briefs to rules with JEV.",
"zh": "可用 JEV 把任务简报匹配到调度规则的 Agent 团队调度器。",
@@ -684,12 +656,40 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "gpt-load-jev-provider",
+ "name": "GPT-Load · JEV Provider",
+ "owner": "tbphp",
+ "category": "sdk-integrations",
+ "starsAtCapture": 6961,
+ "language": "Go",
+ "license": "MIT",
+ "summary": {
+ "en": "A self-hosted AI gateway with a first-class JEV channel and native Decisions routes.",
+ "zh": "自托管 AI 网关内置一等 JEV 渠道与原生 Decisions 路由。",
+ "ja": "セルフホスト AI ゲートウェイに、JEV 専用チャネルとネイティブ Decisions ルートを搭載。"
+ },
+ "decision": {
+ "en": "Registers the fixed TypeSafe endpoint and routes decision creation, model listing, and probes through the native protocol.",
+ "zh": "注册固定 TypeSafe 端点，并通过原生协议处理决策创建、模型列表与探测。",
+ "ja": "固定 TypeSafe エンドポイントを登録し、判断作成・モデル一覧・プローブをネイティブプロトコルで処理します。"
+ },
+ "repoUrl": "https://github.com/tbphp/gpt-load",
+ "evidenceUrl": "https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/tbphp/gpt-load/blob/93502ced3a650018a35ed76c54573e90742b8ccc/internal/channel/modules/jev.go"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "greptimedb-jev-sql",
"name": "GreptimeDB · JEV SQL",
"owner": "GreptimeTeam",
"category": "search-data",
- "starsAtCapture": 6697,
+ "starsAtCapture": 6704,
"language": "Rust",
"license": "Apache-2.0",
"summary": {
@@ -721,7 +721,7 @@
"evidenceUrl": "https://github.com/tamaratran/fast-jev-compaction/blob/e3f262a7f4d42bd8dd32ced30d26176f7cb545b0/src/client.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 6044,
+ "starsAtCapture": 6342,
"summary": {
"en": "Compacts Claude Code tool history while keeping retained text verbatim.",
"zh": "压缩 Claude Code 工具历史，同时原样保留仍有价值的内容。",
@@ -747,7 +747,7 @@
"name": "Open Connector · TypeSafe",
"owner": "oomol-lab",
"category": "sdk-integrations",
- "starsAtCapture": 5858,
+ "starsAtCapture": 5876,
"language": "TypeScript",
"license": "Apache-2.0",
"summary": {
@@ -784,7 +784,7 @@
},
"language": "TypeScript",
"license": "Apache-2.0",
- "starsAtCapture": 5300,
+ "starsAtCapture": 5324,
"summary": {
"en": "Vercel’s open agent framework for packaging skills, evals, and tool-using workflows.",
"zh": "Vercel 开源的 Agent 框架，用于打包 Skill、评测与带工具的工作流。",
@@ -803,7 +803,7 @@
"name": "Kiln · JEV Adapter",
"owner": "Kiln-AI",
"category": "sdk-integrations",
- "starsAtCapture": 5078,
+ "starsAtCapture": 5077,
"language": "Python",
"license": "NOASSERTION",
"summary": {
@@ -835,7 +835,7 @@
"evidenceUrl": "https://github.com/agentgateway/agentgateway/blob/6b0270efd25b5255932943e48b5ca47583d3ad28/examples/llm-guardrail-jev/guardrail.ts",
"language": "TypeScript",
"license": "Apache-2.0",
- "starsAtCapture": 4970,
+ "starsAtCapture": 4990,
"summary": {
"en": "A JEV webhook guardrail example for inspecting model requests and responses.",
"zh": "用 JEV Webhook 检查模型请求与响应的 Guardrail 示例。",
@@ -863,7 +863,7 @@
"evidenceUrl": "https://github.com/latitude-dev/latitude-llm/blob/6b2484c1d74973877fc7d353b8867c7496568d78/packages/domain/flaggers/src/use-cases/run-jev-preclassifier.ts#L1",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 4665,
+ "starsAtCapture": 4668,
"summary": {
"en": "An optional JEV preclassifier for selecting which conversation checks should run.",
"zh": "选择一段会话应执行哪些检查的可选 JEV 预分类器。",
@@ -882,6 +882,64 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "kev",
+ "name": "Kev 0.5B",
+ "owner": "jaredpalmer",
+ "category": "open-models",
+ "repoUrl": "https://github.com/jaredpalmer/kev",
+ "evidenceUrl": "https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md",
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 4463,
+ "summary": {
+ "en": "A tiny trainable Jev-like decision model based on Qwen2.5-0.5B.",
+ "zh": "基于 Qwen2.5-0.5B、可在 MacBook 上训练运行的轻量决策模型。",
+ "ja": "Qwen2.5-0.5B に基づく、小さなトレーニング可能な JEV のような意思決定モデル。"
+ },
+ "decision": {
+ "en": "Exposes a TypeSafe-compatible API for bounded local choices and scores.",
+ "zh": "通过 TypeSafe 兼容接口完成有边界的本地选择与评分。",
+ "ja": "制限されたローカルの選択肢とスコア用の TypeSafe 互換 API を公開します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false,
+ "source": {
+ "type": "x",
+ "label": "Original X case",
+ "url": "https://x.com/jaredpalmer/status/2101028325472841920",
+ "author": "@jaredpalmer",
+ "views": 165852
+ }
+ },
+ {
+ "id": "jev-chat-jarvis",
+ "name": "JEV Chat Jarvis",
+ "owner": "jev-chat",
+ "category": "interfaces",
+ "starsAtCapture": 4149,
+ "language": "Kotlin",
+ "license": "MIT",
+ "summary": {
+ "en": "An Android conversation copilot that reads visible chat context and proposes replies without auto-sending.",
+ "zh": "读取可见聊天上下文、给出候选回复但不自动发送的 Android 对话副驾。",
+ "ja": "表示中の会話を読み、候補返信を提案するが自動送信しない Android 会話コパイロット。"
+ },
+ "decision": {
+ "en": "Uses JEV to judge conversation state and bounded reply choices while the user retains the send action.",
+ "zh": "用 JEV 判断会话状态与有限回复选项，最终发送仍由用户控制。",
+ "ja": "JEV で会話状態と限定された返信候補を判断し、送信操作はユーザーに残します。"
+ },
+ "repoUrl": "https://github.com/jev-chat/jev-chat-jarvis",
+ "evidenceUrl": "https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "semif",
"name": "SemIf",
@@ -892,7 +950,7 @@
"modelUrl": "https://github.com/TheoLeeCJ/SemIf#models",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 3357,
+ "starsAtCapture": 3882,
"summary": {
"en": "An independent open-model implementation of the semantic-if interface pattern.",
"zh": "用开放模型实现“语义 if”接口模式的独立研究项目。",
@@ -911,6 +969,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "fx-typesafe-permission-reviewer",
+ "name": "fx · TypeSafe Permission Reviewer",
+ "owner": "vercel-labs",
+ "category": "safety-review",
+ "starsAtCapture": 3127,
+ "language": "Zig",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "A Zig coding agent with an optional TypeSafe Jev permission reviewer for pending tool actions.",
+ "zh": "Zig 编码 Agent，可选 TypeSafe Jev 权限审查器，用于审核待执行工具动作。",
+ "ja": "保留中のツール操作を審査する TypeSafe Jev 権限レビュアー付き Zig コーディング Agent。"
+ },
+ "decision": {
+ "en": "Posts pending actions to System One as Choice questions and maps allow/deny probabilities into the agent permission gate.",
+ "zh": "把待执行动作作为 Choice 提交到 System One，并把允许/拒绝概率映射进 Agent 权限门。",
+ "ja": "保留中の操作を Choice として System One に送り、許可/拒否確率を Agent の権限ゲートへ写します。"
+ },
+ "repoUrl": "https://github.com/vercel-labs/fx",
+ "evidenceUrl": "https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "ax-typesafe",
"name": "Ax · TypeSafe Client",
@@ -920,7 +1006,7 @@
"evidenceUrl": "https://github.com/ax-llm/ax/blob/5c43344f9ef3016db576fa2c3b59d48ef21b4d71/src/ax/ai/typesafe/client.ts#L1",
"language": "TypeScript",
"license": "Apache-2.0",
- "starsAtCapture": 2942,
+ "starsAtCapture": 2945,
"summary": {
"en": "A TypeSafe integration for boolean, finite-class, and native JEV signatures.",
"zh": "支持布尔、有限分类以及原生 JEV 请求的 TypeSafe 集成。",
@@ -967,42 +1053,12 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "kev",
- "name": "Kev 0.5B",
- "owner": "jaredpalmer",
- "category": "open-models",
- "repoUrl": "https://github.com/jaredpalmer/kev",
- "evidenceUrl": "https://github.com/jaredpalmer/kev/blob/20fa6268c8ceb226530be2fb5266ab2c36b37724/README.md",
- "language": "Python",
- "license": "Apache-2.0",
- "starsAtCapture": 2638,
- "summary": {
- "en": "A tiny trainable Jev-like decision model based on Qwen2.5-0.5B.",
- "zh": "基于 Qwen2.5-0.5B、可在 MacBook 上训练运行的轻量决策模型。",
- "ja": "Qwen2.5-0.5B に基づく、小さなトレーニング可能な JEV のような意思決定モデル。"
- },
- "decision": {
- "en": "Exposes a TypeSafe-compatible API for bounded local choices and scores.",
- "zh": "通过 TypeSafe 兼容接口完成有边界的本地选择与评分。",
- "ja": "制限されたローカルの選択肢とスコア用の TypeSafe 互換 API を公開します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false,
- "source": {
- "type": "x",
- "label": "Original X case",
- "url": "https://x.com/jaredpalmer/status/2101028325472841920",
- "author": "@jaredpalmer",
- "views": 165852
- }
- },
{
"id": "agent-router-typesafe",
"name": "Agent Router · TypeSafe",
"owner": "theagentrouter",
"category": "routing-optimization",
- "starsAtCapture": 2127,
+ "starsAtCapture": 2131,
"language": "Go",
"license": "Apache-2.0",
"summary": {
@@ -1034,7 +1090,7 @@
"evidenceUrl": "https://github.com/jarrodwatts/jev-trader/blob/b587759e459ea049590102e54a0b07800864cdc3/src/model.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 1882,
+ "starsAtCapture": 2080,
"summary": {
"en": "A market-making experiment that can request one JEV trade decision per Monad block.",
"zh": "可在 Monad 每个区块请求一次 JEV 交易判断的做市实验。",
@@ -1063,7 +1119,7 @@
"modelUrl": "https://huggingface.co/C-Tianyu/NanoJev",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 1853,
+ "starsAtCapture": 2018,
"summary": {
"en": "A 0.6B open replica of the JEV interface with parallel decisions and a training pipeline.",
"zh": "提供并行判断、动态候选与完整训练流程的 0.6B 开放 JEV 接口复刻。",
@@ -1083,76 +1139,20 @@
"runtimeVerified": false
},
{
- "id": "llmgateway-system-one",
- "name": "LLMGateway · System One",
- "owner": "theopenco",
- "category": "routing-optimization",
- "starsAtCapture": 1651,
- "language": "TypeScript",
- "license": "NOASSERTION",
- "summary": {
- "en": "LLMGateway implements a native System One route with typed schemas and provider mappings.",
- "zh": "LLMGateway 实现带类型化 Schema 与 Provider 映射的原生 System One 路由。",
- "ja": "LLMGateway は型付き Schema と Provider マッピングを備えたネイティブ System One ルートを実装。"
- },
- "decision": {
- "en": "Validates Noul, Choice, and Score requests, resolves aliases to priced models, and routes them through decision-capable providers.",
- "zh": "校验 Noul、Choice、Score 请求，将别名解析到已定价模型，再路由到支持决策的 Provider。",
- "ja": "Noul・Choice・Score を検証し、別名を価格設定済みモデルへ解決して、判断対応 Provider にルーティングします。"
- },
- "repoUrl": "https://github.com/theopenco/llmgateway",
- "evidenceUrl": "https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "jev-chat-jarvis",
- "name": "JEV Chat Jarvis",
- "owner": "jev-chat",
- "category": "interfaces",
- "starsAtCapture": 1645,
- "language": "Kotlin",
- "license": "MIT",
- "summary": {
- "en": "An Android conversation copilot that reads visible chat context and proposes replies without auto-sending.",
- "zh": "读取可见聊天上下文、给出候选回复但不自动发送的 Android 对话副驾。",
- "ja": "表示中の会話を読み、候補返信を提案するが自動送信しない Android 会話コパイロット。"
- },
- "decision": {
- "en": "Uses JEV to judge conversation state and bounded reply choices while the user retains the send action.",
- "zh": "用 JEV 判断会话状态与有限回复选项，最终发送仍由用户控制。",
- "ja": "JEV で会話状態と限定された返信候補を判断し、送信操作はユーザーに残します。"
- },
- "repoUrl": "https://github.com/jev-chat/jev-chat-jarvis",
- "evidenceUrl": "https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/jev-chat/jev-chat-jarvis/blob/d8720521fa13aa5172a891cba95c5dd8aeb44a11/app/src/main/java/com/jev/probe/jev/JevClient.kt"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "typesafe-skills",
- "name": "TypeSafe Skills",
- "owner": "typesafe-ai",
- "category": "developer-tools",
- "repoUrl": "https://github.com/typesafe-ai/skills",
- "evidenceUrl": "https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md"
+ "id": "typesafe-skills",
+ "name": "TypeSafe Skills",
+ "owner": "typesafe-ai",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/typesafe-ai/skills",
+ "evidenceUrl": "https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/typesafe-ai/skills/blob/65a39f393687675ce170e6094757de20370365b9/skills/typesafe-ai/SKILL.md"
},
"language": "Not declared",
"license": "MIT",
- "starsAtCapture": 1610,
+ "starsAtCapture": 1867,
"summary": {
"en": "Official agent skills for building against TypeSafe’s System One / JEV API.",
"zh": "面向 TypeSafe System One / JEV API 的官方 Agent Skill 集合。",
@@ -1167,29 +1167,29 @@
"runtimeVerified": false
},
{
- "id": "rnskill-jev-office-gate",
- "name": "RNSkill · JEV Office Gate",
- "owner": "Pluviobyte",
- "category": "agent-workflows",
- "starsAtCapture": 1591,
- "language": "Python",
+ "id": "llmgateway-system-one",
+ "name": "LLMGateway · System One",
+ "owner": "theopenco",
+ "category": "routing-optimization",
+ "starsAtCapture": 1653,
+ "language": "TypeScript",
"license": "NOASSERTION",
"summary": {
- "en": "An agent-skill collection containing a JEV gate for bounded office-document checks.",
- "zh": "Agent Skill 合集中包含一个用于有限办公文档检查的 JEV Gate。",
- "ja": "Agent Skill 集に、限定されたオフィス文書チェック用 JEV ゲートを収録。"
+ "en": "LLMGateway implements a native System One route with typed schemas and provider mappings.",
+ "zh": "LLMGateway 实现带类型化 Schema 与 Provider 映射的原生 System One 路由。",
+ "ja": "LLMGateway は型付き Schema と Provider マッピングを備えたネイティブ System One ルートを実装。"
},
"decision": {
- "en": "Turns configured document-review conditions into typed judgments before the workflow proceeds.",
- "zh": "把配置好的文档审查条件转成类型化判断，再决定工作流是否继续。",
- "ja": "設定済みの文書レビュー条件を型付き判断へ変換し、ワークフロー継続可否を決めます。"
+ "en": "Validates Noul, Choice, and Score requests, resolves aliases to priced models, and routes them through decision-capable providers.",
+ "zh": "校验 Noul、Choice、Score 请求，将别名解析到已定价模型，再路由到支持决策的 Provider。",
+ "ja": "Noul・Choice・Score を検証し、別名を価格設定済みモデルへ解決して、判断対応 Provider にルーティングします。"
},
- "repoUrl": "https://github.com/Pluviobyte/rnskill",
- "evidenceUrl": "https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py",
+ "repoUrl": "https://github.com/theopenco/llmgateway",
+ "evidenceUrl": "https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py"
+ "url": "https://github.com/theopenco/llmgateway/blob/24de43b5f307ecbdee9aad386b68aee76de4ae8c/apps/gateway/src/systemone/systemone.ts"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1203,7 +1203,7 @@
"evidenceUrl": "https://github.com/bespokelabsai/nimble/blob/35fe1f4fdbc64fa3dbbf5ac0042f3f5680c1ad28/README.md",
"language": "Python",
"license": "Not declared",
- "starsAtCapture": 1514,
+ "starsAtCapture": 1638,
"summary": {
"en": "An open toolkit for local typed decisions, data curation, and evaluation.",
"zh": "面向本地类型化决策、数据整理与评测的开放工具包。",
@@ -1224,6 +1224,34 @@
"views": 63276
}
},
+ {
+ "id": "rnskill-jev-office-gate",
+ "name": "RNSkill · JEV Office Gate",
+ "owner": "Pluviobyte",
+ "category": "agent-workflows",
+ "starsAtCapture": 1595,
+ "language": "Python",
+ "license": "NOASSERTION",
+ "summary": {
+ "en": "An agent-skill collection containing a JEV gate for bounded office-document checks.",
+ "zh": "Agent Skill 合集中包含一个用于有限办公文档检查的 JEV Gate。",
+ "ja": "Agent Skill 集に、限定されたオフィス文書チェック用 JEV ゲートを収録。"
+ },
+ "decision": {
+ "en": "Turns configured document-review conditions into typed judgments before the workflow proceeds.",
+ "zh": "把配置好的文档审查条件转成类型化判断，再决定工作流是否继续。",
+ "ja": "設定済みの文書レビュー条件を型付き判断へ変換し、ワークフロー継続可否を決めます。"
+ },
+ "repoUrl": "https://github.com/Pluviobyte/rnskill",
+ "evidenceUrl": "https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/Pluviobyte/rnskill/blob/83d1783b892bbaa29a137895ff8338264b6872fa/skills/jev-office-gate/scripts/jev_office_gate.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "agent-desktop-jev",
"name": "Agent Desktop · JEV Skill",
@@ -1233,7 +1261,7 @@
"evidenceUrl": "https://github.com/lahfir/agent-desktop/blob/7a8e4a10281c7319733aa200fd79501f34529716/scripts/jev/act.mjs",
"language": "JavaScript",
"license": "Apache-2.0",
- "starsAtCapture": 1441,
+ "starsAtCapture": 1522,
"summary": {
"en": "An optional JEV skill that chooses native desktop controls and actions from accessibility data.",
"zh": "根据系统无障碍数据选择原生桌面控件与动作的可选 JEV Skill。",
@@ -1253,29 +1281,29 @@
"runtimeVerified": false
},
{
- "id": "vellum-assistant-jev",
- "name": "Vellum Assistant · JEV",
- "owner": "vellum-ai",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/vellum-ai/vellum-assistant",
- "evidenceUrl": "https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts",
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 1295,
- "summary": {
- "en": "An optional JEV provider that submits conversation state and explicit questions.",
- "zh": "提交会话状态与显式问题的可选 JEV Provider。",
- "ja": "会話状態と明示的な質問を送信するオプションの JEV プロバイダー。"
- },
- "decision": {
- "en": "Bundles state and questions for System One and returns structured answers to the assistant.",
- "zh": "把状态与问题打包发送给 System One，再向 Assistant 返回结构化答案。",
- "ja": "System One の状態と質問をバンドルし、構造化された回答をアシスタントに返します。"
- },
+ "id": "yibie-awesome-jev",
+ "name": "Awesome JEV (yibie)",
+ "owner": "yibie",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/yibie/awesome-jev",
+ "evidenceUrl": "https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts"
+ "url": "https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md"
+ },
+ "language": "Python",
+ "license": "Unknown",
+ "starsAtCapture": 1331,
+ "summary": {
+ "en": "A large community directory of public JEV projects, integrations, and discussions.",
+ "zh": "收录公开 JEV 项目、集成与讨论的大型社区目录。",
+ "ja": "公開 JEV プロジェクト、連携、議論をまとめた大規模コミュニティ目录。"
+ },
+ "decision": {
+ "en": "Organizes the ecosystem into browsable implementation and discussion groups.",
+ "zh": "把生态项目按实现与讨论类型整理为可浏览目录。",
+ "ja": "実装と議論を参照しやすい分類に整理します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1289,7 +1317,7 @@
"evidenceUrl": "https://github.com/elvisun/newsjack/tree/092d882fc69912622f620c50eb493afe625f99dc/demos/news-desk-dealer",
"language": "Go",
"license": "MIT",
- "starsAtCapture": 1292,
+ "starsAtCapture": 1316,
"summary": {
"en": "An open-source PR workflow that screens a live news feed for timely brand opportunities.",
"zh": "从实时新闻流中筛选品牌可跟进话题的开源 PR 工作流。",
@@ -1311,29 +1339,29 @@
}
},
{
- "id": "laravel-ai-typesafe",
- "name": "Laravel AI · TypeSafe",
- "owner": "laravel",
+ "id": "vellum-assistant-jev",
+ "name": "Vellum Assistant · JEV",
+ "owner": "vellum-ai",
"category": "sdk-integrations",
- "repoUrl": "https://github.com/laravel/ai",
- "evidenceUrl": "https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php",
- "language": "PHP",
+ "repoUrl": "https://github.com/vellum-ai/vellum-assistant",
+ "evidenceUrl": "https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts",
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 1183,
+ "starsAtCapture": 1303,
"summary": {
- "en": "Laravel AI includes a native TypeSafe classification provider backed by JEV.",
- "zh": "Laravel AI 内置由 JEV 驱动的 TypeSafe 分类 Provider。",
- "ja": "Laravel AI に、JEV を使う TypeSafe 分類プロバイダーが組み込まれています。"
+ "en": "An optional JEV provider that submits conversation state and explicit questions.",
+ "zh": "提交会话状态与显式问题的可选 JEV Provider。",
+ "ja": "会話状態と明示的な質問を送信するオプションの JEV プロバイダー。"
},
"decision": {
- "en": "Maps Laravel classification requests to the System One gateway and a configured JEV model.",
- "zh": "把 Laravel 分类请求映射到 System One Gateway 与配置的 JEV 模型。",
- "ja": "Laravel の分類リクエストを System One ゲートウェイと設定済み JEV モデルへ変換します。"
+ "en": "Bundles state and questions for System One and returns structured answers to the assistant.",
+ "zh": "把状态与问题打包发送给 System One，再向 Assistant 返回结构化答案。",
+ "ja": "System One の状態と質問をバンドルし、構造化された回答をアシスタントに返します。"
},
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php"
+ "url": "https://github.com/vellum-ai/vellum-assistant/blob/ee5ba342719e72b67698c2e1a4a78837321d0b1d/assistant/src/providers/jev/client.ts"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1347,7 +1375,7 @@
"evidenceUrl": "https://github.com/vinnylarouge/jevlike/blob/94f5fd1b0b11d52bbdfdf4e0ee6aa96b568f8452/README.md",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 1170,
+ "starsAtCapture": 1241,
"summary": {
"en": "An independent starter model that scores a changing list of text or visual options in one pass.",
"zh": "一次为动态文本或视觉候选列表评分的独立入门模型。",
@@ -1367,29 +1395,57 @@
"runtimeVerified": false
},
{
- "id": "yibie-awesome-jev",
- "name": "Awesome JEV (yibie)",
- "owner": "yibie",
- "category": "developer-tools",
- "repoUrl": "https://github.com/yibie/awesome-jev",
- "evidenceUrl": "https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md",
+ "id": "laravel-ai-typesafe",
+ "name": "Laravel AI · TypeSafe",
+ "owner": "laravel",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/laravel/ai",
+ "evidenceUrl": "https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php",
+ "language": "PHP",
+ "license": "MIT",
+ "starsAtCapture": 1184,
+ "summary": {
+ "en": "Laravel AI includes a native TypeSafe classification provider backed by JEV.",
+ "zh": "Laravel AI 内置由 JEV 驱动的 TypeSafe 分类 Provider。",
+ "ja": "Laravel AI に、JEV を使う TypeSafe 分類プロバイダーが組み込まれています。"
+ },
+ "decision": {
+ "en": "Maps Laravel classification requests to the System One gateway and a configured JEV model.",
+ "zh": "把 Laravel 分类请求映射到 System One Gateway 与配置的 JEV 模型。",
+ "ja": "Laravel の分類リクエストを System One ゲートウェイと設定済み JEV モデルへ変換します。"
+ },
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/yibie/awesome-jev/blob/a42aea88b30aeb62b3f165395f15336853e01863/README.md"
+ "url": "https://github.com/laravel/ai/blob/f0a5d4f3c5bddda7c8975eb79e92d62811197484/src/Providers/TypeSafeProvider.php"
},
- "language": "Python",
- "license": "Unknown",
- "starsAtCapture": 1046,
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "agent-beacon-jev-evaluator",
+ "name": "Agent Beacon · Jev Evaluator",
+ "owner": "asymptote-labs",
+ "category": "agent-workflows",
+ "starsAtCapture": 1103,
+ "language": "Go",
+ "license": "MIT",
"summary": {
- "en": "A large community directory of public JEV projects, integrations, and discussions.",
- "zh": "收录公开 JEV 项目、集成与讨论的大型社区目录。",
- "ja": "公開 JEV プロジェクト、連携、議論をまとめた大規模コミュニティ目录。"
+ "en": "Cross-harness agent memory that can ask Jev to score bounded, redacted session projections before promoting lessons.",
+ "zh": "跨 Harness 的 Agent 记忆系统，可用 Jev 对脱敏会话投影打分后再提升经验。",
+ "ja": "複数 Harness 向け Agent メモリ。Jev で編集済みセッション投影を採点してから教訓を昇格できます。"
},
"decision": {
- "en": "Organizes the ecosystem into browsable implementation and discussion groups.",
- "zh": "把生态项目按实现与讨论类型整理为可浏览目录。",
- "ja": "実装と議論を参照しやすい分類に整理します。"
+ "en": "Sends Noul rubric questions to TypeSafe System One and stores calibrated probabilities with human review gates.",
+ "zh": "向 TypeSafe System One 发送 Noul 评分问题，并把校准概率与人工复核门一起保存。",
+ "ja": "TypeSafe System One に Noul 採点質問を送り、較正確率を人手レビュー付きで保存します。"
+ },
+ "repoUrl": "https://github.com/asymptote-labs/agent-beacon",
+ "evidenceUrl": "https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1408,7 +1464,7 @@
},
"language": "",
"license": "Apache-2.0",
- "starsAtCapture": 958,
+ "starsAtCapture": 966,
"summary": {
"en": "A Celesto PR-review example prepares sandbox checks and compares a general model with Jev on candidate findings.",
"zh": "Celesto 的 PR 审查示例在沙盒中准备检查，再比较普通模型与 Jev 对候选问题的判断。",
@@ -1422,6 +1478,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "awlevin:typesafe-computer-use",
+ "name": "TypeSafe Computer Use",
+ "owner": "awlevin",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/awlevin/typesafe-computer-use",
+ "evidenceUrl": "https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py"
+ },
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 843,
+ "summary": {
+ "en": "Builds candidate actions from OCR and UI state for Jev to control macOS, calling a text model only when writing is needed.",
+ "zh": "用 OCR 和界面状态构造候选动作，让 Jev 决定如何操作 macOS，需要写文字时再调用文本模型。",
+ "ja": "JEV の OCR および UI 状態から候補アクションを構築して macOS を制御し、書き込みが必要な場合にのみテキスト モデルを呼び出します。"
+ },
+ "decision": {
+ "en": "Selects the next step from deterministically extracted controls and actions before desktop execution.",
+ "zh": "从确定性提取的控件与动作中选择下一步，执行器操作桌面。",
+ "ja": "デスクトップの実行前に、決定的に抽出されたコントロールとアクションから次のステップを選択します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "bastani-inc:atomic",
"name": "Atomic",
@@ -1436,7 +1520,7 @@
},
"language": "TypeScript",
"license": "Custom license",
- "starsAtCapture": 812,
+ "starsAtCapture": 814,
"summary": {
"en": "An optional Jev decision backend in the Atomic coding Agent for bounded structured choices such as routing.",
"zh": "Atomic 编程 Agent 的可选 Jev 决策后端，为路由等流程提供受限的结构化选择。",
@@ -1464,7 +1548,7 @@
},
"language": "TypeScript",
"license": "Not declared",
- "starsAtCapture": 810,
+ "starsAtCapture": 811,
"summary": {
"en": "A terminal generator CLI that ships with an agent skill for creating content from the command line.",
"zh": "终端生成 CLI，并附带可从命令行创作内容的 Agent Skill。",
@@ -1492,7 +1576,7 @@
},
"language": "Python",
"license": "MIT",
- "starsAtCapture": 772,
+ "starsAtCapture": 806,
"summary": {
"en": "An evidence-backed playbook of JEV use cases, patterns, prompts, and starter code.",
"zh": "有证据支撑的 JEV 用例、模式、提示词与起步代码手册。",
@@ -1506,34 +1590,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "awlevin:typesafe-computer-use",
- "name": "TypeSafe Computer Use",
- "owner": "awlevin",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/awlevin/typesafe-computer-use",
- "evidenceUrl": "https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/awlevin/typesafe-computer-use/blob/cc7b5066ae1a07b5e3182e8f87a9b5b6dfdcffc1/typesafe_computer_use/decide.py"
- },
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 765,
- "summary": {
- "en": "Builds candidate actions from OCR and UI state for Jev to control macOS, calling a text model only when writing is needed.",
- "zh": "用 OCR 和界面状态构造候选动作，让 Jev 决定如何操作 macOS，需要写文字时再调用文本模型。",
- "ja": "JEV の OCR および UI 状態から候補アクションを構築して macOS を制御し、書き込みが必要な場合にのみテキスト モデルを呼び出します。"
- },
- "decision": {
- "en": "Selects the next step from deterministically extracted controls and actions before desktop execution.",
- "zh": "从确定性提取的控件与动作中选择下一步，执行器操作桌面。",
- "ja": "デスクトップの実行前に、決定的に抽出されたコントロールとアクションから次のステップを選択します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "hippo-memory-jev",
"name": "Hippo Memory · JEV Reranker",
@@ -1543,7 +1599,7 @@
"evidenceUrl": "https://github.com/kitfunso/hippo-memory/blob/c9eb2c31cd307cdae78c05e1fcb3e6d062efd5c0/src/rerankers/jev.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 752,
+ "starsAtCapture": 756,
"summary": {
"en": "A local-first agent-memory library with an opt-in JEV reranker for recalled memories.",
"zh": "本地优先的 Agent Memory 类库，可选用 JEV 对召回记忆重新排序。",
@@ -1571,7 +1627,7 @@
"evidenceUrl": "https://github.com/githubnext/localjev/blob/3f23e36e1a3bff46c7e83e8e3781d3512bc82021/README.md",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 704,
+ "starsAtCapture": 731,
"summary": {
"en": "A local Jev-compatible server with benchmarks across small local models.",
"zh": "基于多种本地小模型、提供 JEV 兼容接口与评测的本地服务。",
@@ -1597,7 +1653,7 @@
"name": "Distill · JEV Runtime",
"owner": "samuelfaj",
"category": "agent-workflows",
- "starsAtCapture": 683,
+ "starsAtCapture": 684,
"language": "Rust",
"license": "Apache-2.0",
"summary": {
@@ -1662,7 +1718,7 @@
},
"language": "TypeScript",
"license": "FSL-1.1-ALv2",
- "starsAtCapture": 663,
+ "starsAtCapture": 666,
"summary": {
"en": "Optional second-stage search: widen the hybrid pool, then Score-rerank candidates with Workers AI typesafe/jev.",
"zh": "可选的二段检索：先扩大混合召回，再用 Cloudflare Workers AI 上的 `typesafe/jev` Score 重排。",
@@ -1676,6 +1732,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "v-modal-awesome-jev-tools",
+ "name": "Awesome JEV Tools",
+ "owner": "v-modal",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/v-modal/awesome-jev-tools",
+ "evidenceUrl": "https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md"
+ },
+ "language": "",
+ "license": "Unknown",
+ "starsAtCapture": 658,
+ "summary": {
+ "en": "A curated directory focused on tools built for JEV and TypeSafe System One.",
+ "zh": "聚焦 JEV 与 TypeSafe System One 工具的精选目录。",
+ "ja": "JEV と TypeSafe System One 向けツールのキュレーション目录。"
+ },
+ "decision": {
+ "en": "Surfaces practical tools by integration shape and use case.",
+ "zh": "按集成方式与使用场景筛选实用工具。",
+ "ja": "連携方式と用途別に実用ツールを紹介します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "tiptour-macos",
"name": "TipTour macOS",
@@ -1685,7 +1769,7 @@
"evidenceUrl": "https://github.com/milind-soni/tiptour-macos/blob/52582467c883d66484542f3be8e259340eb524f1/README.md",
"language": "Swift",
"license": "Not declared",
- "starsAtCapture": 642,
+ "starsAtCapture": 654,
"summary": {
"en": "A local macOS computer-use companion driven by Jev or Gemini Live.",
"zh": "由 JEV 或 Gemini Live 驱动的本地 macOS 电脑操作伴侣。",
@@ -1707,29 +1791,29 @@
}
},
{
- "id": "v-modal-awesome-jev-tools",
- "name": "Awesome JEV Tools",
- "owner": "v-modal",
- "category": "developer-tools",
- "repoUrl": "https://github.com/v-modal/awesome-jev-tools",
- "evidenceUrl": "https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md",
+ "id": "splash",
+ "name": "Splash",
+ "owner": "incoai",
+ "category": "open-models",
+ "repoUrl": "https://github.com/incoai/splash",
+ "evidenceUrl": "https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/v-modal/awesome-jev-tools/blob/f117e0c368d6e293bf5fdf99244a1173a6dc2ae0/README.md"
+ "url": "https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md"
},
- "language": "",
- "license": "Unknown",
- "starsAtCapture": 628,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 638,
"summary": {
- "en": "A curated directory focused on tools built for JEV and TypeSafe System One.",
- "zh": "聚焦 JEV 与 TypeSafe System One 工具的精选目录。",
- "ja": "JEV と TypeSafe System One 向けツールのキュレーション目录。"
+ "en": "A local Apple Silicon inference engine oriented around decision-style models.",
+ "zh": "面向决策类模型的本地 Apple Silicon 推理引擎。",
+ "ja": "意思決定系モデル向けの Apple Silicon ローカル推論エンジン。"
},
"decision": {
- "en": "Surfaces practical tools by integration shape and use case.",
- "zh": "按集成方式与使用场景筛选实用工具。",
- "ja": "連携方式と用途別に実用ツールを紹介します。"
+ "en": "Runs local inference so apps can obtain typed judgments without a remote JEV hop.",
+ "zh": "在本地完成推理，让应用无需远程 JEV 也能拿到类型化判断。",
+ "ja": "ローカル推論で型付き判断を返し、遠隔 JEV なしでも使えるようにします。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1748,7 +1832,7 @@
},
"language": "Swift",
"license": "MIT",
- "starsAtCapture": 622,
+ "starsAtCapture": 629,
"summary": {
"en": "An optional Jev command-risk advisor inside a native macOS Agent, with a TypeSafeKit client.",
"zh": "原生 macOS Agent 内的可选 Jev 命令风险顾问，配有 TypeSafeKit 客户端。",
@@ -1763,57 +1847,57 @@
"runtimeVerified": false
},
{
- "id": "agentjido:req_llm",
- "name": "Req LLM",
- "owner": "agentjido",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/agentjido/req_llm",
- "evidenceUrl": "https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md",
+ "id": "hermes-jev-skills",
+ "name": "Hermes JEV Skills",
+ "owner": "kerpopule",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/kerpopule/hermes-jev-skills",
+ "evidenceUrl": "https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md"
+ "url": "https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py"
},
- "language": "Elixir",
- "license": "Apache-2.0",
- "starsAtCapture": 581,
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 600,
"summary": {
- "en": "A TypeSafe provider for calling Jev through ReqLLM’s evaluate interface in Elixir.",
- "zh": "ReqLLM 的 TypeSafe provider，让 Elixir 应用通过 evaluate 接口调用 Jev。",
- "ja": "Elixir の ReqLLM 評価インターフェイスから JEV を呼び出す TypeSafe プロバイダー。"
+ "en": "JEV-powered routing, memory, compaction, skill selection, and computer/browser use for Hermes agents.",
+ "zh": "为 Hermes Agent 提供 JEV 驱动的路由、记忆、压缩、Skill 选择与电脑/浏览器操作。",
+ "ja": "Hermes エージェント向けに、JEV 駆動のルーティング・記憶・圧縮・Skill 選択・PC/ブラウザ操作。"
},
"decision": {
- "en": "Sends state and questions, normalizes answers and retains the raw provider response.",
- "zh": "提交状态与判断题，将答案放入统一响应对象，并保留原始 provider 数据。",
- "ja": "状態と質問を送り、回答を正規化しつつ、プロバイダーの元レスポンスも保持します。"
+ "en": "Chooses models, skills, or next computer actions from bounded JEV questions.",
+ "zh": "用有限 JEV 问题选择模型、Skill 或下一步电脑动作。",
+ "ja": "有限の JEV 質問でモデル・Skill・次の PC 操作を選びます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "splash",
- "name": "Splash",
- "owner": "incoai",
- "category": "open-models",
- "repoUrl": "https://github.com/incoai/splash",
- "evidenceUrl": "https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/incoai/splash/blob/f53d5ab543a7accdc332c060fd594a693f33f529/README.md"
- },
- "language": "Python",
- "license": "Apache-2.0",
- "starsAtCapture": 578,
+ "id": "heyjunpenn-awesome-jev",
+ "name": "Awesome JEV (HeyJunPenn)",
+ "owner": "heyjunpenn",
+ "category": "developer-tools",
+ "starsAtCapture": 589,
+ "language": "Astro",
+ "license": "MIT",
"summary": {
- "en": "A local Apple Silicon inference engine oriented around decision-style models.",
- "zh": "面向决策类模型的本地 Apple Silicon 推理引擎。",
- "ja": "意思決定系モデル向けの Apple Silicon ローカル推論エンジン。"
+ "en": "A multilingual, community-maintained catalogue of open-source JEV projects.",
+ "zh": "社区维护的多语言 JEV 开源项目目录。",
+ "ja": "コミュニティ運営の多言語 JEV オープンソース・カタログ。"
},
"decision": {
- "en": "Runs local inference so apps can obtain typed judgments without a remote JEV hop.",
- "zh": "在本地完成推理，让应用无需远程 JEV 也能拿到类型化判断。",
- "ja": "ローカル推論で型付き判断を返し、遠隔 JEV なしでも使えるようにします。"
+ "en": "Indexes implementations and integrations while retaining repository-level provenance for discovery.",
+ "zh": "索引实现与集成，并保留仓库级来源信息用于发现。",
+ "ja": "実装・連携を索引化し、発見用にリポジトリ単位の出典を保持します。"
+ },
+ "repoUrl": "https://github.com/heyjunpenn/awesome-jev",
+ "evidenceUrl": "https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1832,7 +1916,7 @@
},
"language": "JavaScript",
"license": "Not declared",
- "starsAtCapture": 549,
+ "starsAtCapture": 581,
"summary": {
"en": "Computer-use helpers that call JEV to decide the next UI action.",
"zh": "用 JEV 决定下一步界面动作的 Computer Use 辅助工具。",
@@ -1847,29 +1931,29 @@
"runtimeVerified": false
},
{
- "id": "bennykok:omg-dev",
- "name": "Omg.dev",
- "owner": "BennyKok",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/BennyKok/omg.dev",
- "evidenceUrl": "https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts",
+ "id": "agentjido:req_llm",
+ "name": "Req LLM",
+ "owner": "agentjido",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/agentjido/req_llm",
+ "evidenceUrl": "https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts"
+ "url": "https://github.com/agentjido/req_llm/blob/9cb0ee7a0fea5f3520fc953911d352c93193615e/README.md"
},
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 535,
+ "language": "Elixir",
+ "license": "Apache-2.0",
+ "starsAtCapture": 581,
"summary": {
- "en": "An omg.dev mobile testing script can use Jev to read the accessibility tree and choose the next interaction.",
- "zh": "omg.dev 的移动端测试脚本可让 Jev 读取可访问性树并选择下一步交互。",
- "ja": "omg.dev モバイル テスト スクリプトは、JEV を使用してアクセシビリティ ツリーを読み取り、次のインタラクションを選択できます。"
+ "en": "A TypeSafe provider for calling Jev through ReqLLM’s evaluate interface in Elixir.",
+ "zh": "ReqLLM 的 TypeSafe provider，让 Elixir 应用通过 evaluate 接口调用 Jev。",
+ "ja": "Elixir の ReqLLM 評価インターフェイスから JEV を呼び出す TypeSafe プロバイダー。"
},
"decision": {
- "en": "Chooses controls and checks completion or blockage before the test runner operates the UI.",
- "zh": "判断要操作的控件、步骤是否完成或是否受阻，再由测试执行器操作界面。",
- "ja": "テスト ランナーが UI を操作する前に、コントロールを選択し、完了または障害をチェックします。"
+ "en": "Sends state and questions, normalizes answers and retains the raw provider response.",
+ "zh": "提交状态与判断题，将答案放入统一响应对象，并保留原始 provider 数据。",
+ "ja": "状態と質問を送り、回答を正規化しつつ、プロバイダーの元レスポンスも保持します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -1888,7 +1972,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 502,
+ "starsAtCapture": 556,
"summary": {
"en": "Reviews a Git diff or a codebase in stages and displays review leads in a local dashboard.",
"zh": "分阶段检查 Git diff 或整个代码库，在本地面板里展示可复核的审查线索。",
@@ -1902,6 +1986,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "bennykok:omg-dev",
+ "name": "Omg.dev",
+ "owner": "BennyKok",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/BennyKok/omg.dev",
+ "evidenceUrl": "https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/BennyKok/omg.dev/blob/a00f56684a569ee417be787c148eaa9874946f2b/mobile/scripts/jev.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 535,
+ "summary": {
+ "en": "An omg.dev mobile testing script can use Jev to read the accessibility tree and choose the next interaction.",
+ "zh": "omg.dev 的移动端测试脚本可让 Jev 读取可访问性树并选择下一步交互。",
+ "ja": "omg.dev モバイル テスト スクリプトは、JEV を使用してアクセシビリティ ツリーを読み取り、次のインタラクションを選択できます。"
+ },
+ "decision": {
+ "en": "Chooses controls and checks completion or blockage before the test runner operates the UI.",
+ "zh": "判断要操作的控件、步骤是否完成或是否受阻，再由测试执行器操作界面。",
+ "ja": "テスト ランナーが UI を操作する前に、コントロールを選択し、完了または障害をチェックします。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "minecraft-agent-jev",
"name": "Astra + JEV Minecraft Agent",
@@ -1916,7 +2028,7 @@
},
"language": "JavaScript",
"license": "Unknown",
- "starsAtCapture": 485,
+ "starsAtCapture": 524,
"summary": {
"en": "A Minecraft agent where a planner sets objectives and JEV selects legal game actions.",
"zh": "由规划模型设定目标、JEV 选择合法游戏动作的 Minecraft Agent。",
@@ -1944,7 +2056,7 @@
},
"language": "Python",
"license": "MIT",
- "starsAtCapture": 472,
+ "starsAtCapture": 517,
"summary": {
"en": "An independent supervisor loop that reads worker diffs, logs and tests, asks Jev Nouls about stuck / off-track / verify, then applies a Python policy.",
"zh": "独立监督环读取编码工人的 diff、日志和测试，用 Jev Noul 判断卡住、跑偏、该验证，再由 Python 策略干预。",
@@ -1958,6 +2070,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "von-open-system-one",
+ "name": "Von",
+ "owner": "wfzyx",
+ "category": "open-models",
+ "repoUrl": "https://github.com/wfzyx/von",
+ "evidenceUrl": "https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md"
+ },
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 498,
+ "summary": {
+ "en": "An open, local System One decision model with a JEV-compatible interface.",
+ "zh": "提供 JEV 兼容接口的本地开放 System One 决策模型。",
+ "ja": "JEV 互換インターフェースを持つローカルのオープン System One 判断モデル。"
+ },
+ "decision": {
+ "en": "Scores bounded typed choices locally without autoregressive text generation.",
+ "zh": "在本地为有限类型化选项评分，不做自回归文本生成。",
+ "ja": "自己回帰テキスト生成を行わず、限定された型付き候補をローカルで評価します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "featherless-ai:simple-jev",
"name": "Simple JEV",
@@ -1972,7 +2112,7 @@
},
"language": "Python",
"license": "Unknown",
- "starsAtCapture": 458,
+ "starsAtCapture": 485,
"summary": {
"en": "Adapter turning open LLM endpoints into Jev-compatible classification services without training a separate classifier head.",
"zh": "将任意开源大语言模型转化为分类器与 Jev 兼容端点的适配服务，无需额外训练专用分类头。",
@@ -2000,7 +2140,7 @@
},
"language": "CSS",
"license": "MIT",
- "starsAtCapture": 431,
+ "starsAtCapture": 476,
"summary": {
"en": "A curated list of official and community resources for TypeSafe, System One, and JEV.",
"zh": "TypeSafe、System One 与 JEV 的官方与社区资源精选列表。",
@@ -2019,7 +2159,7 @@
"name": "Awesome TypeSafe JEV",
"owner": "AbdelStark",
"category": "developer-tools",
- "starsAtCapture": 431,
+ "starsAtCapture": 476,
"language": "HTML",
"license": "MIT",
"summary": {
@@ -2043,141 +2183,85 @@
"runtimeVerified": false
},
{
- "id": "notque:vexjoy-agent",
- "name": "Vexjoy Agent",
- "owner": "notque",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/notque/vexjoy-agent",
- "evidenceUrl": "https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md"
- },
+ "id": "wuyoscar-jev-skill",
+ "name": "Awesome Jev Skills",
+ "owner": "wuyoscar",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/wuyoscar/jev-skill",
+ "evidenceUrl": "https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py",
"language": "Python",
"license": "MIT",
- "starsAtCapture": 423,
+ "starsAtCapture": 437,
"summary": {
- "en": "An optional Jev routing path that matches VexJoy requests to specialist Agents, skills and workflows.",
- "zh": "给 VexJoy 的任务分派增加一条 Jev 路线。输入需求后，判断该选哪位专长 Agent、哪项技能和哪条工作流。",
- "ja": "VexJoy リクエストを専門エージェント、スキル、ワークフローに照合するオプションの JEV ルーティング パス。"
+ "en": "A collection of agent skills and workflows with a standard-library JEV decision client.",
+ "zh": "一套 Agent Skills 与工作流，内含仅依赖标准库的 JEV 决策客户端。",
+ "ja": "標準ライブラリだけで動く JEV クライアントを含む Agent スキルとワークフロー集です。"
},
"decision": {
- "en": "After deterministic routing guards, Jev judges the remaining candidates and required workflow components.",
- "zh": "先判断是否需要路由，再从实际清单中选择候选；非法选择或调用失败交回原有流程。",
- "ja": "確定的ルーティング ガードの後、JEV は残りの候補と必要なワークフロー コンポーネントを判断します。"
+ "en": "Runs typed decisions through TypeSafe or OpenRouter and applies confidence-aware review rules.",
+ "zh": "通过 TypeSafe 或 OpenRouter 执行类型化决策，并应用基于置信度的复核规则。",
+ "ja": "TypeSafe または OpenRouter で型付き判断を実行し、信頼度に応じた確認ルールを適用します。"
},
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "smithers",
- "name": "Smithers",
- "owner": "smithersai",
- "category": "agent-workflows",
- "repoUrl": "https://github.com/smithersai/smithers",
- "evidenceUrl": "https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts"
- },
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 418,
- "summary": {
- "en": "An agentic TypeScript workflow framework with a first-class JEV classify path.",
- "zh": "以 TypeScript 配置定义工作流的 Agent 框架，内置 JEV 分类路径。",
- "ja": "TypeScript 設定でワークフローを定義し、JEV 分類パスを持つエージェント枠組み。"
- },
- "decision": {
- "en": "Classifies workflow steps or review verdicts through a dedicated JEV relay.",
- "zh": "通过专用 JEV 中继对工作流步骤或审查结论做分类。",
- "ja": "専用の JEV 中継でワークフロー手順やレビュー判定を分類します。"
+ "url": "https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "classifier-dev",
- "name": "Classifier.dev",
- "owner": "mrmps",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/mrmps/classifier-dev",
- "evidenceUrl": "https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts",
+ "id": "notque:vexjoy-agent",
+ "name": "Vexjoy Agent",
+ "owner": "notque",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/notque/vexjoy-agent",
+ "evidenceUrl": "https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts"
+ "url": "https://github.com/notque/vexjoy-agent/blob/ab51ee7da567e84f72f76d73b6e7b83cec5ef060/README.md"
},
- "language": "TypeScript",
+ "language": "Python",
"license": "MIT",
- "starsAtCapture": 406,
+ "starsAtCapture": 423,
"summary": {
- "en": "A zero-shot classification service with JEV as its primary typed-decision backend.",
- "zh": "以 JEV 为主要类型化决策后端的零样本分类服务。",
- "ja": "JEV を主要な型付き判断バックエンドに使うゼロショット分類サービス。"
+ "en": "An optional Jev routing path that matches VexJoy requests to specialist Agents, skills and workflows.",
+ "zh": "给 VexJoy 的任务分派增加一条 Jev 路线。输入需求后，判断该选哪位专长 Agent、哪项技能和哪条工作流。",
+ "ja": "VexJoy リクエストを専門エージェント、スキル、ワークフローに照合するオプションの JEV ルーティング パス。"
},
"decision": {
- "en": "Classifies labels and dimensions, with explicit fallback and observability around JEV calls.",
- "zh": "用 JEV 判断标签与维度，并提供明确的回退与可观测性。",
- "ja": "JEV でラベルと次元を分類し、明示的なフォールバックと観測性を備えます。"
+ "en": "After deterministic routing guards, Jev judges the remaining candidates and required workflow components.",
+ "zh": "先判断是否需要路由，再从实际清单中选择候选；非法选择或调用失败交回原有流程。",
+ "ja": "確定的ルーティング ガードの後、JEV は残りの候補と必要なワークフロー コンポーネントを判断します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "hermes-jev-skills",
- "name": "Hermes JEV Skills",
- "owner": "kerpopule",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/kerpopule/hermes-jev-skills",
- "evidenceUrl": "https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py",
+ "id": "smithers",
+ "name": "Smithers",
+ "owner": "smithersai",
+ "category": "agent-workflows",
+ "repoUrl": "https://github.com/smithersai/smithers",
+ "evidenceUrl": "https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/kerpopule/hermes-jev-skills/blob/a5d89014fbeca5102c5e7ccd3915afcd856505d8/jevkit/choose.py"
- },
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 401,
- "summary": {
- "en": "JEV-powered routing, memory, compaction, skill selection, and computer/browser use for Hermes agents.",
- "zh": "为 Hermes Agent 提供 JEV 驱动的路由、记忆、压缩、Skill 选择与电脑/浏览器操作。",
- "ja": "Hermes エージェント向けに、JEV 駆動のルーティング・記憶・圧縮・Skill 選択・PC/ブラウザ操作。"
- },
- "decision": {
- "en": "Chooses models, skills, or next computer actions from bounded JEV questions.",
- "zh": "用有限 JEV 问题选择模型、Skill 或下一步电脑动作。",
- "ja": "有限の JEV 質問でモデル・Skill・次の PC 操作を選びます。"
+ "url": "https://github.com/smithersai/smithers/blob/394ada6a3fb815b4b62cfc7bc2242e238d42ff44/apps/server/src/jev.ts"
},
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "wuyoscar-jev-skill",
- "name": "Awesome Jev Skills",
- "owner": "wuyoscar",
- "category": "developer-tools",
- "repoUrl": "https://github.com/wuyoscar/jev-skill",
- "evidenceUrl": "https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py",
- "language": "Python",
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 398,
+ "starsAtCapture": 420,
"summary": {
- "en": "A collection of agent skills and workflows with a standard-library JEV decision client.",
- "zh": "一套 Agent Skills 与工作流，内含仅依赖标准库的 JEV 决策客户端。",
- "ja": "標準ライブラリだけで動く JEV クライアントを含む Agent スキルとワークフロー集です。"
+ "en": "An agentic TypeScript workflow framework with a first-class JEV classify path.",
+ "zh": "以 TypeScript 配置定义工作流的 Agent 框架，内置 JEV 分类路径。",
+ "ja": "TypeScript 設定でワークフローを定義し、JEV 分類パスを持つエージェント枠組み。"
},
"decision": {
- "en": "Runs typed decisions through TypeSafe or OpenRouter and applies confidence-aware review rules.",
- "zh": "通过 TypeSafe 或 OpenRouter 执行类型化决策，并应用基于置信度的复核规则。",
- "ja": "TypeSafe または OpenRouter で型付き判断を実行し、信頼度に応じた確認ルールを適用します。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/wuyoscar/jev-skill/blob/4d6efbc5b87a4172524ad4ab4590aef077fdc13b/skills/jev/scripts/jev.py"
+ "en": "Classifies workflow steps or review verdicts through a dedicated JEV relay.",
+ "zh": "通过专用 JEV 中继对工作流步骤或审查结论做分类。",
+ "ja": "専用の JEV 中継でワークフロー手順やレビュー判定を分類します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2196,7 +2280,7 @@
},
"language": "",
"license": "MIT",
- "starsAtCapture": 385,
+ "starsAtCapture": 416,
"summary": {
"en": "Searches the web with Jev choosing sources and time ranges, then ranking returned links.",
"zh": "用自然语言搜网页：Jev 选择搜索来源和时间范围，再给返回的链接排序。",
@@ -2211,62 +2295,32 @@
"runtimeVerified": false
},
{
- "id": "von-open-system-one",
- "name": "Von",
- "owner": "wfzyx",
- "category": "open-models",
- "repoUrl": "https://github.com/wfzyx/von",
- "evidenceUrl": "https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md",
+ "id": "classifier-dev",
+ "name": "Classifier.dev",
+ "owner": "mrmps",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/mrmps/classifier-dev",
+ "evidenceUrl": "https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/wfzyx/von/blob/14d09878e89b103bfbbe641f9bed02e4d72c8830/README.md"
- },
- "language": "Python",
- "license": "Apache-2.0",
- "starsAtCapture": 374,
- "summary": {
- "en": "An open, local System One decision model with a JEV-compatible interface.",
- "zh": "提供 JEV 兼容接口的本地开放 System One 决策模型。",
- "ja": "JEV 互換インターフェースを持つローカルのオープン System One 判断モデル。"
- },
- "decision": {
- "en": "Scores bounded typed choices locally without autoregressive text generation.",
- "zh": "在本地为有限类型化选项评分，不做自回归文本生成。",
- "ja": "自己回帰テキスト生成を行わず、限定された型付き候補をローカルで評価します。"
+ "url": "https://github.com/mrmps/classifier-dev/blob/33ca63816f2bc7e93c3f2d0715f7896500370739/src/jev.ts"
},
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "jev-experiments",
- "name": "Jev Experiments",
- "owner": "dabit3",
- "category": "search-data",
- "repoUrl": "https://github.com/dabit3/jev-experiments",
- "evidenceUrl": "https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist",
"language": "TypeScript",
- "license": "Not declared",
- "starsAtCapture": 358,
+ "license": "MIT",
+ "starsAtCapture": 411,
"summary": {
- "en": "Small Jev experiments including intent search and agent-assist flows.",
- "zh": "包含意图搜索与 Agent Assist 流程的一组 JEV 实验。",
- "ja": "インテント検索やエージェント支援フローを含む小規模な JEV 実験。"
+ "en": "A zero-shot classification service with JEV as its primary typed-decision backend.",
+ "zh": "以 JEV 为主要类型化决策后端的零样本分类服务。",
+ "ja": "JEV を主要な型付き判断バックエンドに使うゼロショット分類サービス。"
},
"decision": {
- "en": "Uses fast intent judgments after retrieval instead of asking a large model to inspect everything.",
- "zh": "在初步召回后快速判断意图，避免让大模型通读全部内容。",
- "ja": "大規模なモデルにすべてを検査させる代わりに、取得後に迅速な意図判断を使用します。"
+ "en": "Classifies labels and dimensions, with explicit fallback and observability around JEV calls.",
+ "zh": "用 JEV 判断标签与维度，并提供明确的回退与可观测性。",
+ "ja": "JEV でラベルと次元を分類し、明示的なフォールバックと観測性を備えます。"
},
"verification": "source-reviewed",
- "runtimeVerified": false,
- "source": {
- "type": "x",
- "label": "Original X case",
- "url": "https://x.com/dabit3/status/2100960281769738433",
- "author": "@dabit3",
- "views": 131659
- }
+ "runtimeVerified": false
},
{
"id": "tax-doc-classifier",
@@ -2277,7 +2331,7 @@
"evidenceUrl": "https://github.com/kyotofin/tax-doc-classifier/blob/6afcf701395466d7c936ec8178daf017b9d96b0c/src/backend.ts",
"language": "TypeScript",
"license": "Apache-2.0",
- "starsAtCapture": 348,
+ "starsAtCapture": 399,
"summary": {
"en": "Classifies extracted tax-document pages against a fixed IRS form catalogue.",
"zh": "把税务文档页面归入预先定义的 IRS 表格目录。",
@@ -2298,34 +2352,6 @@
"views": 223890
}
},
- {
- "id": "fhshaik:typesafe-mario",
- "name": "TypeSafe Mario",
- "owner": "fhshaik",
- "category": "domain-tools",
- "repoUrl": "https://github.com/fhshaik/typesafe-mario",
- "evidenceUrl": "https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md"
- },
- "language": "Python",
- "license": "Unknown",
- "starsAtCapture": 338,
- "summary": {
- "en": "An experimental NES Mario controller that gives Jev structured emulator RAM and telemetry instead of screenshots.",
- "zh": "从 NES 模拟器 RAM 和状态数据中提取环境，让 Jev 选择超级马力欧的手柄按键。",
- "ja": "JEV 構造化エミュレータ RAM とスクリーンショットの代わりにテレメトリを提供する実験的な NES マリオ コントローラ。"
- },
- "decision": {
- "en": "Reads motion, enemies, terrain and recent controls, then selects a predefined legal action.",
- "zh": "读取运动、敌人、地形及近期操作状态，从预设合法操作中选一项。",
- "ja": "動き、敵、地形、最近の操作を読み取り、事前に定義された法的措置を選択します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "logicrw-awesome-jev-projects",
"name": "Awesome JEV Projects",
@@ -2340,7 +2366,7 @@
},
"language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 334,
+ "starsAtCapture": 398,
"summary": {
"en": "A source-backed JEV ecosystem radar with plain-language discovery and GitHub sync.",
"zh": "有源码依据的 JEV 生态雷达，含白话发现页与 GitHub 同步。",
@@ -2368,7 +2394,7 @@
},
"language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 334,
+ "starsAtCapture": 387,
"summary": {
"en": "A Skill for Codex browser workflows: Jev chooses navigation, clicks, and scrolling while Codex handles text input and final checks.",
"zh": "给 Codex 浏览器工作流加一个 Skill：Jev 选导航、点击和滚动，Codex 保留文字输入与最终核验。",
@@ -2383,32 +2409,34 @@
"runtimeVerified": false
},
{
- "id": "wrongstack:wrongstack",
- "name": "WrongStack",
- "owner": "WrongStack",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/WrongStack/WrongStack",
- "evidenceUrl": "https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md"
- },
+ "id": "jev-experiments",
+ "name": "Jev Experiments",
+ "owner": "dabit3",
+ "category": "search-data",
+ "repoUrl": "https://github.com/dabit3/jev-experiments",
+ "evidenceUrl": "https://github.com/dabit3/jev-experiments/tree/c469e5bfdc73eb3e1999bba2569e66b579a970fd/agent-assist",
"language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 331,
+ "license": "Not declared",
+ "starsAtCapture": 370,
"summary": {
- "en": "An optional Jev dispatch classifier for choosing among WrongStack specialist Agents.",
- "zh": "给 WrongStack 编程 Agent 增加一个可选分派助手。遇到多个相近的专长 Agent 时，用 Jev 判断谁更适合当前任务。",
- "ja": "WrongStack の専門 Agent から担当を選ぶための、任意で使える JEV ディスパッチ分類器。"
+ "en": "Small Jev experiments including intent search and agent-assist flows.",
+ "zh": "包含意图搜索与 Agent Assist 流程的一组 JEV 实验。",
+ "ja": "インテント検索やエージェント支援フローを含む小規模な JEV 実験。"
},
"decision": {
- "en": "Jev evaluates the task against eligible specialists; local dispatch rules use the result.",
- "zh": "从候选角色中选一个，同时判断有没有任何候选真正适合；不合适时交回原有策略。",
- "ja": "JEV は、適格なスペシャリストと照らし合わせてタスクを評価します。ローカルディスパッチルールは結果を使用します。"
+ "en": "Uses fast intent judgments after retrieval instead of asking a large model to inspect everything.",
+ "zh": "在初步召回后快速判断意图，避免让大模型通读全部内容。",
+ "ja": "大規模なモデルにすべてを検査させる代わりに、取得後に迅速な意図判断を使用します。"
},
"verification": "source-reviewed",
- "runtimeVerified": false
+ "runtimeVerified": false,
+ "source": {
+ "type": "x",
+ "label": "Original X case",
+ "url": "https://x.com/dabit3/status/2100960281769738433",
+ "author": "@dabit3",
+ "views": 131659
+ }
},
{
"id": "droidrun:mobile-jev",
@@ -2424,7 +2452,7 @@
},
"language": "",
"license": "MIT",
- "starsAtCapture": 329,
+ "starsAtCapture": 356,
"summary": {
"en": "Controls an Android phone through Mobilerun, with a web studio and CLI showing Jev decisions.",
"zh": "通过 Mobilerun 控制 Android 手机，网页面板与命令行可查看 Jev 的操作过程。",
@@ -2439,29 +2467,29 @@
"runtimeVerified": false
},
{
- "id": "cognesy:instructor-php",
- "name": "Instructor Php",
- "owner": "cognesy",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/cognesy/instructor-php",
- "evidenceUrl": "https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md",
+ "id": "fhshaik:typesafe-mario",
+ "name": "TypeSafe Mario",
+ "owner": "fhshaik",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/fhshaik/typesafe-mario",
+ "evidenceUrl": "https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md"
+ "url": "https://github.com/fhshaik/typesafe-mario/blob/ca22449ed187118d19326d1f54b01b6636578aa4/README.md"
},
- "language": "PHP",
- "license": "MIT",
- "starsAtCapture": 327,
+ "language": "Python",
+ "license": "Unknown",
+ "starsAtCapture": 355,
"summary": {
- "en": "A TypeSafe Decision driver within Instructor PHP’s Polyglot module.",
- "zh": "Instructor PHP 的 Polyglot 模块内置 TypeSafe Decision 驱动。",
- "ja": "Instructor PHP の Polyglot モジュール内の TypeSafe Decision ドライバー。"
+ "en": "An experimental NES Mario controller that gives Jev structured emulator RAM and telemetry instead of screenshots.",
+ "zh": "从 NES 模拟器 RAM 和状态数据中提取环境，让 Jev 选择超级马力欧的手柄按键。",
+ "ja": "JEV 構造化エミュレータ RAM とスクリーンショットの代わりにテレメトリを提供する実験的な NES マリオ コントローラ。"
},
"decision": {
- "en": "Converts application state and typed questions into Jev requests and maps responses to PHP decision objects.",
- "zh": "将业务状态和类型化问题转换为 Jev 请求，再映射为 PHP 决策响应。",
- "ja": "アプリケーション状態と型付き質問を JEV リクエストへ変換し、レスポンスを PHP の意思決定オブジェクトへ割り当てます。"
+ "en": "Reads motion, enemies, terrain and recent controls, then selects a predefined legal action.",
+ "zh": "读取运动、敌人、地形及近期操作状态，从预设合法操作中选一项。",
+ "ja": "動き、敵、地形、最近の操作を読み取り、事前に定義された法的措置を選択します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2480,7 +2508,7 @@
},
"language": "Not declared",
"license": "CC0-1.0",
- "starsAtCapture": 316,
+ "starsAtCapture": 354,
"summary": {
"en": "A source-backed list of projects built with JEV / TypeSafe System One.",
"zh": "基于源码核对的 JEV / TypeSafe System One 项目精选列表。",
@@ -2508,7 +2536,7 @@
},
"language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 316,
+ "starsAtCapture": 349,
"summary": {
"en": "A Claude Code and CLI proxy that asks Jev to score task complexity and pick a model from the account’s available set.",
"zh": "Claude Code / CLI 代理：Jev 给任务复杂度打分，并在当前账号可用的模型里选一个，再由本地策略决定是否更换。",
@@ -2523,113 +2551,113 @@
"runtimeVerified": false
},
{
- "id": "razorback16:openjev",
- "name": "Openjev",
- "owner": "razorback16",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/razorback16/openjev",
- "evidenceUrl": "https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py"
- },
+ "id": "rizzo-flow",
+ "name": "Rizzo Flow",
+ "owner": "Rizzo-AI-Academy",
+ "category": "open-models",
+ "repoUrl": "https://github.com/Rizzo-AI-Academy/rizzo-flow",
+ "evidenceUrl": "https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py",
"language": "Python",
"license": "Apache-2.0",
- "starsAtCapture": 286,
+ "starsAtCapture": 348,
"summary": {
- "en": "An independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model.",
- "zh": "一个兼容 Jev System One 接口的独立决策服务，使用开源 DiffusionGemma 模型运行。",
- "ja": "JEV API と互換性のある独立型 System One 意思決定サーバーで、オープンな DiffusionGemma モデルを実行します。"
+ "en": "An independent local decision model exposing a TypeSafe-compatible System One interface.",
+ "zh": "独立的本地决策模型，提供兼容 TypeSafe 的 System One 接口。",
+ "ja": "TypeSafe 互換の System One インターフェースを公開する独立ローカル判断モデルです。"
},
"decision": {
- "en": "Accepts Jev-style state and Noul, Choice, and Score questions, producing probabilities with a local model.",
- "zh": "接受与 Jev 相似的状态和 Noul、Choice、Score 问题，由本地模型产生概率结果。",
- "ja": "JEV スタイルの状態と Noul、Choice、Score の質問を受け入れ、ローカル モデルで確率を生成します。"
+ "en": "Translates Noul, Choice, and Score wire questions into local model requests without claiming JEV model parity.",
+ "zh": "把 Noul、Choice 与 Score 协议问题转换为本地模型请求，不宣称与 JEV 模型等价。",
+ "ja": "Noul、Choice、Score の質問をローカルモデル向けに変換し、JEV と同等のモデルだとは主張しません。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "realzachi:pg-jev",
- "name": "Pg JEV",
- "owner": "realZachi",
- "category": "search-data",
- "repoUrl": "https://github.com/realZachi/pg-jev",
- "evidenceUrl": "https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql",
+ "id": "wrongstack:wrongstack",
+ "name": "WrongStack",
+ "owner": "WrongStack",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/WrongStack/WrongStack",
+ "evidenceUrl": "https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql"
+ "url": "https://github.com/WrongStack/WrongStack/blob/4cf97c0aa4f855751949ee8e99c719d06f4c26e0/README.md"
},
- "language": "Shell",
- "license": "PostgreSQL",
- "starsAtCapture": 286,
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 332,
"summary": {
- "en": "Adds natural-language filtering, classification, and ranking of rows to PostgreSQL queries.",
- "zh": "在 PostgreSQL 查询中用自然语言给数据行筛选、分类和排序。",
- "ja": "PostgreSQL クエリに行の自然言語フィルタリング、分類、ランク付けを追加します。"
+ "en": "An optional Jev dispatch classifier for choosing among WrongStack specialist Agents.",
+ "zh": "给 WrongStack 编程 Agent 增加一个可选分派助手。遇到多个相近的专长 Agent 时，用 Jev 判断谁更适合当前任务。",
+ "ja": "WrongStack の専門 Agent から担当を選ぶための、任意で使える JEV ディスパッチ分類器。"
},
"decision": {
- "en": "Sends row content to Jev for a match decision, category, or score used by SQL conditions and ordering.",
- "zh": "将行内容交给 Jev，返回是否匹配、类别或等级分数，供 SQL 条件与排序使用。",
- "ja": "SQL 条件および順序付けで使用される一致判定、カテゴリ、またはスコアのために、行の内容を JEV に送信します。"
+ "en": "Jev evaluates the task against eligible specialists; local dispatch rules use the result.",
+ "zh": "从候选角色中选一个，同时判断有没有任何候选真正适合；不合适时交回原有策略。",
+ "ja": "JEV は、適格なスペシャリストと照らし合わせてタスクを評価します。ローカルディスパッチルールは結果を使用します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "third-hand",
- "name": "Third Hand",
- "owner": "shhivv",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/shhivv/third-hand",
- "evidenceUrl": "https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift",
+ "id": "cognesy:instructor-php",
+ "name": "Instructor Php",
+ "owner": "cognesy",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/cognesy/instructor-php",
+ "evidenceUrl": "https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift"
+ "url": "https://github.com/cognesy/instructor-php/blob/bb1160ce2360dcf1434788c46ee97809fe782ea3/README.md"
},
- "language": "Swift",
+ "language": "PHP",
"license": "MIT",
- "starsAtCapture": 283,
+ "starsAtCapture": 327,
"summary": {
- "en": "A Swift client stack that calls JEV while driving local computer-use style helpers.",
- "zh": "Swift 客户端栈：在驱动本地 computer-use 辅助逻辑时调用 JEV。",
- "ja": "ローカル Computer Use 補助を動かしつつ JEV を呼ぶ Swift クライアント。"
+ "en": "A TypeSafe Decision driver within Instructor PHP’s Polyglot module.",
+ "zh": "Instructor PHP 的 Polyglot 模块内置 TypeSafe Decision 驱动。",
+ "ja": "Instructor PHP の Polyglot モジュール内の TypeSafe Decision ドライバー。"
},
"decision": {
- "en": "Requests typed judgments from JEV before applying local automation actions.",
- "zh": "在执行本地自动化动作前向 JEV 请求类型化判断。",
- "ja": "ローカル自動化の前に JEV へ型付き判断を求めます。"
+ "en": "Converts application state and typed questions into Jev requests and maps responses to PHP decision objects.",
+ "zh": "将业务状态和类型化问题转换为 Jev 请求，再映射为 PHP 决策响应。",
+ "ja": "アプリケーション状態と型付き質問を JEV リクエストへ変換し、レスポンスを PHP の意思決定オブジェクトへ割り当てます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "yonatangross:orchestkit",
- "name": "Orchestkit",
- "owner": "yonatangross",
- "category": "developer-tools",
- "repoUrl": "https://github.com/yonatangross/orchestkit",
- "evidenceUrl": "https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts",
+ "id": "razorback16:openjev",
+ "name": "Openjev",
+ "owner": "razorback16",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/razorback16/openjev",
+ "evidenceUrl": "https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts"
+ "url": "https://github.com/razorback16/openjev/blob/91d5005effcf8cc0ecccaa9538ceabbb130fef59/openjev/api.py"
},
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 281,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 319,
"summary": {
- "en": "OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold.",
- "zh": "OrchestKit 可选用 Jev 给编程会话分类，符合阈值时用结果决定会话颜色。",
- "ja": "OrchestKit はオプションで JEV を使用してコーディング セッションを分類し、信頼度がしきい値を満たしたときにその色を設定できます。"
+ "en": "An independent System One decision server compatible with Jev’s API, running an open DiffusionGemma model.",
+ "zh": "一个兼容 Jev System One 接口的独立决策服务，使用开源 DiffusionGemma 模型运行。",
+ "ja": "JEV API と互換性のある独立型 System One 意思決定サーバーで、オープンな DiffusionGemma モデルを実行します。"
},
"decision": {
- "en": "Classifies the first task prompt and branch state by work type; local policy accepts the result or falls back.",
- "zh": "将首条任务提示与分支状态分类为开发、排错等工作类型；本地规则选择采用或回退。",
- "ja": "最初のタスクプロンプトとブランチ状態を作業種別ごとに分類し、ローカルポリシーが結果の採用またはフォールバックを決めます。"
+ "en": "Accepts Jev-style state and Noul, Choice, and Score questions, producing probabilities with a local model.",
+ "zh": "接受与 Jev 相似的状态和 Noul、Choice、Score 问题，由本地模型产生概率结果。",
+ "ja": "JEV スタイルの状態と Noul、Choice、Score の質問を受け入れ、ローカル モデルで確率を生成します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2648,7 +2676,7 @@
},
"language": "Python",
"license": "Apache-2.0",
- "starsAtCapture": 279,
+ "starsAtCapture": 316,
"summary": {
"en": "One-pass typed decisions with calibrated probabilities, fine-tuned from Qwen3.5-2B.",
"zh": "一次前向的类型化决策与校准概率，基于 Qwen3.5-2B 微调。",
@@ -2663,87 +2691,29 @@
"runtimeVerified": false
},
{
- "id": "rizzo-flow",
- "name": "Rizzo Flow",
- "owner": "Rizzo-AI-Academy",
- "category": "open-models",
- "repoUrl": "https://github.com/Rizzo-AI-Academy/rizzo-flow",
- "evidenceUrl": "https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py",
- "language": "Python",
- "license": "Apache-2.0",
- "starsAtCapture": 273,
- "summary": {
- "en": "An independent local decision model exposing a TypeSafe-compatible System One interface.",
- "zh": "独立的本地决策模型，提供兼容 TypeSafe 的 System One 接口。",
- "ja": "TypeSafe 互換の System One インターフェースを公開する独立ローカル判断モデルです。"
- },
- "decision": {
- "en": "Translates Noul, Choice, and Score wire questions into local model requests without claiming JEV model parity.",
- "zh": "把 Noul、Choice 与 Score 协议问题转换为本地模型请求，不宣称与 JEV 模型等价。",
- "ja": "Noul、Choice、Score の質問をローカルモデル向けに変換し、JEV と同等のモデルだとは主張しません。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/Rizzo-AI-Academy/rizzo-flow/blob/d34665b7a28c62b79f37939f2fd83f5fe659fbf9/src/rizzo_flow/compat.py"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "jev-align",
- "name": "jev-align",
- "owner": "sutro-sh",
- "category": "safety-review",
- "repoUrl": "https://github.com/sutro-sh/jev-align",
- "evidenceUrl": "https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md",
- "language": "Python",
- "license": "Apache-2.0",
- "starsAtCapture": 271,
- "summary": {
- "en": "A CLI for calibrating Jev classifiers from human examples with GEPA.",
- "zh": "用人工正反例和 GEPA 校准 JEV 分类标准的 CLI。",
- "ja": "GEPA を使用して人間の例から JEV 分類器を調整するための CLI。"
- },
- "decision": {
- "en": "Optimizes the decision prompt against examples instead of relying on a generic criterion.",
- "zh": "围绕明确样例优化判断标准，而不是依赖泛化提示词。",
- "ja": "一般的な基準に依存するのではなく、例に基づいて決定プロンプトを最適化します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false,
- "source": {
- "type": "x",
- "label": "Original X case",
- "url": "https://x.com/sethkimmel3/status/2101357768640987302",
- "author": "@sethkimmel3",
- "views": 68757
- }
- },
- {
- "id": "openjev-sglang",
- "name": "OpenJEV SGLang",
- "owner": "ekzhang",
- "category": "open-models",
- "repoUrl": "https://github.com/ekzhang/openjev-sglang",
- "evidenceUrl": "https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py",
+ "id": "realzachi:pg-jev",
+ "name": "Pg JEV",
+ "owner": "realZachi",
+ "category": "search-data",
+ "repoUrl": "https://github.com/realZachi/pg-jev",
+ "evidenceUrl": "https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py"
+ "url": "https://github.com/realZachi/pg-jev/blob/afd11fa856d7a2b831a1bfd8ee7f869ce8efcd62/sql/jev--0.2.0.sql"
},
- "language": "Python",
- "license": "Not declared",
- "starsAtCapture": 258,
+ "language": "Shell",
+ "license": "PostgreSQL",
+ "starsAtCapture": 314,
"summary": {
- "en": "A JEV-compatible API endpoint built on open models with a prefill-only serving path.",
- "zh": "基于开源模型的 JEV 兼容 API，采用 prefill-only 服务路径。",
- "ja": "オープンモデル上の JEV 互換 API。prefill-only の提供経路を使います。"
+ "en": "Adds natural-language filtering, classification, and ranking of rows to PostgreSQL queries.",
+ "zh": "在 PostgreSQL 查询中用自然语言给数据行筛选、分类和排序。",
+ "ja": "PostgreSQL クエリに行の自然言語フィルタリング、分類、ランク付けを追加します。"
},
"decision": {
- "en": "Serves System One–style probability answers from open models instead of hosted JEV.",
- "zh": "用开源模型提供 System One 风格的概率答案，替代托管 JEV。",
- "ja": "ホスト型 JEV の代わりに、オープンモデルで System One 風の確率回答を返します。"
+ "en": "Sends row content to Jev for a match decision, category, or score used by SQL conditions and ordering.",
+ "zh": "将行内容交给 Jev，返回是否匹配、类别或等级分数，供 SQL 条件与排序使用。",
+ "ja": "SQL 条件および順序付けで使用される一致判定、カテゴリ、またはスコアのために、行の内容を JEV に送信します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2762,7 +2732,7 @@
},
"language": "",
"license": "CC0-1.0",
- "starsAtCapture": 257,
+ "starsAtCapture": 303,
"summary": {
"en": "A curated list of JEV applications, libraries, and System One resources.",
"zh": "JEV 应用、类库与 System One 资源精选清单。",
@@ -2777,29 +2747,199 @@
"runtimeVerified": false
},
{
- "id": "cequence-io:openai-scala-client",
- "name": "Openai Scala Client",
- "owner": "cequence-io",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/cequence-io/openai-scala-client",
- "evidenceUrl": "https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md"
- },
- "language": "Scala",
+ "id": "laya-node-runtime",
+ "name": "Laya Node Runtime",
+ "owner": "receptron",
+ "category": "open-models",
+ "starsAtCapture": 295,
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 248,
"summary": {
- "en": "A dedicated TypeSafe module in a Scala client that supports multiple AI providers.",
- "zh": "让 Scala 应用也能接入 Jev。这个多模型客户端新增了独立 TypeSafe 模块，用状态和判断题获取结构化答案。",
- "ja": "複数の AI プロバイダーをサポートする Scala クライアントの専用 TypeSafe モジュール。"
+ "en": "A Node.js and TypeScript runtime for the open Laya JEV-compatible decision model.",
+ "zh": "面向开放 Laya JEV 兼容决策模型的 Node.js / TypeScript Runtime。",
+ "ja": "オープンな Laya JEV 互換判断モデル向け Node.js / TypeScript Runtime。"
},
"decision": {
- "en": "Sends shared state and typed questions to Jev and parses structured answers.",
- "zh": "调用 System One 返回单选、评分和是非概率；可把受支持的封闭 JSON Schema 转成判断题。",
- "ja": "共有状態と型付きの質問を JEV に送信し、構造化された回答を解析します。"
+ "en": "Runs the ONNX model locally and returns typed decision probabilities without a hosted JEV call.",
+ "zh": "在本地运行 ONNX 模型并返回类型化决策概率，无需调用托管 JEV。",
+ "ja": "ONNX モデルをローカル実行し、ホスト型 JEV を呼ばずに型付き判断確率を返します。"
+ },
+ "repoUrl": "https://github.com/receptron/laya",
+ "evidenceUrl": "https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "jev-mcp",
+ "name": "JEV MCP",
+ "owner": "jkudish",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/jkudish/jev-mcp",
+ "evidenceUrl": "https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 290,
+ "summary": {
+ "en": "Eight MCP tools for evidence checks, content screening, search, reranking, classification, comparison and extraction.",
+ "zh": "给 Agent 提供核对引用、筛查内容、查找、重排、分类、比较和提取等八个 MCP 判断工具。",
+ "ja": "証拠チェック、コンテンツ スクリーニング、検索、再ランキング、分類、比較、抽出のための 8 つの MCP ツール。"
+ },
+ "decision": {
+ "en": "Uses typed questions to assess evidence support, content risk and candidate relevance.",
+ "zh": "用 Choice、Noul 等问题评估证据支持、内容风险和候选相关性。",
+ "ja": "型付きの質問 を使用して、証拠の裏付け、コンテンツのリスク、および候補者の関連性を評価します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "third-hand",
+ "name": "Third Hand",
+ "owner": "shhivv",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/shhivv/third-hand",
+ "evidenceUrl": "https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/shhivv/third-hand/blob/430394b35dbb44ff8b303bf19da29b0828d92bd2/Sources/ThirdHand/JevClient.swift"
+ },
+ "language": "Swift",
+ "license": "MIT",
+ "starsAtCapture": 288,
+ "summary": {
+ "en": "A Swift client stack that calls JEV while driving local computer-use style helpers.",
+ "zh": "Swift 客户端栈：在驱动本地 computer-use 辅助逻辑时调用 JEV。",
+ "ja": "ローカル Computer Use 補助を動かしつつ JEV を呼ぶ Swift クライアント。"
+ },
+ "decision": {
+ "en": "Requests typed judgments from JEV before applying local automation actions.",
+ "zh": "在执行本地自动化动作前向 JEV 请求类型化判断。",
+ "ja": "ローカル自動化の前に JEV へ型付き判断を求めます。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "docjev",
+ "name": "DocJEV",
+ "owner": "jerryjliu",
+ "category": "search-data",
+ "starsAtCapture": 287,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "A document classifier and splitter with a dedicated JEV engine.",
+ "zh": "带专用 JEV Engine 的文档分类与切分工具。",
+ "ja": "専用 JEV エンジンを備えた文書分類・分割ツール。"
+ },
+ "decision": {
+ "en": "Classifies documents and split boundaries from bounded schemas, with cache and alternate-engine support.",
+ "zh": "根据有限 Schema 判断文档类别与切分边界，并支持缓存及备用引擎。",
+ "ja": "限定 Schema から文書カテゴリと分割境界を判断し、キャッシュと代替エンジンにも対応します。"
+ },
+ "repoUrl": "https://github.com/jerryjliu/docjev",
+ "evidenceUrl": "https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "yonatangross:orchestkit",
+ "name": "Orchestkit",
+ "owner": "yonatangross",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/yonatangross/orchestkit",
+ "evidenceUrl": "https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/yonatangross/orchestkit/blob/569095f35ce61a4881197dfb8117f4c3312e2bee/src/hooks/src/lib/session-category-provider.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 283,
+ "summary": {
+ "en": "OrchestKit can optionally use Jev to classify coding sessions and set their colors when confidence meets a threshold.",
+ "zh": "OrchestKit 可选用 Jev 给编程会话分类，符合阈值时用结果决定会话颜色。",
+ "ja": "OrchestKit はオプションで JEV を使用してコーディング セッションを分類し、信頼度がしきい値を満たしたときにその色を設定できます。"
+ },
+ "decision": {
+ "en": "Classifies the first task prompt and branch state by work type; local policy accepts the result or falls back.",
+ "zh": "将首条任务提示与分支状态分类为开发、排错等工作类型；本地规则选择采用或回退。",
+ "ja": "最初のタスクプロンプトとブランチ状態を作業種別ごとに分類し、ローカルポリシーが結果の採用またはフォールバックを決めます。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "jev-align",
+ "name": "jev-align",
+ "owner": "sutro-sh",
+ "category": "safety-review",
+ "repoUrl": "https://github.com/sutro-sh/jev-align",
+ "evidenceUrl": "https://github.com/sutro-sh/jev-align/blob/49753df924d30c0d3642b58e0b9b1e89921dc102/README.md",
+ "language": "Python",
+ "license": "Apache-2.0",
+ "starsAtCapture": 276,
+ "summary": {
+ "en": "A CLI for calibrating Jev classifiers from human examples with GEPA.",
+ "zh": "用人工正反例和 GEPA 校准 JEV 分类标准的 CLI。",
+ "ja": "GEPA を使用して人間の例から JEV 分類器を調整するための CLI。"
+ },
+ "decision": {
+ "en": "Optimizes the decision prompt against examples instead of relying on a generic criterion.",
+ "zh": "围绕明确样例优化判断标准，而不是依赖泛化提示词。",
+ "ja": "一般的な基準に依存するのではなく、例に基づいて決定プロンプトを最適化します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false,
+ "source": {
+ "type": "x",
+ "label": "Original X case",
+ "url": "https://x.com/sethkimmel3/status/2101357768640987302",
+ "author": "@sethkimmel3",
+ "views": 68757
+ }
+ },
+ {
+ "id": "openjev-sglang",
+ "name": "OpenJEV SGLang",
+ "owner": "ekzhang",
+ "category": "open-models",
+ "repoUrl": "https://github.com/ekzhang/openjev-sglang",
+ "evidenceUrl": "https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/ekzhang/openjev-sglang/blob/604664a22b2cf44c6cc499e503092ae4e3c24c03/src/openjev/__init__.py"
+ },
+ "language": "Python",
+ "license": "Not declared",
+ "starsAtCapture": 272,
+ "summary": {
+ "en": "A JEV-compatible API endpoint built on open models with a prefill-only serving path.",
+ "zh": "基于开源模型的 JEV 兼容 API，采用 prefill-only 服务路径。",
+ "ja": "オープンモデル上の JEV 互換 API。prefill-only の提供経路を使います。"
+ },
+ "decision": {
+ "en": "Serves System One–style probability answers from open models instead of hosted JEV.",
+ "zh": "用开源模型提供 System One 风格的概率答案，替代托管 JEV。",
+ "ja": "ホスト型 JEV の代わりに、オープンモデルで System One 風の確率回答を返します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2818,7 +2958,7 @@
},
"language": "Python",
"license": "Not declared",
- "starsAtCapture": 246,
+ "starsAtCapture": 268,
"summary": {
"en": "A calibrated ~151M non-autoregressive local decision engine aiming at JEV-style typed answers.",
"zh": "约 151M 的校准非自回归本地决策引擎，目标对齐类 JEV 类型化答案。",
@@ -2833,29 +2973,57 @@
"runtimeVerified": false
},
{
- "id": "jev-mcp",
- "name": "JEV MCP",
- "owner": "jkudish",
+ "id": "system-one-adapter-python",
+ "name": "System One Adapter (Python)",
+ "owner": "typesafe-ai",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/typesafe-ai/system-one-adapter-python",
+ "evidenceUrl": "https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py"
+ },
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 266,
+ "summary": {
+ "en": "A drop-in TypeSafeClient replacement that backs System One calls with ordinary LLM APIs.",
+ "zh": "可替换 TypeSafeClient 的适配器，用普通 LLM API 承接 System One 调用。",
+ "ja": "通常の LLM API で System One 呼び出しを受ける、TypeSafeClient 差し替えアダプタ。"
+ },
+ "decision": {
+ "en": "Maps Noul, Choice, and Score questions onto LLM providers and normalizes probabilities.",
+ "zh": "把 Noul / Choice / Score 问题映射到 LLM 提供方并归一化概率。",
+ "ja": "Noul / Choice / Score を LLM プロバイダへ写し、確率を正規化します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "typesafe-mcp",
+ "name": "TypeSafe MCP",
+ "owner": "itsmostafa",
"category": "developer-tools",
- "repoUrl": "https://github.com/jkudish/jev-mcp",
- "evidenceUrl": "https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127",
+ "repoUrl": "https://github.com/itsmostafa/typesafe-mcp",
+ "evidenceUrl": "https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/jkudish/jev-mcp/blob/89f88b90c3180184d56961160e7178e64c98e468/src/provider.ts#L127"
+ "url": "https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md"
},
- "language": "TypeScript",
+ "language": "Go",
"license": "MIT",
- "starsAtCapture": 245,
+ "starsAtCapture": 262,
"summary": {
- "en": "Eight MCP tools for evidence checks, content screening, search, reranking, classification, comparison and extraction.",
- "zh": "给 Agent 提供核对引用、筛查内容、查找、重排、分类、比较和提取等八个 MCP 判断工具。",
- "ja": "証拠チェック、コンテンツ スクリーニング、検索、再ランキング、分類、比較、抽出のための 8 つの MCP ツール。"
+ "en": "An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions.",
+ "zh": "让 Claude Code、Claude Desktop、Codex 和 Pi 通过 MCP 或扩展向 Jev 提问，获取结构化判断。",
+ "ja": "Claude Code、Claude Desktop、Codex、Pi から JEV へ型付き質問を送れる MCP サーバー。"
},
"decision": {
- "en": "Uses typed questions to assess evidence support, content risk and candidate relevance.",
- "zh": "用 Choice、Noul 等问题评估证据支持、内容风险和候选相关性。",
- "ja": "型付きの質問 を使用して、証拠の裏付け、コンテンツのリスク、および候補者の関連性を評価します。"
+ "en": "Submits state and Choice, Score or Noul questions and returns structured answers with probabilities.",
+ "zh": "把状态和 Choice、Score、Noul 问题交给 Jev，返回答案与概率。",
+ "ja": "状態、選択、スコア、または Noul の質問を送信し、確率を含む構造化された回答を返します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2874,7 +3042,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 244,
+ "starsAtCapture": 253,
"summary": {
"en": "Pi’s programmable runtime includes an optional Jev loop for observing state, making decisions and running bounded actions.",
"zh": "给 Pi 的工具运行时加上可编程的 Jev 决策循环。先写好观察、判断和执行步骤，再让它按预算在前台或后台运行。",
@@ -2889,85 +3057,113 @@
"runtimeVerified": false
},
{
- "id": "system-one-adapter-python",
- "name": "System One Adapter (Python)",
- "owner": "typesafe-ai",
+ "id": "cequence-io:openai-scala-client",
+ "name": "Openai Scala Client",
+ "owner": "cequence-io",
"category": "sdk-integrations",
- "repoUrl": "https://github.com/typesafe-ai/system-one-adapter-python",
- "evidenceUrl": "https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py",
+ "repoUrl": "https://github.com/cequence-io/openai-scala-client",
+ "evidenceUrl": "https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/typesafe-ai/system-one-adapter-python/blob/adffc2eab300a4fa3c0e92252d4ffd6ceaa53700/src/system_one_adapter/_client.py"
+ "url": "https://github.com/cequence-io/openai-scala-client/blob/cfabe8842344da713615d0121a7804840649eb46/README.md"
},
- "language": "Python",
+ "language": "Scala",
"license": "MIT",
- "starsAtCapture": 243,
+ "starsAtCapture": 249,
"summary": {
- "en": "A drop-in TypeSafeClient replacement that backs System One calls with ordinary LLM APIs.",
- "zh": "可替换 TypeSafeClient 的适配器，用普通 LLM API 承接 System One 调用。",
- "ja": "通常の LLM API で System One 呼び出しを受ける、TypeSafeClient 差し替えアダプタ。"
+ "en": "A dedicated TypeSafe module in a Scala client that supports multiple AI providers.",
+ "zh": "让 Scala 应用也能接入 Jev。这个多模型客户端新增了独立 TypeSafe 模块，用状态和判断题获取结构化答案。",
+ "ja": "複数の AI プロバイダーをサポートする Scala クライアントの専用 TypeSafe モジュール。"
},
"decision": {
- "en": "Maps Noul, Choice, and Score questions onto LLM providers and normalizes probabilities.",
- "zh": "把 Noul / Choice / Score 问题映射到 LLM 提供方并归一化概率。",
- "ja": "Noul / Choice / Score を LLM プロバイダへ写し、確率を正規化します。"
+ "en": "Sends shared state and typed questions to Jev and parses structured answers.",
+ "zh": "调用 System One 返回单选、评分和是非概率；可把受支持的封闭 JSON Schema 转成判断题。",
+ "ja": "共有状態と型付きの質問を JEV に送信し、構造化された回答を解析します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "jkudish:jev-browser",
- "name": "JEV Browser",
- "owner": "jkudish",
+ "id": "moritzkremb:jev-voice-browser",
+ "name": "JEV Voice Browser",
+ "owner": "moritzkremb",
"category": "browser-computer-use",
- "repoUrl": "https://github.com/jkudish/jev-browser",
- "evidenceUrl": "https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1",
+ "repoUrl": "https://github.com/moritzkremb/jev-voice-browser",
+ "evidenceUrl": "https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1"
+ "url": "https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123"
},
"language": "",
"license": "MIT",
- "starsAtCapture": 231,
+ "starsAtCapture": 247,
"summary": {
- "en": "Drives a browser from a task and URL, returning the final page, screenshot and action trace.",
- "zh": "给定任务与网址后驱动浏览器，返回最终页面、截图和逐步操作记录。",
- "ja": "タスクと URL からブラウザを起動し、最終ページ、スクリーンショット、アクション トレースを返します。"
+ "en": "Controls a Playwright browser by sending incremental speech transcripts to Jev.",
+ "zh": "用语音控制 Playwright 浏览器，将逐步转写的口令交给 Jev 判断。",
+ "ja": "増分音声トランスクリプトを JEV に送信することで、Playwright ブラウザを制御します。"
},
"decision": {
- "en": "Chooses DOM actions and judges goal completion or stalled progress; a separate model can supply typed text.",
- "zh": "从 DOM 控件选择点击、输入、选择或滚动，并判断目标是否完成、流程是否卡住。",
- "ja": "DOM アクションを選択し、目標の完了または進行の停滞を判断します。別のモデルでは、入力されたテキストを提供できます。"
+ "en": "Chooses intent, elements, URLs or verbatim spans and judges command completeness and sensitive actions.",
+ "zh": "选择意图、元素、网址或原文片段，并判断口令是否完整、是否涉及敏感动作。",
+ "ja": "インテント、要素、URL、または逐語的な範囲を選択し、コマンドの完全性と機密性の高いアクションを判断します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "typesafe-mcp",
- "name": "TypeSafe MCP",
- "owner": "itsmostafa",
- "category": "developer-tools",
- "repoUrl": "https://github.com/itsmostafa/typesafe-mcp",
- "evidenceUrl": "https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md",
+ "id": "jev-visual",
+ "name": "JEV Visual",
+ "owner": "hr98w",
+ "category": "open-models",
+ "repoUrl": "https://github.com/hr98w/jev-visual",
+ "evidenceUrl": "https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py"
+ },
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 243,
+ "summary": {
+ "en": "An educational JEV-like visual inference experiment on Apple Silicon with shared context scoring.",
+ "zh": "在 Apple Silicon 上的教学向类 JEV 视觉推理实验，共享上下文直接给候选打分。",
+ "ja": "Apple Silicon 上の教育向け JEV 風ビジョン推論。共有コンテキストで候補を直接採点。"
+ },
+ "decision": {
+ "en": "Scores visual or textual candidates in one forward pass and returns calibrated probabilities.",
+ "zh": "一次前向为视觉或文本候选打分，并返回校准概率。",
+ "ja": "1 回の前向き計算で視覚／テキスト候補を採点し、校正確率を返します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "jkudish:jev-browser",
+ "name": "JEV Browser",
+ "owner": "jkudish",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/jkudish/jev-browser",
+ "evidenceUrl": "https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/itsmostafa/typesafe-mcp/blob/0c9f35d9b1859189fc7e7d01947061f311ca6dde/README.md"
+ "url": "https://github.com/jkudish/jev-browser/blob/8d90c51bedbe7cd07596bfaa532ded019a31d2a8/src/navigate.ts#L1"
},
- "language": "Go",
+ "language": "",
"license": "MIT",
- "starsAtCapture": 227,
+ "starsAtCapture": 240,
"summary": {
- "en": "An MCP server that lets Claude Code, Claude Desktop, Codex and Pi ask Jev typed questions.",
- "zh": "让 Claude Code、Claude Desktop、Codex 和 Pi 通过 MCP 或扩展向 Jev 提问，获取结构化判断。",
- "ja": "Claude Code、Claude Desktop、Codex、Pi から JEV へ型付き質問を送れる MCP サーバー。"
+ "en": "Drives a browser from a task and URL, returning the final page, screenshot and action trace.",
+ "zh": "给定任务与网址后驱动浏览器，返回最终页面、截图和逐步操作记录。",
+ "ja": "タスクと URL からブラウザを起動し、最終ページ、スクリーンショット、アクション トレースを返します。"
},
"decision": {
- "en": "Submits state and Choice, Score or Noul questions and returns structured answers with probabilities.",
- "zh": "把状态和 Choice、Score、Noul 问题交给 Jev，返回答案与概率。",
- "ja": "状態、選択、スコア、または Noul の質問を送信し、確率を含む構造化された回答を返します。"
+ "en": "Chooses DOM actions and judges goal completion or stalled progress; a separate model can supply typed text.",
+ "zh": "从 DOM 控件选择点击、输入、选择或滚动，并判断目标是否完成、流程是否卡住。",
+ "ja": "DOM アクションを選択し、目標の完了または進行の停滞を判断します。別のモデルでは、入力されたテキストを提供できます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -2986,7 +3182,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 223,
+ "starsAtCapture": 229,
"summary": {
"en": "A self-hosted, versioned Agent skills library with optional Jev recommendations.",
"zh": "自建一个有版本管理的 Agent 技能库，还能选配 Jev 推荐：告诉它当前任务，从你有权限使用的技能里挑更相关的。",
@@ -3001,57 +3197,57 @@
"runtimeVerified": false
},
{
- "id": "jev-visual",
- "name": "JEV Visual",
- "owner": "hr98w",
- "category": "open-models",
- "repoUrl": "https://github.com/hr98w/jev-visual",
- "evidenceUrl": "https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/hr98w/jev-visual/blob/19af545f096e8db4c4dd5d47aed42d92ec252111/jev_visual/scoring.py"
- },
+ "id": "quackd-jev-stepper",
+ "name": "quackd · Jev Stepper",
+ "owner": "rokbenko",
+ "category": "domain-tools",
+ "starsAtCapture": 229,
"language": "Python",
- "license": "MIT",
- "starsAtCapture": 221,
+ "license": "Apache-2.0",
"summary": {
- "en": "An educational JEV-like visual inference experiment on Apple Silicon with shared context scoring.",
- "zh": "在 Apple Silicon 上的教学向类 JEV 视觉推理实验，共享上下文直接给候选打分。",
- "ja": "Apple Silicon 上の教育向け JEV 風ビジョン推論。共有コンテキストで候補を直接採点。"
+ "en": "A robot orchestration CLI with an optional System One stepper that chooses among permitted discrete robot calls.",
+ "zh": "机器人编排 CLI，可选 System One 步进器在已许可的离散机器人调用中做选择。",
+ "ja": "許可済みの離散ロボット呼び出しから選ぶ任意 System One ステッパー付きロボット制御 CLI。"
},
"decision": {
- "en": "Scores visual or textual candidates in one forward pass and returns calibrated probabilities.",
- "zh": "一次前向为视觉或文本候选打分，并返回校准概率。",
- "ja": "1 回の前向き計算で視覚／テキスト候補を採点し、校正確率を返します。"
+ "en": "Routes Choice and Noul questions through the TypeSafe System One client while the executor keeps safety gates.",
+ "zh": "通过 TypeSafe System One 客户端路由 Choice 与 Noul，执行器仍保留安全门。",
+ "ja": "Choice と Noul を TypeSafe System One クライアント経由で送り、実行側が安全ゲートを維持します。"
+ },
+ "repoUrl": "https://github.com/rokbenko/quackd",
+ "evidenceUrl": "https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "moritzkremb:jev-voice-browser",
- "name": "JEV Voice Browser",
- "owner": "moritzkremb",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/moritzkremb/jev-voice-browser",
- "evidenceUrl": "https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123",
+ "id": "jev-codex-router",
+ "name": "JEV Codex Router",
+ "owner": "0xNatoshi",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/0xNatoshi/jev-codex-router",
+ "evidenceUrl": "https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/moritzkremb/jev-voice-browser/blob/054db0f3dbf537af63a8117632d3f941ccd520e1/src/jev.js#L123"
+ "url": "https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py"
},
- "language": "",
+ "language": "Python",
"license": "MIT",
- "starsAtCapture": 216,
+ "starsAtCapture": 229,
"summary": {
- "en": "Controls a Playwright browser by sending incremental speech transcripts to Jev.",
- "zh": "用语音控制 Playwright 浏览器，将逐步转写的口令交给 Jev 判断。",
- "ja": "増分音声トランスクリプトを JEV に送信することで、Playwright ブラウザを制御します。"
+ "en": "Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode.",
+ "zh": "每轮先让 Jev 判断任务类型与难度，再由本地策略为 Codex 选模型、思考深度和速度档。",
+ "ja": "JEV で Codex の各ターンを分類し、ローカルルールに従ってモデル、推論強度、速度モードを選びます。"
},
"decision": {
- "en": "Chooses intent, elements, URLs or verbatim spans and judges command completeness and sensitive actions.",
- "zh": "选择意图、元素、网址或原文片段，并判断口令是否完整、是否涉及敏感动作。",
- "ja": "インテント、要素、URL、または逐語的な範囲を選択し、コマンドの完全性と機密性の高いアクションを判断します。"
+ "en": "Classifies task difficulty and reasoning needs before local policy selects a model configuration.",
+ "zh": "对当前轮次的任务层级与推理需求分类，本地规则据此选择模型配置。",
+ "ja": "ローカルポリシーがモデル設定を選ぶ前に、タスクの難易度と必要な推論強度を分類します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3070,7 +3266,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 214,
+ "starsAtCapture": 226,
"summary": {
"en": "The JavaScript and TypeScript SDK published by TypeSafe, with typed Jev requests and answers.",
"zh": "TypeSafe 组织发布的 JavaScript 与 TypeScript SDK，提供 Jev 请求和回答类型。",
@@ -3085,33 +3281,33 @@
"runtimeVerified": false
},
{
- "id": "abide",
- "name": "Abide",
- "owner": "coldteadotai",
- "category": "safety-review",
- "repoUrl": "https://github.com/coldteadotai/abide",
- "evidenceUrl": "https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md",
- "language": "TypeScript",
+ "id": "jeff",
+ "name": "Jeff",
+ "owner": "logan-markewich",
+ "category": "open-models",
+ "repoUrl": "https://github.com/logan-markewich/jeff",
+ "evidenceUrl": "https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md",
+ "language": "Python",
"license": "MIT",
- "starsAtCapture": 207,
+ "starsAtCapture": 219,
"summary": {
- "en": "A coding-agent rule layer for constraints that ordinary linters cannot express.",
- "zh": "为普通 Linter 难以表达的项目规则增加 Agent 级约束。",
- "ja": "通常のリンターでは表現できない制約を扱う、コーディング Agent 向けルール層。"
+ "en": "A self-hosted Jev-compatible service powered by GliFormer.",
+ "zh": "基于 GliFormer、可自托管的 JEV 兼容服务。",
+ "ja": "GliFormer を利用したセルフホスト型の JEV 互換サービス。"
},
"decision": {
- "en": "Scores each agent turn against semantic rules and returns targeted corrections.",
- "zh": "逐轮对照语义规则评分，并把需要修正的约束返回给 Agent。",
- "ja": "セマンティック ルールに対してエージェントが反抗するたびにスコアを付け、対象を絞った修正を返します。"
+ "en": "Preserves the decision API while running the classifier locally.",
+ "zh": "保留简洁的决策 API，同时在本地运行分类器。",
+ "ja": "分類子をローカルで実行している間、意思決定 API を保持します。"
},
"verification": "source-reviewed",
"runtimeVerified": false,
"source": {
"type": "x",
"label": "Original X case",
- "url": "https://x.com/OhansEmmanuel/status/2101034822760288452",
- "author": "@OhansEmmanuel",
- "views": 108458
+ "url": "https://x.com/LoganMarkewich/status/2101142971223781587",
+ "author": "@LoganMarkewich",
+ "views": 29163
}
},
{
@@ -3128,7 +3324,7 @@
},
"language": "TypeScript",
"license": "AGPL-3.0",
- "starsAtCapture": 207,
+ "starsAtCapture": 213,
"summary": {
"en": "A product that turns work artifacts into publishable content with agent skill hooks.",
"zh": "把工作产物转成可发布内容的产品，并挂接 Agent Skill。",
@@ -3143,33 +3339,33 @@
"runtimeVerified": false
},
{
- "id": "jeff",
- "name": "Jeff",
- "owner": "logan-markewich",
- "category": "open-models",
- "repoUrl": "https://github.com/logan-markewich/jeff",
- "evidenceUrl": "https://github.com/logan-markewich/jeff/blob/34b32f99a727c47b679adde33f4702a001e02979/README.md",
- "language": "Python",
+ "id": "abide",
+ "name": "Abide",
+ "owner": "coldteadotai",
+ "category": "safety-review",
+ "repoUrl": "https://github.com/coldteadotai/abide",
+ "evidenceUrl": "https://github.com/coldteadotai/abide/blob/ec3352e873163b74aca1ac9cf3bd0ea69a97723a/README.md",
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 205,
+ "starsAtCapture": 210,
"summary": {
- "en": "A self-hosted Jev-compatible service powered by GliFormer.",
- "zh": "基于 GliFormer、可自托管的 JEV 兼容服务。",
- "ja": "GliFormer を利用したセルフホスト型の JEV 互換サービス。"
+ "en": "A coding-agent rule layer for constraints that ordinary linters cannot express.",
+ "zh": "为普通 Linter 难以表达的项目规则增加 Agent 级约束。",
+ "ja": "通常のリンターでは表現できない制約を扱う、コーディング Agent 向けルール層。"
},
"decision": {
- "en": "Preserves the decision API while running the classifier locally.",
- "zh": "保留简洁的决策 API，同时在本地运行分类器。",
- "ja": "分類子をローカルで実行している間、意思決定 API を保持します。"
+ "en": "Scores each agent turn against semantic rules and returns targeted corrections.",
+ "zh": "逐轮对照语义规则评分，并把需要修正的约束返回给 Agent。",
+ "ja": "セマンティック ルールに対してエージェントが反抗するたびにスコアを付け、対象を絞った修正を返します。"
},
"verification": "source-reviewed",
"runtimeVerified": false,
"source": {
"type": "x",
"label": "Original X case",
- "url": "https://x.com/LoganMarkewich/status/2101142971223781587",
- "author": "@LoganMarkewich",
- "views": 29163
+ "url": "https://x.com/OhansEmmanuel/status/2101034822760288452",
+ "author": "@OhansEmmanuel",
+ "views": 108458
}
},
{
@@ -3177,7 +3373,7 @@
"name": "SiftRank · JEV",
"owner": "noperator",
"category": "search-data",
- "starsAtCapture": 202,
+ "starsAtCapture": 209,
"language": "Go",
"license": "MIT",
"summary": {
@@ -3201,29 +3397,29 @@
"runtimeVerified": false
},
{
- "id": "docjev",
- "name": "DocJEV",
- "owner": "jerryjliu",
- "category": "search-data",
- "starsAtCapture": 201,
- "language": "Python",
- "license": "Apache-2.0",
- "summary": {
- "en": "A document classifier and splitter with a dedicated JEV engine.",
- "zh": "带专用 JEV Engine 的文档分类与切分工具。",
- "ja": "専用 JEV エンジンを備えた文書分類・分割ツール。"
- },
- "decision": {
- "en": "Classifies documents and split boundaries from bounded schemas, with cache and alternate-engine support.",
- "zh": "根据有限 Schema 判断文档类别与切分边界，并支持缓存及备用引擎。",
- "ja": "限定 Schema から文書カテゴリと分割境界を判断し、キャッシュと代替エンジンにも対応します。"
- },
- "repoUrl": "https://github.com/jerryjliu/docjev",
- "evidenceUrl": "https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py",
+ "id": "omnijev-awesome-gallery",
+ "name": "Awesome JEV Gallery",
+ "owner": "OmniJev",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/OmniJev/awesome-jev-gallery",
+ "evidenceUrl": "https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/jerryjliu/docjev/blob/7e6b48d3f7eec8db7b582e7276f2778c0d12ca3d/src/jev_docs/engines/jev.py"
+ "url": "https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md"
+ },
+ "language": "JavaScript",
+ "license": "NOASSERTION",
+ "starsAtCapture": 208,
+ "summary": {
+ "en": "A gallery of System One papers, open reproductions, and independent evaluations.",
+ "zh": "汇总 System One 论文、开放复现与独立评测的 Gallery。",
+ "ja": "System One の論文、オープン再現、独立評価を集めたギャラリー。"
+ },
+ "decision": {
+ "en": "Connects model implementations with research and evaluation evidence.",
+ "zh": "把模型实现与研究、评测证据关联起来。",
+ "ja": "モデル実装を研究および評価の根拠と結び付けます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3242,7 +3438,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 196,
+ "starsAtCapture": 206,
"summary": {
"en": "A local MCP code-quality reviewer returning structured scores to coding Agents.",
"zh": "供编码 Agent 使用的本地 MCP 代码质量检查器，返回多个维度的结构化评分。",
@@ -3256,34 +3452,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "laya-node-runtime",
- "name": "Laya Node Runtime",
- "owner": "receptron",
- "category": "open-models",
- "starsAtCapture": 195,
- "language": "TypeScript",
- "license": "MIT",
- "summary": {
- "en": "A Node.js and TypeScript runtime for the open Laya JEV-compatible decision model.",
- "zh": "面向开放 Laya JEV 兼容决策模型的 Node.js / TypeScript Runtime。",
- "ja": "オープンな Laya JEV 互換判断モデル向け Node.js / TypeScript Runtime。"
- },
- "decision": {
- "en": "Runs the ONNX model locally and returns typed decision probabilities without a hosted JEV call.",
- "zh": "在本地运行 ONNX 模型并返回类型化决策概率，无需调用托管 JEV。",
- "ja": "ONNX モデルをローカル実行し、ホスト型 JEV を呼ばずに型付き判断確率を返します。"
- },
- "repoUrl": "https://github.com/receptron/laya",
- "evidenceUrl": "https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/receptron/laya/blob/6478649e723122ca24bbf5fb69ed1010023c9750/src/laya.ts"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "typesafe-ai:typesafe-sdk-python",
"name": "TypeSafe SDK Python",
@@ -3298,7 +3466,7 @@
},
"language": "Python",
"license": "MIT",
- "starsAtCapture": 188,
+ "starsAtCapture": 205,
"summary": {
"en": "Official TypeSafe Python SDK with synchronous and asynchronous clients for Jev System One, plus question and answer types.",
"zh": "TypeSafe 官方 Python SDK，提供 Jev System One 的同步、异步客户端及问题和回答类型。",
@@ -3312,6 +3480,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "kitze:unclutter",
+ "name": "Unclutter",
+ "owner": "kitze",
+ "category": "safety-review",
+ "repoUrl": "https://github.com/kitze/unclutter",
+ "evidenceUrl": "https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 203,
+ "summary": {
+ "en": "A browser extension using Jev to identify ads, promotions, and subscription popups and save reusable hiding rules.",
+ "zh": "用 Jev 帮浏览器扩展识别网页中的广告、促销与订阅弹窗，并保存可复用的隐藏规则。",
+ "ja": "JEV を使用して広告、プロモーション、購読ポップアップを識別し、再利用可能な非表示ルールを保存するブラウザ拡張機能。"
+ },
+ "decision": {
+ "en": "Evaluates candidate page elements before the extension applies local hiding rules.",
+ "zh": "对候选页面元素做结构化判断，再由扩展应用本地隐藏规则。",
+ "ja": "拡張機能がローカルの非表示ルールを適用する前に、候補のページ要素を評価します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "fatwang2-awesome-jev",
"name": "Awesome JEV (fatwang2)",
@@ -3326,7 +3522,7 @@
},
"language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 187,
+ "starsAtCapture": 198,
"summary": {
"en": "A source-backed JEV directory plus a reusable JEV-only GitHub review workflow.",
"zh": "有源码依据的 JEV 目录，并附带可复用的纯 JEV GitHub Review 工作流。",
@@ -3340,6 +3536,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "embodied-jev",
+ "name": "Embodied JEV",
+ "owner": "FBddcz",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/FBddcz/embodied-jev",
+ "evidenceUrl": "https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py",
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 191,
+ "summary": {
+ "en": "An embodied-agent evaluation stack comparing JEV with local and general-model decision policies.",
+ "zh": "面向具身 Agent 的评测栈，用于比较 JEV、本地模型与通用模型决策策略。",
+ "ja": "JEV、ローカルモデル、汎用モデルの判断ポリシーを比較する身体化 Agent 評価基盤です。"
+ },
+ "decision": {
+ "en": "Offers legal actions as bounded choices, validates returned probabilities, and keeps execution under local policy.",
+ "zh": "把合法动作作为有限选项提交，校验返回概率，并由本地策略控制执行。",
+ "ja": "許可済み操作を有限候補として提示し、返却確率を検証したうえで実行はローカルポリシーが管理します。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "knuckles92:openwhisper",
"name": "OpenWhisper",
@@ -3369,29 +3593,57 @@
"runtimeVerified": false
},
{
- "id": "jev-codex-router",
- "name": "JEV Codex Router",
- "owner": "0xNatoshi",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/0xNatoshi/jev-codex-router",
- "evidenceUrl": "https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py",
+ "id": "ai-python-jev-evaluate",
+ "name": "AI SDK Python · Jev Evaluate",
+ "owner": "vercel-labs",
+ "category": "sdk-integrations",
+ "starsAtCapture": 185,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "Vercel's AI SDK for Python with an experimental evaluate path that asks typed questions through typesafe-ai/jev.",
+ "zh": "Vercel Python AI SDK 的实验性 evaluate 路径，可通过 typesafe-ai/jev 提问类型化问题。",
+ "ja": "typesafe-ai/jev 経由で型付き質問を送る実験的 evaluate を備えた Vercel Python AI SDK。"
+ },
+ "decision": {
+ "en": "Batches Choice, Score, and Boolean questions against shared state via the AI Gateway evaluation protocol.",
+ "zh": "通过 AI Gateway 评估协议，对共享状态批量提交 Choice、Score 与布尔问题。",
+ "ja": "AI Gateway の評価プロトコルで、共有状態に対し Choice・Score・真偽質問をまとめて送ります。"
+ },
+ "repoUrl": "https://github.com/vercel-labs/ai-python",
+ "evidenceUrl": "https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/0xNatoshi/jev-codex-router/blob/8292b519659280884627a962c826ac7721136a64/server/jev_server.py"
+ "url": "https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py"
},
- "language": "Python",
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "compact-adviser",
+ "name": "Compact Adviser",
+ "owner": "kunchenguid",
+ "category": "agent-workflows",
+ "repoUrl": "https://github.com/kunchenguid/compact-adviser",
+ "evidenceUrl": "https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md"
+ },
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 184,
+ "starsAtCapture": 180,
"summary": {
- "en": "Uses Jev to classify each Codex turn, then applies local rules to choose the model, reasoning effort, and speed mode.",
- "zh": "每轮先让 Jev 判断任务类型与难度，再由本地策略为 Codex 选模型、思考深度和速度档。",
- "ja": "JEV で Codex の各ターンを分類し、ローカルルールに従ってモデル、推論強度、速度モードを選びます。"
+ "en": "A coding-agent skill/plugin that advises when to compact context to save tokens.",
+ "zh": "编码 Agent 的 Skill/插件：建议何时压缩上下文以节省 token。",
+ "ja": "いつコンテキストを圧縮してトークンを節約するかを助言するコーディング Agent Skill。"
},
"decision": {
- "en": "Classifies task difficulty and reasoning needs before local policy selects a model configuration.",
- "zh": "对当前轮次的任务层级与推理需求分类，本地规则据此选择模型配置。",
- "ja": "ローカルポリシーがモデル設定を選ぶ前に、タスクの難易度と必要な推論強度を分類します。"
+ "en": "Judges whether the session looks complete enough to compact safely.",
+ "zh": "判断当前会话是否已足够完整、可以安全压缩。",
+ "ja": "セッションが十分完了して安全に圧縮できるかを判断します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3410,7 +3662,7 @@
},
"language": "",
"license": "MIT",
- "starsAtCapture": 177,
+ "starsAtCapture": 178,
"summary": {
"en": "Interlinked adds optional Jev judgments and evidence checks to local coding-agent checks.",
"zh": "Interlinked 在编程 Agent 的本地检查之外，提供可选 Jev 判断与证据检查。",
@@ -3425,88 +3677,34 @@
"runtimeVerified": false
},
{
- "id": "kitze:unclutter",
- "name": "Unclutter",
- "owner": "kitze",
- "category": "safety-review",
- "repoUrl": "https://github.com/kitze/unclutter",
- "evidenceUrl": "https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/kitze/unclutter/blob/9ef9beccc1e57b4e3115ae68644b8fc9c19c29f6/lib/jev.ts"
- },
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 176,
- "summary": {
- "en": "A browser extension using Jev to identify ads, promotions, and subscription popups and save reusable hiding rules.",
- "zh": "用 Jev 帮浏览器扩展识别网页中的广告、促销与订阅弹窗，并保存可复用的隐藏规则。",
- "ja": "JEV を使用して広告、プロモーション、購読ポップアップを識別し、再利用可能な非表示ルールを保存するブラウザ拡張機能。"
- },
- "decision": {
- "en": "Evaluates candidate page elements before the extension applies local hiding rules.",
- "zh": "对候选页面元素做结构化判断，再由扩展应用本地隐藏规则。",
- "ja": "拡張機能がローカルの非表示ルールを適用する前に、候補のページ要素を評価します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "compact-adviser",
- "name": "Compact Adviser",
- "owner": "kunchenguid",
+ "id": "jev-gateway",
+ "name": "jev-gateway",
+ "owner": "vinilana",
"category": "agent-workflows",
- "repoUrl": "https://github.com/kunchenguid/compact-adviser",
- "evidenceUrl": "https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/kunchenguid/compact-adviser/blob/d1655faa16a22b68bff60c3d7deb0123e1e52a53/packages/codex-plugin/skills/compact-adviser/SKILL.md"
- },
+ "repoUrl": "https://github.com/vinilana/jev-gateway",
+ "evidenceUrl": "https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 173,
+ "starsAtCapture": 179,
"summary": {
- "en": "A coding-agent skill/plugin that advises when to compact context to save tokens.",
- "zh": "编码 Agent 的 Skill/插件：建议何时压缩上下文以节省 token。",
- "ja": "いつコンテキストを圧縮してトークンを節約するかを助言するコーディング Agent Skill。"
+ "en": "A gateway for inserting Jev decisions into Claude Code and Codex tool flows.",
+ "zh": "把 JEV 判断接入 Claude Code 与 Codex 工具流程的网关。",
+ "ja": "JEV 決定を Claude Code および Codex ツール フローに挿入するためのゲートウェイ。"
},
"decision": {
- "en": "Judges whether the session looks complete enough to compact safely.",
- "zh": "判断当前会话是否已足够完整、可以安全压缩。",
- "ja": "セッションが十分完了して安全に圧縮できるかを判断します。"
+ "en": "Delegates bounded tool-choice judgments while the coding agent owns the broader task.",
+ "zh": "把有限的工具选择判断分离出来，完整任务仍由编码 Agent 负责。",
+ "ja": "コーディングエージェントがより広範なタスクを所有する一方で、限定されたツール選択の判断を委任します。"
},
"verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "embodied-jev",
- "name": "Embodied JEV",
- "owner": "FBddcz",
- "category": "domain-tools",
- "repoUrl": "https://github.com/FBddcz/embodied-jev",
- "evidenceUrl": "https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py",
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 170,
- "summary": {
- "en": "An embodied-agent evaluation stack comparing JEV with local and general-model decision policies.",
- "zh": "面向具身 Agent 的评测栈，用于比较 JEV、本地模型与通用模型决策策略。",
- "ja": "JEV、ローカルモデル、汎用モデルの判断ポリシーを比較する身体化 Agent 評価基盤です。"
- },
- "decision": {
- "en": "Offers legal actions as bounded choices, validates returned probabilities, and keeps execution under local policy.",
- "zh": "把合法动作作为有限选项提交，校验返回概率，并由本地策略控制执行。",
- "ja": "許可済み操作を有限候補として提示し、返却確率を検証したうえで実行はローカルポリシーが管理します。"
- },
+ "runtimeVerified": false,
"source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/FBddcz/embodied-jev/blob/687210b62f3fad831295ebb1b83dab0b4e013baa/src/embodied_jev/policies.py"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
+ "type": "x",
+ "label": "Original X case",
+ "url": "https://x.com/oviniciuslana/status/2101057880640127230",
+ "author": "@oviniciuslana",
+ "views": 53439
+ }
},
{
"id": "djev-spark",
@@ -3522,7 +3720,7 @@
},
"language": "HTML",
"license": "Not declared",
- "starsAtCapture": 168,
+ "starsAtCapture": 176,
"summary": {
"en": "A DGX Spark container recipe for running DiffusionGemma NVFP4 structured decisions.",
"zh": "在 DGX Spark 上跑 DiffusionGemma NVFP4 结构化决策的容器配方。",
@@ -3537,85 +3735,57 @@
"runtimeVerified": false
},
{
- "id": "perch",
- "name": "Perch",
- "owner": "lakeday-org",
- "category": "developer-tools",
- "repoUrl": "https://github.com/lakeday-org/perch",
- "evidenceUrl": "https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js"
- },
- "language": "JavaScript",
- "license": "MIT",
- "starsAtCapture": 167,
- "summary": {
- "en": "Semantic code linting that asks JEV / System One whether findings should fire.",
- "zh": "语义代码检查：用 JEV / System One 判断某条发现是否该触发。",
- "ja": "発見を発火すべきかを JEV / System One に問うセマンティック lint。"
- },
- "decision": {
- "en": "Scores lint candidates against project rules before reporting issues.",
- "zh": "在报告问题前，按项目规则给 lint 候选项打分。",
- "ja": "問題報告の前に、プロジェクト規則で lint 候補を採点します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "michaelliv:runline",
- "name": "Runline",
- "owner": "Michaelliv",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/Michaelliv/runline",
- "evidenceUrl": "https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md",
+ "id": "billionsbobby-jevrouter",
+ "name": "JevRouter",
+ "owner": "BillionsBobby",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/BillionsBobby/JevRouter",
+ "evidenceUrl": "https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md"
+ "url": "https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts"
},
"language": "TypeScript",
- "license": "Unknown",
- "starsAtCapture": 163,
+ "license": "MIT",
+ "starsAtCapture": 173,
"summary": {
- "en": "A TypeSafe plugin exposing Jev decisions as callable actions in Runline Agent JavaScript.",
- "zh": "Runline 的 TypeSafe 插件，把 Jev 判断作为 Agent JavaScript 可调用的动作。",
- "ja": "Runline Agent JavaScript で呼び出し可能なアクションとして JEV 決定を公開する TypeSafe プラグイン。"
+ "en": "A lightweight JEV router for models, tools, and subagents.",
+ "zh": "用于模型、工具与子 Agent 的轻量 JEV 路由器。",
+ "ja": "モデル、ツール、サブ Agent 向けの軽量 JEV ルーター。"
},
"decision": {
- "en": "evaluate, choice, score and noul actions submit questions and retain answers and usage.",
- "zh": "evaluate、choice、score、noul 等动作提交问题并保留模型答案和用量。",
- "ja": "評価、選択、スコア、および noul アクションは質問を送信し、回答と使用法を保持します。"
+ "en": "Chooses a model, tool, or subagent from caller-supplied candidates.",
+ "zh": "从调用方提供的候选中选择模型、工具或子 Agent。",
+ "ja": "呼び出し側が与えた候補からモデル、ツール、サブ Agent を選びます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "omnijev-awesome-gallery",
- "name": "Awesome JEV Gallery",
- "owner": "OmniJev",
+ "id": "perch",
+ "name": "Perch",
+ "owner": "lakeday-org",
"category": "developer-tools",
- "repoUrl": "https://github.com/OmniJev/awesome-jev-gallery",
- "evidenceUrl": "https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md",
+ "repoUrl": "https://github.com/lakeday-org/perch",
+ "evidenceUrl": "https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/OmniJev/awesome-jev-gallery/blob/2b84d22ab0fb9e54098788955187feb023c40ab3/README.md"
+ "url": "https://github.com/lakeday-org/perch/blob/54a38d6034264dc507e97294b82316c347fe5a5e/src/systemone.js"
},
"language": "JavaScript",
- "license": "NOASSERTION",
- "starsAtCapture": 159,
+ "license": "MIT",
+ "starsAtCapture": 170,
"summary": {
- "en": "A gallery of System One papers, open reproductions, and independent evaluations.",
- "zh": "汇总 System One 论文、开放复现与独立评测的 Gallery。",
- "ja": "System One の論文、オープン再現、独立評価を集めたギャラリー。"
+ "en": "Semantic code linting that asks JEV / System One whether findings should fire.",
+ "zh": "语义代码检查：用 JEV / System One 判断某条发现是否该触发。",
+ "ja": "発見を発火すべきかを JEV / System One に問うセマンティック lint。"
},
"decision": {
- "en": "Connects model implementations with research and evaluation evidence.",
- "zh": "把模型实现与研究、评测证据关联起来。",
- "ja": "モデル実装を研究および評価の根拠と結び付けます。"
+ "en": "Scores lint candidates against project rules before reporting issues.",
+ "zh": "在报告问题前，按项目规则给 lint 候选项打分。",
+ "ja": "問題報告の前に、プロジェクト規則で lint 候補を採点します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3634,7 +3804,7 @@
},
"language": "JavaScript",
"license": "Unknown",
- "starsAtCapture": 155,
+ "starsAtCapture": 169,
"summary": {
"en": "A browser driving simulator where Jev chooses among locally generated paths and speeds.",
"zh": "在浏览器里开一辆小车，让 Jev 从提前算好的路线和速度里选下一步。",
@@ -3649,85 +3819,85 @@
"runtimeVerified": false
},
{
- "id": "heyjunpenn-awesome-jev",
- "name": "Awesome JEV (HeyJunPenn)",
- "owner": "heyjunpenn",
- "category": "developer-tools",
- "starsAtCapture": 148,
- "language": "Astro",
- "license": "MIT",
- "summary": {
- "en": "A multilingual, community-maintained catalogue of open-source JEV projects.",
- "zh": "社区维护的多语言 JEV 开源项目目录。",
- "ja": "コミュニティ運営の多言語 JEV オープンソース・カタログ。"
- },
- "decision": {
- "en": "Indexes implementations and integrations while retaining repository-level provenance for discovery.",
- "zh": "索引实现与集成，并保留仓库级来源信息用于发现。",
- "ja": "実装・連携を索引化し、発見用にリポジトリ単位の出典を保持します。"
- },
- "repoUrl": "https://github.com/heyjunpenn/awesome-jev",
- "evidenceUrl": "https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md",
+ "id": "michaelliv:runline",
+ "name": "Runline",
+ "owner": "Michaelliv",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/Michaelliv/runline",
+ "evidenceUrl": "https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/heyjunpenn/awesome-jev/blob/eabb8446bdadd6744adce2e30871ef0691cf34e5/README.md"
+ "url": "https://github.com/Michaelliv/runline/blob/6bdddfa82cd95b6b9a07e57fd93a271ae83a0d1b/README.md"
+ },
+ "language": "TypeScript",
+ "license": "Unknown",
+ "starsAtCapture": 163,
+ "summary": {
+ "en": "A TypeSafe plugin exposing Jev decisions as callable actions in Runline Agent JavaScript.",
+ "zh": "Runline 的 TypeSafe 插件，把 Jev 判断作为 Agent JavaScript 可调用的动作。",
+ "ja": "Runline Agent JavaScript で呼び出し可能なアクションとして JEV 決定を公開する TypeSafe プラグイン。"
+ },
+ "decision": {
+ "en": "evaluate, choice, score and noul actions submit questions and retain answers and usage.",
+ "zh": "evaluate、choice、score、noul 等动作提交问题并保留模型答案和用量。",
+ "ja": "評価、選択、スコア、および noul アクションは質問を送信し、回答と使用法を保持します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "billionsbobby-jevrouter",
- "name": "JevRouter",
- "owner": "BillionsBobby",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/BillionsBobby/JevRouter",
- "evidenceUrl": "https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts",
+ "id": "hellogumbo-awesome-jev",
+ "name": "Awesome Jev",
+ "owner": "hellogumbo",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/hellogumbo/awesome-jev",
+ "evidenceUrl": "https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/BillionsBobby/JevRouter/blob/715970774ae8070e958e83ac9b1a780b32a9184c/src/provider.ts"
+ "url": "https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json"
},
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 145,
+ "language": "JavaScript",
+ "license": "CC0-1.0",
+ "starsAtCapture": 149,
"summary": {
- "en": "A lightweight JEV router for models, tools, and subagents.",
- "zh": "用于模型、工具与子 Agent 的轻量 JEV 路由器。",
- "ja": "モデル、ツール、サブ Agent 向けの軽量 JEV ルーター。"
+ "en": "The source catalogue behind awesomejev.com, covering repos, integrations, demos, and research.",
+ "zh": "awesomejev.com 的源数据目录，覆盖仓库、集成、演示与研究。",
+ "ja": "awesomejev.com の元カタログで、リポジトリ、連携、デモ、研究を収録。"
},
"decision": {
- "en": "Chooses a model, tool, or subagent from caller-supplied candidates.",
- "zh": "从调用方提供的候选中选择模型、工具或子 Agent。",
- "ja": "呼び出し側が与えた候補からモデル、ツール、サブ Agent を選びます。"
+ "en": "Indexes the wider ecosystem while separating repositories, sites, and articles.",
+ "zh": "索引更广的生态，并区分仓库、网站与文章。",
+ "ja": "広いエコシステムを索引化し、リポジトリ、サイト、記事を区別します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "neurolink-typesafe",
- "name": "NeuroLink · TypeSafe",
- "owner": "Juspay",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/Juspay/neurolink",
- "evidenceUrl": "https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts",
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 137,
+ "id": "llm2jev",
+ "name": "LLM2Jev",
+ "owner": "Yinsongxu",
+ "category": "open-models",
+ "starsAtCapture": 142,
+ "language": "Python",
+ "license": "Apache-2.0",
"summary": {
- "en": "NeuroLink ships a TypeSafe provider and classifier-routing support for JEV.",
- "zh": "NeuroLink 提供 TypeSafe Provider 与基于 JEV 的分类路由能力。",
- "ja": "NeuroLink は TypeSafe プロバイダーと JEV 分類ルーティングを提供します。"
+ "en": "An adapter that turns local language models into JEV-compatible structured decision engines.",
+ "zh": "把本地语言模型转成 JEV 兼容结构化决策引擎的适配器。",
+ "ja": "ローカル言語モデルを JEV 互換の構造化判断エンジンへ変換するアダプタ。"
},
"decision": {
- "en": "Validates typed decision requests, calls JEV directly or through a gateway, and normalizes answers for routing.",
- "zh": "校验类型化决策请求，直连或经网关调用 JEV，再将答案标准化用于路由。",
- "ja": "型付き判断リクエストを検証し、JEV を直接またはゲートウェイ経由で呼び出してルーティング用に整形します。"
+ "en": "Uses prefill-only binary inference to produce Choice, Score, and Noul probabilities.",
+ "zh": "通过仅预填充的二元推理生成 Choice、Score 与 Noul 概率。",
+ "ja": "prefill のみの二値推論で Choice・Score・Noul の確率を生成します。"
},
+ "repoUrl": "https://github.com/Yinsongxu/LLM2Jev",
+ "evidenceUrl": "https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts"
+ "url": "https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3746,7 +3916,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 134,
+ "starsAtCapture": 140,
"summary": {
"en": "A Claude Code plugin that trims long Bash output with TypeSafe JEV before the model sees it.",
"zh": "Claude Code 插件：在模型看到长 Bash 输出前，用 TypeSafe JEV 裁剪。",
@@ -3774,7 +3944,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 134,
+ "starsAtCapture": 140,
"summary": {
"en": "A Pi extension that flags tool risks before execution and checks output for secrets and failure types.",
"zh": "Pi 编码 Agent 扩展：执行前提示工具风险，执行后检查秘密泄露与失败类型。",
@@ -3789,29 +3959,29 @@
"runtimeVerified": false
},
{
- "id": "hackclub:ai",
- "name": "AI",
- "owner": "hackclub",
+ "id": "neurolink-typesafe",
+ "name": "NeuroLink · TypeSafe",
+ "owner": "Juspay",
"category": "sdk-integrations",
- "repoUrl": "https://github.com/hackclub/ai",
- "evidenceUrl": "https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md"
- },
+ "repoUrl": "https://github.com/Juspay/neurolink",
+ "evidenceUrl": "https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts",
"language": "TypeScript",
- "license": "Unknown",
- "starsAtCapture": 133,
+ "license": "MIT",
+ "starsAtCapture": 138,
"summary": {
- "en": "A Jev forwarding endpoint in the Hack Club AI proxy, using its authentication, limits and usage logging.",
- "zh": "Hack Club AI 代理中的 Jev 转发接口，复用已有鉴权、限额和用量记录。",
- "ja": "Hack Club AI プロキシの JEV 転送エンドポイント。認証、制限、使用状況ログを使用します。"
+ "en": "NeuroLink ships a TypeSafe provider and classifier-routing support for JEV.",
+ "zh": "NeuroLink 提供 TypeSafe Provider 与基于 JEV 的分类路由能力。",
+ "ja": "NeuroLink は TypeSafe プロバイダーと JEV 分類ルーティングを提供します。"
},
"decision": {
- "en": "Forwards authorized structured requests to TypeSafe and returns answers with usage accounting.",
- "zh": "把获准用户的结构化请求转发到 TypeSafe，返回模型结果并记录用量。",
- "ja": "承認された構造化リクエストを TypeSafe に転送し、使用量アカウンティングを含む回答を返します。"
+ "en": "Validates typed decision requests, calls JEV directly or through a gateway, and normalizes answers for routing.",
+ "zh": "校验类型化决策请求，直连或经网关调用 JEV，再将答案标准化用于路由。",
+ "ja": "型付き判断リクエストを検証し、JEV を直接またはゲートウェイ経由で呼び出してルーティング用に整形します。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/Juspay/neurolink/blob/e2d87c355cf9e725ae5edc3c6db6c05fa659aaa0/src/lib/providers/typesafe.ts"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3830,7 +4000,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 132,
+ "starsAtCapture": 136,
"summary": {
"en": "Adds checks for project rules, out-of-scope actions, repeated failures, and completion claims to Pi Agents.",
"zh": "给 Pi Agent 加项目规则、越界操作、重复失败和完成声明的检查。",
@@ -3845,85 +4015,85 @@
"runtimeVerified": false
},
{
- "id": "building-with-jev-skill",
- "name": "Building with JEV Skill",
- "owner": "dbreunig",
- "category": "developer-tools",
- "repoUrl": "https://github.com/dbreunig/building-with-jev-skill",
- "evidenceUrl": "https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md",
+ "id": "dasheng-jev-reading",
+ "name": "Dasheng",
+ "owner": "wquguru",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/wquguru/dasheng",
+ "evidenceUrl": "https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md"
+ "url": "https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js"
},
- "language": "Not declared",
- "license": "Not declared",
- "starsAtCapture": 128,
+ "language": "JavaScript",
+ "license": "Unknown",
+ "starsAtCapture": 135,
"summary": {
- "en": "An agent skill for writing and improving programs that call JEV / System One.",
- "zh": "帮助编写与改进调用 JEV / System One 程序的 Agent Skill。",
- "ja": "JEV / System One を呼ぶプログラムの作成・改善を助ける Agent Skill。"
+ "en": "A reading-practice app that combines streaming ASR with per-word JEV judgments.",
+ "zh": "把流式 ASR 与 JEV 逐词判断结合的朗读练习应用。",
+ "ja": "ストリーミング ASR と単語ごとの JEV 判断を組み合わせた音読練習アプリ。"
},
"decision": {
- "en": "Advises how to frame state and questions so JEV returns usable typed answers.",
- "zh": "指导如何组织 state 与 questions，让 JEV 返回可用的类型化答案。",
- "ja": "使える型付き回答になるよう、state と questions の組み立て方を助言します。"
+ "en": "Judges word matches, error classes, and sentence similarity from aligned text.",
+ "zh": "根据对齐文本判断单词匹配、错误类别与句子相似度。",
+ "ja": "整列済みテキストから単語一致、誤り分類、文の類似度を判断します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "arc-cua-typesafe",
- "name": "Arc CUA · TypeSafe Policy",
- "owner": "shhivv",
- "category": "browser-computer-use",
- "starsAtCapture": 127,
+ "id": "romanslack:jev-drone",
+ "name": "JEV Drone",
+ "owner": "RomanSlack",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/RomanSlack/jev-drone",
+ "evidenceUrl": "https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184"
+ },
"language": "Python",
"license": "MIT",
+ "starsAtCapture": 134,
"summary": {
- "en": "A fast computer-use action layer with a TypeSafe policy implementation.",
- "zh": "带 TypeSafe Policy 实现的高速 Computer Use 动作层。",
- "ja": "TypeSafe Policy 実装を備えた高速 Computer Use アクション層。"
+ "en": "A MuJoCo drone experiment deriving scene features from camera buffers for Jev tactical advice.",
+ "zh": "MuJoCo 无人机仿真实验：从相机缓冲区提取场景，Jev 提供战术动作建议。",
+ "ja": "JEV 戦術的なアドバイスのためにカメラ バッファーからシーンの特徴を導出する MuJoCo ドローン実験。"
},
"decision": {
- "en": "Selects the next allowed UI action from observed state while the local harness owns execution.",
- "zh": "根据观察状态选择下一步允许的 UI 动作，执行仍由本地 Harness 负责。",
- "ja": "観測状態から次の許可済み UI 操作を選び、実行はローカル Harness が担います。"
- },
- "repoUrl": "https://github.com/shhivv/arc-cua",
- "evidenceUrl": "https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py"
+ "en": "Uses range sectors, obstacle heights and target state to choose maneuvers and assess risk or target loss.",
+ "zh": "根据距离扇区、障碍高度与目标状态选择机动、评估风险和目标丢失状态。",
+ "ja": "範囲セクター、障害物の高さ、目標の状態を使用して、操縦を選択し、リスクや目標の損失を評価します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "hellogumbo-awesome-jev",
- "name": "Awesome Jev",
- "owner": "hellogumbo",
- "category": "developer-tools",
- "repoUrl": "https://github.com/hellogumbo/awesome-jev",
- "evidenceUrl": "https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json",
+ "id": "hackclub:ai",
+ "name": "AI",
+ "owner": "hackclub",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/hackclub/ai",
+ "evidenceUrl": "https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/hellogumbo/awesome-jev/blob/e18fa74fa53e27395a074cf94c7a127e912fb090/data/projects.json"
+ "url": "https://github.com/hackclub/ai/blob/a76ea2cb159f707a60107935a5b2e0dbdc7455f5/README.md"
},
- "language": "JavaScript",
- "license": "CC0-1.0",
- "starsAtCapture": 127,
+ "language": "TypeScript",
+ "license": "Unknown",
+ "starsAtCapture": 133,
"summary": {
- "en": "The source catalogue behind awesomejev.com, covering repos, integrations, demos, and research.",
- "zh": "awesomejev.com 的源数据目录，覆盖仓库、集成、演示与研究。",
- "ja": "awesomejev.com の元カタログで、リポジトリ、連携、デモ、研究を収録。"
+ "en": "A Jev forwarding endpoint in the Hack Club AI proxy, using its authentication, limits and usage logging.",
+ "zh": "Hack Club AI 代理中的 Jev 转发接口，复用已有鉴权、限额和用量记录。",
+ "ja": "Hack Club AI プロキシの JEV 転送エンドポイント。認証、制限、使用状況ログを使用します。"
},
"decision": {
- "en": "Indexes the wider ecosystem while separating repositories, sites, and articles.",
- "zh": "索引更广的生态，并区分仓库、网站与文章。",
- "ja": "広いエコシステムを索引化し、リポジトリ、サイト、記事を区別します。"
+ "en": "Forwards authorized structured requests to TypeSafe and returns answers with usage accounting.",
+ "zh": "把获准用户的结构化请求转发到 TypeSafe，返回模型结果并记录用量。",
+ "ja": "承認された構造化リクエストを TypeSafe に転送し、使用量アカウンティングを含む回答を返します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -3942,7 +4112,7 @@
},
"language": "JavaScript",
"license": "CC0-1.0",
- "starsAtCapture": 125,
+ "starsAtCapture": 130,
"summary": {
"en": "A TypeSafe JEV resource list centered on typed, confidence-aware decisions.",
"zh": "围绕类型化、置信度感知决策整理的 TypeSafe JEV 资源列表。",
@@ -3957,62 +4127,90 @@
"runtimeVerified": false
},
{
- "id": "llm2jev",
- "name": "LLM2Jev",
- "owner": "Yinsongxu",
- "category": "open-models",
- "starsAtCapture": 125,
+ "id": "jev-semgrep",
+ "name": "jev-semgrep",
+ "owner": "uehaj",
+ "category": "search-data",
+ "repoUrl": "https://github.com/uehaj/jev-semgrep",
+ "evidenceUrl": "https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md",
+ "language": "JavaScript",
+ "license": "Not declared",
+ "starsAtCapture": 130,
+ "summary": {
+ "en": "Meaning-based grep with multilingual AND, OR, NOT, and probability thresholds.",
+ "zh": "支持跨语言 AND、OR、NOT 与概率阈值的“按意思搜索”grep。",
+ "ja": "多言語の AND、OR、NOT、確率しきい値を使用した意味ベースの grep。"
+ },
+ "decision": {
+ "en": "Scores whether each line satisfies a proposition, then applies grep controls locally.",
+ "zh": "判断每行是否满足命题，再由本地代码应用 grep 控制项。",
+ "ja": "各行が命題を満たすかどうかをスコアリングし、grep コントロールをローカルに適用します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false,
+ "source": {
+ "type": "x",
+ "label": "Original X case",
+ "url": "https://x.com/uehaj/status/2101198713834381716",
+ "author": "@uehaj",
+ "views": 81889
+ }
+ },
+ {
+ "id": "arc-cua-typesafe",
+ "name": "Arc CUA · TypeSafe Policy",
+ "owner": "shhivv",
+ "category": "browser-computer-use",
+ "starsAtCapture": 128,
"language": "Python",
- "license": "Apache-2.0",
+ "license": "MIT",
"summary": {
- "en": "An adapter that turns local language models into JEV-compatible structured decision engines.",
- "zh": "把本地语言模型转成 JEV 兼容结构化决策引擎的适配器。",
- "ja": "ローカル言語モデルを JEV 互換の構造化判断エンジンへ変換するアダプタ。"
+ "en": "A fast computer-use action layer with a TypeSafe policy implementation.",
+ "zh": "带 TypeSafe Policy 实现的高速 Computer Use 动作层。",
+ "ja": "TypeSafe Policy 実装を備えた高速 Computer Use アクション層。"
},
"decision": {
- "en": "Uses prefill-only binary inference to produce Choice, Score, and Noul probabilities.",
- "zh": "通过仅预填充的二元推理生成 Choice、Score 与 Noul 概率。",
- "ja": "prefill のみの二値推論で Choice・Score・Noul の確率を生成します。"
+ "en": "Selects the next allowed UI action from observed state while the local harness owns execution.",
+ "zh": "根据观察状态选择下一步允许的 UI 动作，执行仍由本地 Harness 负责。",
+ "ja": "観測状態から次の許可済み UI 操作を選び、実行はローカル Harness が担います。"
},
- "repoUrl": "https://github.com/Yinsongxu/LLM2Jev",
- "evidenceUrl": "https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py",
+ "repoUrl": "https://github.com/shhivv/arc-cua",
+ "evidenceUrl": "https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/Yinsongxu/LLM2Jev/blob/6fbbf74fc604f92fc9b0b67f0989157bb2908229/src/llm2jev/inference/binary.py"
+ "url": "https://github.com/shhivv/arc-cua/blob/6d47ce6c906d7d8586d3553e277c1c3611c4f6af/src/arc_cua/policies/typesafe.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "jev-semgrep",
- "name": "jev-semgrep",
- "owner": "uehaj",
- "category": "search-data",
- "repoUrl": "https://github.com/uehaj/jev-semgrep",
- "evidenceUrl": "https://github.com/uehaj/jev-semgrep/blob/ba6ef50f85d0c5d6caa4db102ee4db4a08c85dd2/README.md",
- "language": "JavaScript",
+ "id": "building-with-jev-skill",
+ "name": "Building with JEV Skill",
+ "owner": "dbreunig",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/dbreunig/building-with-jev-skill",
+ "evidenceUrl": "https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/dbreunig/building-with-jev-skill/blob/04fe3666c6b8b8abfec1271c0e581c823a181f6d/skills/jev/SKILL.md"
+ },
+ "language": "Not declared",
"license": "Not declared",
- "starsAtCapture": 124,
+ "starsAtCapture": 128,
"summary": {
- "en": "Meaning-based grep with multilingual AND, OR, NOT, and probability thresholds.",
- "zh": "支持跨语言 AND、OR、NOT 与概率阈值的“按意思搜索”grep。",
- "ja": "多言語の AND、OR、NOT、確率しきい値を使用した意味ベースの grep。"
+ "en": "An agent skill for writing and improving programs that call JEV / System One.",
+ "zh": "帮助编写与改进调用 JEV / System One 程序的 Agent Skill。",
+ "ja": "JEV / System One を呼ぶプログラムの作成・改善を助ける Agent Skill。"
},
"decision": {
- "en": "Scores whether each line satisfies a proposition, then applies grep controls locally.",
- "zh": "判断每行是否满足命题，再由本地代码应用 grep 控制项。",
- "ja": "各行が命題を満たすかどうかをスコアリングし、grep コントロールをローカルに適用します。"
+ "en": "Advises how to frame state and questions so JEV returns usable typed answers.",
+ "zh": "指导如何组织 state 与 questions，让 JEV 返回可用的类型化答案。",
+ "ja": "使える型付き回答になるよう、state と questions の組み立て方を助言します。"
},
"verification": "source-reviewed",
- "runtimeVerified": false,
- "source": {
- "type": "x",
- "label": "Original X case",
- "url": "https://x.com/uehaj/status/2101198713834381716",
- "author": "@uehaj",
- "views": 81889
- }
+ "runtimeVerified": false
},
{
"id": "promethe-us-awesome-jev",
@@ -4042,6 +4240,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "reflex-open-decision-model",
+ "name": "Reflex",
+ "owner": "kshetrajna12",
+ "category": "open-models",
+ "repoUrl": "https://github.com/kshetrajna12/reflex",
+ "evidenceUrl": "https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md"
+ },
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 122,
+ "summary": {
+ "en": "A small open decision model that recreates the JEV/System One interface on Qwen.",
+ "zh": "在 Qwen 上复现 JEV/System One 接口的小型开放决策模型。",
+ "ja": "Qwen 上で JEV/System One インターフェースを再現する小型オープン判断モデル。"
+ },
+ "decision": {
+ "en": "Returns calibrated probabilities for state plus typed questions.",
+ "zh": "对状态与类型化问题返回校准概率。",
+ "ja": "状態と型付き質問に対して較正済み確率を返します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "danieljvdm:effect-agent",
"name": "Effect Agent",
@@ -4070,62 +4296,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "dasheng-jev-reading",
- "name": "Dasheng",
- "owner": "wquguru",
- "category": "domain-tools",
- "repoUrl": "https://github.com/wquguru/dasheng",
- "evidenceUrl": "https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/wquguru/dasheng/blob/1bacff4a075527e6c02da242a72d117e7cb3286b/lib/jev.js"
- },
- "language": "JavaScript",
- "license": "Unknown",
- "starsAtCapture": 119,
- "summary": {
- "en": "A reading-practice app that combines streaming ASR with per-word JEV judgments.",
- "zh": "把流式 ASR 与 JEV 逐词判断结合的朗读练习应用。",
- "ja": "ストリーミング ASR と単語ごとの JEV 判断を組み合わせた音読練習アプリ。"
- },
- "decision": {
- "en": "Judges word matches, error classes, and sentence similarity from aligned text.",
- "zh": "根据对齐文本判断单词匹配、错误类别与句子相似度。",
- "ja": "整列済みテキストから単語一致、誤り分類、文の類似度を判断します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "romanslack:jev-drone",
- "name": "JEV Drone",
- "owner": "RomanSlack",
- "category": "domain-tools",
- "repoUrl": "https://github.com/RomanSlack/jev-drone",
- "evidenceUrl": "https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/RomanSlack/jev-drone/blob/cbeb53ce4f17a06ea490ae43effcdad231143610/tactics.py#L184"
- },
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 119,
- "summary": {
- "en": "A MuJoCo drone experiment deriving scene features from camera buffers for Jev tactical advice.",
- "zh": "MuJoCo 无人机仿真实验：从相机缓冲区提取场景，Jev 提供战术动作建议。",
- "ja": "JEV 戦術的なアドバイスのためにカメラ バッファーからシーンの特徴を導出する MuJoCo ドローン実験。"
- },
- "decision": {
- "en": "Uses range sectors, obstacle heights and target state to choose maneuvers and assess risk or target loss.",
- "zh": "根据距离扇区、障碍高度与目标状态选择机动、评估风险和目标丢失状态。",
- "ja": "範囲セクター、障害物の高さ、目標の状態を使用して、操縦を選択し、リスクや目標の損失を評価します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "ldbumble:taskuary",
"name": "Taskuary",
@@ -4140,7 +4310,7 @@
},
"language": "Python",
"license": "MIT",
- "starsAtCapture": 116,
+ "starsAtCapture": 118,
"summary": {
"en": "An optional Jev judgment module in Taskuary for checking user-defined conditions on task state.",
"zh": "Taskuary 的可选 Jev 判断模块，对任务运行状态检查用户定义的条件。",
@@ -4155,87 +4325,29 @@
"runtimeVerified": false
},
{
- "id": "jev-gateway",
- "name": "jev-gateway",
- "owner": "vinilana",
+ "id": "jev-dsh-decision",
+ "name": "JEV DSH Decision",
+ "owner": "Devin-AXIS",
"category": "agent-workflows",
- "repoUrl": "https://github.com/vinilana/jev-gateway",
- "evidenceUrl": "https://github.com/vinilana/jev-gateway/blob/9463952bf118773fb955427d2725a98f76546233/README.md",
- "language": "TypeScript",
- "license": "MIT",
"starsAtCapture": 115,
+ "language": "JavaScript",
+ "license": "NOASSERTION",
"summary": {
- "en": "A gateway for inserting Jev decisions into Claude Code and Codex tool flows.",
- "zh": "把 JEV 判断接入 Claude Code 与 Codex 工具流程的网关。",
- "ja": "JEV 決定を Claude Code および Codex ツール フローに挿入するためのゲートウェイ。"
- },
- "decision": {
- "en": "Delegates bounded tool-choice judgments while the coding agent owns the broader task.",
- "zh": "把有限的工具选择判断分离出来，完整任务仍由编码 Agent 负责。",
- "ja": "コーディングエージェントがより広範なタスクを所有する一方で、限定されたツール選択の判断を委任します。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false,
- "source": {
- "type": "x",
- "label": "Original X case",
- "url": "https://x.com/oviniciuslana/status/2101057880640127230",
- "author": "@oviniciuslana",
- "views": 53439
- }
- },
- {
- "id": "stanley-code-jev",
- "name": "Stanley Code",
- "owner": "devagrawal09",
- "category": "developer-tools",
- "repoUrl": "https://github.com/devagrawal09/stanley-code",
- "evidenceUrl": "https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts"
- },
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 111,
- "summary": {
- "en": "Bounded JEV workflows for coding agents with a dedicated adapter.",
- "zh": "为编码 Agent 提供带专用适配器的有限 JEV 工作流。",
- "ja": "専用アダプターを備えたコーディング Agent 向けの限定 JEV ワークフロー。"
+ "en": "A structured JEV decision plugin for DeepSeek Harness and compatible coding-agent hosts.",
+ "zh": "面向 DeepSeek Harness 及兼容编码 Agent Host 的结构化 JEV 决策插件。",
+ "ja": "DeepSeek Harness と互換コーディング Agent ホスト向けの構造化 JEV 判断プラグイン。"
},
"decision": {
- "en": "Turns review and coding checkpoints into typed JEV judgments.",
- "zh": "把审查与编码检查点转成类型化 JEV 判断。",
- "ja": "レビューとコーディングのチェックポイントを型付き JEV 判断に変換します。"
+ "en": "Packages bounded questions as a service and returns typed results to the host harness.",
+ "zh": "把有限问题封装为服务，并向 Host Harness 返回类型化结果。",
+ "ja": "限定質問をサービス化し、ホスト Harness へ型付き結果を返します。"
},
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "reflex-open-decision-model",
- "name": "Reflex",
- "owner": "kshetrajna12",
- "category": "open-models",
- "repoUrl": "https://github.com/kshetrajna12/reflex",
- "evidenceUrl": "https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md",
+ "repoUrl": "https://github.com/Devin-AXIS/jev-dsh-decision",
+ "evidenceUrl": "https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/kshetrajna12/reflex/blob/e21b3b23afdfeee7021a6604fa38f57e7ff5187f/README.md"
- },
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 110,
- "summary": {
- "en": "A small open decision model that recreates the JEV/System One interface on Qwen.",
- "zh": "在 Qwen 上复现 JEV/System One 接口的小型开放决策模型。",
- "ja": "Qwen 上で JEV/System One インターフェースを再現する小型オープン判断モデル。"
- },
- "decision": {
- "en": "Returns calibrated probabilities for state plus typed questions.",
- "zh": "对状态与类型化问题返回校准概率。",
- "ja": "状態と型付き質問に対して較正済み確率を返します。"
+ "url": "https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4254,7 +4366,7 @@
},
"language": "Rust",
"license": "NOASSERTION",
- "starsAtCapture": 109,
+ "starsAtCapture": 114,
"summary": {
"en": "A Rust CLI that ranks agent skills against live session context with JEV.",
"zh": "用 JEV 根据实时会话上下文排序 Agent Skills 的 Rust CLI。",
@@ -4268,34 +4380,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "vinilana:jev-eval-agent",
- "name": "JEV Eval Agent",
- "owner": "vinilana",
- "category": "domain-tools",
- "repoUrl": "https://github.com/vinilana/jev-eval-agent",
- "evidenceUrl": "https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154"
- },
- "language": "HTML",
- "license": "Unknown",
- "starsAtCapture": 103,
- "summary": {
- "en": "Agent tool evaluation harness comparing standard LLM tool selection against Jev routing across 100 mocked tools.",
- "zh": "智能体工具选择基准测试平台，在包含 100 个模拟工具的个人助理环境下对比大模型直接选工具与 Jev 路由的效率。",
- "ja": "標準 LLM ツールの選択と 100 個のモック ツールにわたる JEV ルーティングを比較するエージェント ツール評価ハーネス。"
- },
- "decision": {
- "en": "Employs Jev for two-stage discrete pruning over tool catalogs, narrowing extensive tool lists down to relevant candidates in milliseconds.",
- "zh": "利用 Jev 对上百个工具元数据进行两级离散筛选和相关性判定，把候选工具集从百级快速裁剪至少量候选。",
- "ja": "ツール カタログに対する 2 段階の個別のプルーニングに JEV を採用し、広範なツール リストをミリ秒単位で関連する候補に絞り込みます。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "kraayenjon-awesome-jev",
"name": "Awesome JEV (kraayenjon)",
@@ -4305,7 +4389,7 @@
"evidenceUrl": "https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md",
"language": "",
"license": "NOASSERTION",
- "starsAtCapture": 100,
+ "starsAtCapture": 113,
"summary": {
"en": "A curated guide to JEV use cases, projects, SDKs, tools, and learning resources.",
"zh": "涵盖 JEV 用例、项目、SDK、工具与学习资料的精选指南。",
@@ -4315,67 +4399,67 @@
"en": "Explains typed Choice, Score, and Noul patterns and links them to public ecosystem implementations.",
"zh": "讲解 Choice、Score 与 Noul 类型化模式，并关联公开生态实现。",
"ja": "Choice、Score、Noul の型付きパターンを説明し、公開実装へ案内します。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "formanator-jev",
- "name": "Formanator",
- "owner": "timrogers",
- "category": "domain-tools",
- "repoUrl": "https://github.com/timrogers/formanator",
- "evidenceUrl": "https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs",
+ },
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs"
+ "url": "https://github.com/kraayenjon/awesome-jev/blob/22570dcd8662ae039860a1dfad7d7aec4aff8e15/README.md"
},
- "language": "Rust",
- "license": "MIT",
- "starsAtCapture": 99,
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "awesome-jev-use-cases",
+ "name": "Awesome JEV Use Cases",
+ "owner": "walidboulanouar",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/walidboulanouar/awesome-jev-use-cases",
+ "evidenceUrl": "https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md",
+ "language": "",
+ "license": "CC0-1.0",
+ "starsAtCapture": 114,
"summary": {
- "en": "A CLI and MCP client for benefit claims with optional JEV receipt classification.",
- "zh": "支持可选 JEV 收据分类的福利报销 CLI 与 MCP 客户端。",
- "ja": "任意の JEV レシート分類を備えた福利厚生申請 CLI / MCP クライアント。"
+ "en": "A source-linked catalogue of shipped JEV demos, repositories, reported metrics, and model limits.",
+ "zh": "带原始来源的 JEV 已发布 Demo、仓库、自报指标与模型限制目录。",
+ "ja": "公開済み JEV デモ、リポジトリ、報告値、モデル制約を出典付きで整理したカタログです。"
},
"decision": {
- "en": "Selects the benefit and receipt category, falling back when confidence is low.",
- "zh": "选择福利与收据类别，置信度不足时回退。",
- "ja": "福利厚生とレシートのカテゴリを選び、信頼度が低い場合はフォールバックします。"
+ "en": "Keeps shipped examples separate from unbuilt ideas and links each entry back to its original evidence.",
+ "zh": "把已发布案例与未实现想法分开，并把每条记录链接回原始证据。",
+ "ja": "公開済み事例と未実装アイデアを分け、各項目を元の根拠へリンクします。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "mrnugget:jev-shell-history",
- "name": "JEV Shell History",
- "owner": "mrnugget",
+ "id": "stanley-code-jev",
+ "name": "Stanley Code",
+ "owner": "devagrawal09",
"category": "developer-tools",
- "repoUrl": "https://github.com/mrnugget/jev-shell-history",
- "evidenceUrl": "https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196",
+ "repoUrl": "https://github.com/devagrawal09/stanley-code",
+ "evidenceUrl": "https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196"
+ "url": "https://github.com/devagrawal09/stanley-code/blob/fd092558ebea389c81d44f9b10e826d9a72afaa3/src/adapters/jev.ts"
},
"language": "TypeScript",
- "license": "Unknown",
- "starsAtCapture": 96,
+ "license": "MIT",
+ "starsAtCapture": 113,
"summary": {
- "en": "Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context.",
- "zh": "类似于 Fish 终端样式的 Zsh 历史命令建议工具，利用 Jev 对已有历史记录根据当前上下文进行智能打分排序。",
- "ja": "JEV によってランク付けされた Fish スタイルの Zsh 履歴提案ツール。コンテキストに基づいてローカル履歴から候補コマンドを順序付けします。"
+ "en": "Bounded JEV workflows for coding agents with a dedicated adapter.",
+ "zh": "为编码 Agent 提供带专用适配器的有限 JEV 工作流。",
+ "ja": "専用アダプターを備えたコーディング Agent 向けの限定 JEV ワークフロー。"
},
"decision": {
- "en": "Packages current command input and historical candidates into a Jev request, ranking the most plausible command for inline suggestion.",
- "zh": "将当前敲入的命令前缀与本地 Zsh 历史候选组装为 Jev 请求，由 Jev 评估最佳补全项，仅做行内建议而不自动执行。",
- "ja": "現在のコマンド入力と過去の候補を JEV リクエストにパッケージ化し、インライン提案として最も妥当なコマンドをランク付けします。"
+ "en": "Turns review and coding checkpoints into typed JEV judgments.",
+ "zh": "把审查与编码检查点转成类型化 JEV 判断。",
+ "ja": "レビューとコーディングのチェックポイントを型付き JEV 判断に変換します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4394,7 +4478,7 @@
},
"language": "TypeScript",
"license": "NOASSERTION",
- "starsAtCapture": 95,
+ "starsAtCapture": 110,
"summary": {
"en": "A live Hyperliquid trading agent with JEV as a bounded decision layer.",
"zh": "以 JEV 作为有限决策层的 Hyperliquid 实盘交易 Agent。",
@@ -4422,7 +4506,7 @@
},
"language": "Rust",
"license": "MIT",
- "starsAtCapture": 94,
+ "starsAtCapture": 106,
"summary": {
"en": "A code-quality and coverage tool that uses JEV for bounded quality checks.",
"zh": "使用 JEV 做有限质量检查的代码质量与覆盖率工具。",
@@ -4437,197 +4521,141 @@
"runtimeVerified": false
},
{
- "id": "daseinlabs-open-jev",
- "name": "Open JEV",
- "owner": "daseinlabs",
- "category": "open-models",
- "repoUrl": "https://github.com/daseinlabs/open-jev",
- "evidenceUrl": "https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md",
+ "id": "vinilana:jev-eval-agent",
+ "name": "JEV Eval Agent",
+ "owner": "vinilana",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/vinilana/jev-eval-agent",
+ "evidenceUrl": "https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md"
+ "url": "https://github.com/vinilana/jev-eval-agent/blob/037de1120c84b4b63cdf748e2acf258ff66d7731/agent/lib/jev-router.ts#L1-L154"
},
- "language": "Python",
+ "language": "HTML",
"license": "Unknown",
- "starsAtCapture": 91,
+ "starsAtCapture": 105,
"summary": {
- "en": "An open JEV-style implementation with custom fine-tuning support.",
- "zh": "支持自定义微调的开放 JEV 风格实现。",
- "ja": "カスタム微調整に対応するオープンな JEV 形式の実装。"
+ "en": "Agent tool evaluation harness comparing standard LLM tool selection against Jev routing across 100 mocked tools.",
+ "zh": "智能体工具选择基准测试平台，在包含 100 个模拟工具的个人助理环境下对比大模型直接选工具与 Jev 路由的效率。",
+ "ja": "標準 LLM ツールの選択と 100 個のモック ツールにわたる JEV ルーティングを比較するエージェント ツール評価ハーネス。"
},
"decision": {
- "en": "Runs bounded decision questions on a locally trainable model.",
- "zh": "用可本地训练的模型处理有限决策问题。",
- "ja": "ローカル学習可能なモデルで限定的な判断質問を処理します。"
+ "en": "Employs Jev for two-stage discrete pruning over tool catalogs, narrowing extensive tool lists down to relevant candidates in milliseconds.",
+ "zh": "利用 Jev 对上百个工具元数据进行两级离散筛选和相关性判定，把候选工具集从百级快速裁剪至少量候选。",
+ "ja": "ツール カタログに対する 2 段階の個別のプルーニングに JEV を採用し、広範なツール リストをミリ秒単位で関連する候補に絞り込みます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "rokcso:bluenoise",
- "name": "Bluenoise",
- "owner": "rokcso",
- "category": "agent-workflows",
- "repoUrl": "https://github.com/rokcso/bluenoise",
- "evidenceUrl": "https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts",
+ "id": "mrnugget:jev-shell-history",
+ "name": "JEV Shell History",
+ "owner": "mrnugget",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/mrnugget/jev-shell-history",
+ "evidenceUrl": "https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts"
+ "url": "https://github.com/mrnugget/jev-shell-history/blob/4b2b75d26c0ccf5726263904514a22a8e11659ea/src/suggest.ts#L1-L196"
},
"language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 90,
+ "license": "Unknown",
+ "starsAtCapture": 104,
"summary": {
- "en": "An X/Twitter filtering extension using local rules by default, with optional Jev checks for unmatched replies.",
- "zh": "为 X/Twitter 过滤帖子与回复的浏览器扩展；默认用本地规则，可选择用 Jev 检查未匹配的回复。",
- "ja": "デフォルトでローカル ルールを使用する X/Twitter フィルタリング拡張機能。オプションで JEV が一致しない返信をチェックします。"
+ "en": "Fish-style Zsh history suggestion tool ranked by Jev, ordering candidate commands from local history based on context.",
+ "zh": "类似于 Fish 终端样式的 Zsh 历史命令建议工具，利用 Jev 对已有历史记录根据当前上下文进行智能打分排序。",
+ "ja": "JEV によってランク付けされた Fish スタイルの Zsh 履歴提案ツール。コンテキストに基づいてローカル履歴から候補コマンドを順序付けします。"
},
"decision": {
- "en": "When experimental AI is enabled, Jev evaluates replies missed by rules and local thresholds decide whether to hide them.",
- "zh": "启用实验性 AI 后，将未命中规则的回复交给 Jev 评估，再由本地阈值决定是否隐藏。",
- "ja": "実験的な AI が有効になっている場合、JEV はルールによって見逃された応答を評価し、ローカルのしきい値によってそれらを非表示にするかどうかを決定します。"
+ "en": "Packages current command input and historical candidates into a Jev request, ranking the most plausible command for inline suggestion.",
+ "zh": "将当前敲入的命令前缀与本地 Zsh 历史候选组装为 Jev 请求，由 Jev 评估最佳补全项，仅做行内建议而不自动执行。",
+ "ja": "現在のコマンド入力と過去の候補を JEV リクエストにパッケージ化し、インライン提案として最も妥当なコマンドをランク付けします。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "pithings:advocaat",
- "name": "Advocaat",
- "owner": "pithings",
- "category": "sdk-integrations",
- "repoUrl": "https://github.com/pithings/advocaat",
- "evidenceUrl": "https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md",
+ "id": "daseinlabs-open-jev",
+ "name": "Open JEV",
+ "owner": "daseinlabs",
+ "category": "open-models",
+ "repoUrl": "https://github.com/daseinlabs/open-jev",
+ "evidenceUrl": "https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md"
+ "url": "https://github.com/daseinlabs/open-jev/blob/8a4fbdf712e78c5ef45509a16aacb81facdd79be/README.md"
},
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 89,
+ "language": "Python",
+ "license": "Unknown",
+ "starsAtCapture": 103,
"summary": {
- "en": "A small TypeScript client for asking Jev multiple typed questions about the same data.",
- "zh": "用简短的 TypeScript 调用向 Jev 提问。把同一份数据里的多个判断一次写好，直接拿到概率、选项和分数。",
- "ja": "同じデータについて複数の型付き質問を JEV に送るための、小さな TypeScript クライアント。"
+ "en": "An open JEV-style implementation with custom fine-tuning support.",
+ "zh": "支持自定义微调的开放 JEV 风格实现。",
+ "ja": "カスタム微調整に対応するオープンな JEV 形式の実装。"
},
"decision": {
- "en": "Combines yes/no, Choice and Score questions and maps responses to typed values.",
- "zh": "将带类型的问题转成 System One 请求，并把命名答案映射回调用方。",
- "ja": "Noul、Choice、Score の質問を組み合わせ、回答を型付きの値へ変換します。"
+ "en": "Runs bounded decision questions on a locally trainable model.",
+ "zh": "用可本地训练的模型处理有限决策问题。",
+ "ja": "ローカル学習可能なモデルで限定的な判断質問を処理します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "savka777-jev-use",
- "name": "JEV Use",
- "owner": "savka777",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/savka777/jev-use",
- "evidenceUrl": "https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md"
- },
- "language": "Swift",
+ "id": "siliconlab-openjev",
+ "name": "OpenJev",
+ "owner": "SiliconLabAI",
+ "category": "open-models",
+ "starsAtCapture": 101,
+ "language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 86,
- "summary": {
- "en": "A macOS computer-use harness that drives Accessibility controls without a vision model.",
- "zh": "无需视觉模型、通过无障碍控件操作 macOS 的 Computer Use 工具。",
- "ja": "視覚モデルなしで Accessibility 要素を操作する macOS Computer Use ツール。"
- },
- "decision": {
- "en": "Chooses a visible accessibility element and the next allowed action.",
- "zh": "选择可见无障碍元素与下一步允许的动作。",
- "ja": "表示中の Accessibility 要素と次の許可済み操作を選びます。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "awesome-jev-use-cases",
- "name": "Awesome JEV Use Cases",
- "owner": "walidboulanouar",
- "category": "developer-tools",
- "repoUrl": "https://github.com/walidboulanouar/awesome-jev-use-cases",
- "evidenceUrl": "https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md",
- "language": "",
- "license": "CC0-1.0",
- "starsAtCapture": 84,
"summary": {
- "en": "A source-linked catalogue of shipped JEV demos, repositories, reported metrics, and model limits.",
- "zh": "带原始来源的 JEV 已发布 Demo、仓库、自报指标与模型限制目录。",
- "ja": "公開済み JEV デモ、リポジトリ、報告値、モデル制約を出典付きで整理したカタログです。"
+ "en": "An open-source JEV-compatible implementation for typed decision requests.",
+ "zh": "面向类型化决策请求的开源 JEV 兼容实现。",
+ "ja": "型付き判断リクエスト向けのオープンソース JEV 互換実装。"
},
"decision": {
- "en": "Keeps shipped examples separate from unbuilt ideas and links each entry back to its original evidence.",
- "zh": "把已发布案例与未实现想法分开，并把每条记录链接回原始证据。",
- "ja": "公開済み事例と未実装アイデアを分け、各項目を元の根拠へリンクします。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/walidboulanouar/awesome-jev-use-cases/blob/80bd47333db768044cec10b1327c9ca3bb247994/README.md"
+ "en": "Accepts bounded question schemas and returns structured answers through a compatible interface.",
+ "zh": "接受有限问题 Schema，并通过兼容接口返回结构化答案。",
+ "ja": "限定質問 Schema を受け取り、互換インターフェースで構造化回答を返します。"
},
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "jev-chat",
- "name": "JEV Chat",
- "owner": "w3cj",
- "category": "interfaces",
- "repoUrl": "https://github.com/w3cj/jev-chat",
- "evidenceUrl": "https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts",
+ "repoUrl": "https://github.com/SiliconLabAI/OpenJev",
+ "evidenceUrl": "https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts"
- },
- "language": "TypeScript",
- "license": "MIT",
- "starsAtCapture": 84,
- "summary": {
- "en": "A tool-using chat interface where JEV chooses only from code-supplied reply and tool options.",
- "zh": "由 JEV 仅从代码提供的回复与工具选项中选择的工具型聊天界面。",
- "ja": "JEV がコード提供の返答・ツール候補だけから選ぶツール利用チャット。"
- },
- "decision": {
- "en": "Selects tools, arguments, and reply fragments from bounded pools.",
- "zh": "从有限候选池中选择工具、参数与回复片段。",
- "ja": "限定プールからツール、引数、返答断片を選びます。"
+ "url": "https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "chetaslua:jevmeter",
- "name": "Jevmeter",
- "owner": "ChetasLua",
- "category": "interfaces",
- "repoUrl": "https://github.com/ChetasLua/jevmeter",
- "evidenceUrl": "https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py",
+ "id": "formanator-jev",
+ "name": "Formanator",
+ "owner": "timrogers",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/timrogers/formanator",
+ "evidenceUrl": "https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py"
+ "url": "https://github.com/timrogers/formanator/blob/548740edef37a4169eef163e4988e64f43a560ec/src/typesafe.rs"
},
- "language": "Python",
+ "language": "Rust",
"license": "MIT",
- "starsAtCapture": 81,
+ "starsAtCapture": 99,
"summary": {
- "en": "Creates edited videos with score meters by asking Jev to rate transcript sentences against selected rubrics.",
- "zh": "把视频转成带评分仪表的视频成片：Jev 按选定规则给字幕句子评分，再由渲染器叠加显示。",
- "ja": "選択したルーブリックに対してトランスクリプト文を評価するよう JEV に依頼することで、スコア メーター付きの編集済みビデオを作成します。"
+ "en": "A CLI and MCP client for benefit claims with optional JEV receipt classification.",
+ "zh": "支持可选 JEV 收据分类的福利报销 CLI 与 MCP 客户端。",
+ "ja": "任意の JEV レシート分類を備えた福利厚生申請 CLI / MCP クライアント。"
},
"decision": {
- "en": "Scores transcript sentences with preset questions and scales for meters placed on the video timeline.",
- "zh": "逐句读取转录文本，按预设问题和尺度返回分数，供时间轴上的仪表使用。",
- "ja": "事前に設定された質問を含むトランスクリプト文をスコア化し、ビデオ タイムライン上に配置されたメーターをスケールします。"
+ "en": "Selects the benefit and receipt category, falling back when confidence is low.",
+ "zh": "选择福利与收据类别，置信度不足时回退。",
+ "ja": "福利厚生とレシートのカテゴリを選び、信頼度が低い場合はフォールバックします。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4646,7 +4674,7 @@
},
"language": "Jupyter Notebook",
"license": "MIT",
- "starsAtCapture": 81,
+ "starsAtCapture": 100,
"summary": {
"en": "A Neo4j graph navigator that uses JEV to classify the next relationship.",
"zh": "使用 JEV 判断下一条关系的 Neo4j 图导航器。",
@@ -4661,29 +4689,85 @@
"runtimeVerified": false
},
{
- "id": "giuliosmall:pg_typesafe",
- "name": "Pg TypeSafe",
- "owner": "giuliosmall",
- "category": "search-data",
- "repoUrl": "https://github.com/giuliosmall/pg_typesafe",
- "evidenceUrl": "https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c",
+ "id": "fastbrowse-jev",
+ "name": "fastbrowse · Jev",
+ "owner": "agent-labs-dev",
+ "category": "browser-computer-use",
+ "starsAtCapture": 97,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A browser agent that indexes page controls and lets Jev pick the next action while an LLM plans and reads.",
+ "zh": "浏览器 Agent：索引页面控件，由 Jev 选择下一步动作，LLM 负责规划与阅读。",
+ "ja": "ページ上の操作候補を索引化し、Jev が次の操作を選び、LLM が計画・読解するブラウザ Agent。"
+ },
+ "decision": {
+ "en": "Issues Choice and Noul questions over candidate DOM actions and verifies answer claims against quoted page text.",
+ "zh": "对候选 DOM 动作发出 Choice/Noul 问题，并用页面原文引用校验回答主张。",
+ "ja": "候補 DOM 操作に Choice/Noul を投げ、ページ原文の引用で回答主張を検証します。"
+ },
+ "repoUrl": "https://github.com/agent-labs-dev/fastbrowse",
+ "evidenceUrl": "https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c"
+ "url": "https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py"
},
- "language": "C",
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "savka777-jev-use",
+ "name": "JEV Use",
+ "owner": "savka777",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/savka777/jev-use",
+ "evidenceUrl": "https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/savka777/jev-use/blob/8907f85354addfa3d2b78f6a462087e4c73310b8/README.md"
+ },
+ "language": "Swift",
"license": "MIT",
- "starsAtCapture": 81,
+ "starsAtCapture": 92,
"summary": {
- "en": "A pre-alpha PostgreSQL C extension for calling Jev from SQL for classification, yes/no judgments, and scoring.",
- "zh": "一个预览阶段的 PostgreSQL C 扩展，让 SQL 直接调用 Jev 做分类、是非判断和评分。",
- "ja": "分類、はい/いいえの判断、スコアリングのために SQL から JEV を呼び出すためのプレアルファ版の PostgreSQL C 拡張機能。"
+ "en": "A macOS computer-use harness that drives Accessibility controls without a vision model.",
+ "zh": "无需视觉模型、通过无障碍控件操作 macOS 的 Computer Use 工具。",
+ "ja": "視覚モデルなしで Accessibility 要素を操作する macOS Computer Use ツール。"
},
"decision": {
- "en": "Converts SQL input into System One requests and returns answers through database functions.",
- "zh": "将 SQL 输入转换为 System One 请求，并把答案返回给数据库函数。",
- "ja": "SQL 入力を System One リクエストに変換し、データベース関数を通じて回答を返します。"
+ "en": "Chooses a visible accessibility element and the next allowed action.",
+ "zh": "选择可见无障碍元素与下一步允许的动作。",
+ "ja": "表示中の Accessibility 要素と次の許可済み操作を選びます。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "jev-chat",
+ "name": "JEV Chat",
+ "owner": "w3cj",
+ "category": "interfaces",
+ "repoUrl": "https://github.com/w3cj/jev-chat",
+ "evidenceUrl": "https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/w3cj/jev-chat/blob/e543aba8c21b57a28a748ef41966502130f0f69e/apps/server/src/jev/pools.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 91,
+ "summary": {
+ "en": "A tool-using chat interface where JEV chooses only from code-supplied reply and tool options.",
+ "zh": "由 JEV 仅从代码提供的回复与工具选项中选择的工具型聊天界面。",
+ "ja": "JEV がコード提供の返答・ツール候補だけから選ぶツール利用チャット。"
+ },
+ "decision": {
+ "en": "Selects tools, arguments, and reply fragments from bounded pools.",
+ "zh": "从有限候选池中选择工具、参数与回复片段。",
+ "ja": "限定プールからツール、引数、返答断片を選びます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4702,7 +4786,7 @@
},
"language": "JavaScript",
"license": "Unknown",
- "starsAtCapture": 81,
+ "starsAtCapture": 91,
"summary": {
"en": "YouTube extension detecting sponsored segments from live audio and transcripts using Jev, skipping promotional blocks automatically.",
"zh": "结合实时音频与字幕由 Jev 驱动的 YouTube 视频赞助广告片段检测与自动跳过扩展。",
@@ -4717,57 +4801,85 @@
"runtimeVerified": false
},
{
- "id": "jev-dsh-decision",
- "name": "JEV DSH Decision",
- "owner": "Devin-AXIS",
- "category": "agent-workflows",
- "starsAtCapture": 78,
- "language": "JavaScript",
- "license": "NOASSERTION",
+ "id": "pithings:advocaat",
+ "name": "Advocaat",
+ "owner": "pithings",
+ "category": "sdk-integrations",
+ "repoUrl": "https://github.com/pithings/advocaat",
+ "evidenceUrl": "https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/pithings/advocaat/blob/bc46287fc1102b95852a81d679c6e34a2c44f4a2/README.md"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 90,
"summary": {
- "en": "A structured JEV decision plugin for DeepSeek Harness and compatible coding-agent hosts.",
- "zh": "面向 DeepSeek Harness 及兼容编码 Agent Host 的结构化 JEV 决策插件。",
- "ja": "DeepSeek Harness と互換コーディング Agent ホスト向けの構造化 JEV 判断プラグイン。"
+ "en": "A small TypeScript client for asking Jev multiple typed questions about the same data.",
+ "zh": "用简短的 TypeScript 调用向 Jev 提问。把同一份数据里的多个判断一次写好，直接拿到概率、选项和分数。",
+ "ja": "同じデータについて複数の型付き質問を JEV に送るための、小さな TypeScript クライアント。"
},
"decision": {
- "en": "Packages bounded questions as a service and returns typed results to the host harness.",
- "zh": "把有限问题封装为服务，并向 Host Harness 返回类型化结果。",
- "ja": "限定質問をサービス化し、ホスト Harness へ型付き結果を返します。"
+ "en": "Combines yes/no, Choice and Score questions and maps responses to typed values.",
+ "zh": "将带类型的问题转成 System One 请求，并把命名答案映射回调用方。",
+ "ja": "Noul、Choice、Score の質問を組み合わせ、回答を型付きの値へ変換します。"
},
- "repoUrl": "https://github.com/Devin-AXIS/jev-dsh-decision",
- "evidenceUrl": "https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs",
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "rokcso:bluenoise",
+ "name": "Bluenoise",
+ "owner": "rokcso",
+ "category": "agent-workflows",
+ "repoUrl": "https://github.com/rokcso/bluenoise",
+ "evidenceUrl": "https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/Devin-AXIS/jev-dsh-decision/blob/adc88caa9bf174d367e79a5f254c7936bcef088e/service/jev.mjs"
+ "url": "https://github.com/rokcso/bluenoise/blob/ef81ea7a7c3677501d6de8f9235a4d6a866b573a/entrypoints/background.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 90,
+ "summary": {
+ "en": "An X/Twitter filtering extension using local rules by default, with optional Jev checks for unmatched replies.",
+ "zh": "为 X/Twitter 过滤帖子与回复的浏览器扩展；默认用本地规则，可选择用 Jev 检查未匹配的回复。",
+ "ja": "デフォルトでローカル ルールを使用する X/Twitter フィルタリング拡張機能。オプションで JEV が一致しない返信をチェックします。"
+ },
+ "decision": {
+ "en": "When experimental AI is enabled, Jev evaluates replies missed by rules and local thresholds decide whether to hide them.",
+ "zh": "启用实验性 AI 后，将未命中规则的回复交给 Jev 评估，再由本地阈值决定是否隐藏。",
+ "ja": "実験的な AI が有効になっている場合、JEV はルールによって見逃された応答を評価し、ローカルのしきい値によってそれらを非表示にするかどうかを決定します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "windtunnel-jev",
- "name": "WindTunnel · JEV Benchmarks",
- "owner": "nekuda-ai",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/nekuda-ai/WindTunnel",
- "evidenceUrl": "https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs",
- "language": "HTML",
- "license": "Apache-2.0",
- "starsAtCapture": 76,
+ "id": "jevbench",
+ "name": "JevBench",
+ "owner": "fstandhartinger",
+ "category": "developer-tools",
+ "repoUrl": "https://github.com/fstandhartinger/jevbench",
+ "evidenceUrl": "https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py",
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 83,
"summary": {
- "en": "A browser-agent benchmark harness with frozen JEV experiments across accessibility and WebMCP interfaces.",
- "zh": "浏览器 Agent 基准测试工具，包含基于无障碍树与 WebMCP 接口的冻结 JEV 实验。",
- "ja": "アクセシビリティと WebMCP を対象に固定条件の JEV 実験を行うブラウザー Agent ベンチマークです。"
+ "en": "A reproducible benchmark comparing JEV with open decision models, classifiers, and rerankers.",
+ "zh": "用于比较 JEV、开放决策模型、分类器与重排模型的可复现基准。",
+ "ja": "JEV、オープン判断モデル、分類器、リランカーを比較する再現可能なベンチマークです。"
},
"decision": {
- "en": "Combines JEV action selection with a bounded browser executor and records drift, cost, and failure evidence.",
- "zh": "把 JEV 动作选择与有限浏览器执行器结合，并记录漂移、成本与失败证据。",
- "ja": "JEV の操作選択を有限ブラウザー実行器と組み合わせ、ドリフト、費用、失敗の証拠を記録します。"
+ "en": "Uses a dedicated TypeSafe adapter to run the same bounded tasks and preserve model, latency, token, and cost evidence.",
+ "zh": "通过专用 TypeSafe Adapter 执行相同有限任务，并保留模型、延迟、Token 与成本证据。",
+ "ja": "専用 TypeSafe アダプターで同じ有限タスクを実行し、モデル、遅延、トークン、費用の記録を残します。"
},
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs"
+ "url": "https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4781,7 +4893,7 @@
"evidenceUrl": "https://github.com/can1357/jegrep/blob/a280f14f6da8163bde67e0c49f58b23517a02882/src/jev.rs",
"language": "Rust",
"license": "MIT",
- "starsAtCapture": 75,
+ "starsAtCapture": 83,
"summary": {
"en": "Semantic grep for live code trees: describe what you need, get files and original line ranges without building an index.",
"zh": "面向实时代码树的语义 grep：用自然语言描述目标，无需建索引即可返回文件与原始行号范围。",
@@ -4801,29 +4913,113 @@
"runtimeVerified": false
},
{
- "id": "grok-bot-jev",
- "name": "Grok Bot JEV",
- "owner": "Bodila51",
- "category": "routing-optimization",
- "repoUrl": "https://github.com/Bodila51/grok-bot-jev",
- "evidenceUrl": "https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md"
- },
- "language": "Python",
- "license": "MIT",
- "starsAtCapture": 74,
+ "id": "giuliosmall:pg_typesafe",
+ "name": "Pg TypeSafe",
+ "owner": "giuliosmall",
+ "category": "search-data",
+ "repoUrl": "https://github.com/giuliosmall/pg_typesafe",
+ "evidenceUrl": "https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/giuliosmall/pg_typesafe/blob/4b5bfc1df11b18c3f07bb10804eeb47e4508ec6a/typesafe.c"
+ },
+ "language": "C",
+ "license": "MIT",
+ "starsAtCapture": 82,
+ "summary": {
+ "en": "A pre-alpha PostgreSQL C extension for calling Jev from SQL for classification, yes/no judgments, and scoring.",
+ "zh": "一个预览阶段的 PostgreSQL C 扩展，让 SQL 直接调用 Jev 做分类、是非判断和评分。",
+ "ja": "分類、はい/いいえの判断、スコアリングのために SQL から JEV を呼び出すためのプレアルファ版の PostgreSQL C 拡張機能。"
+ },
+ "decision": {
+ "en": "Converts SQL input into System One requests and returns answers through database functions.",
+ "zh": "将 SQL 输入转换为 System One 请求，并把答案返回给数据库函数。",
+ "ja": "SQL 入力を System One リクエストに変換し、データベース関数を通じて回答を返します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "chetaslua:jevmeter",
+ "name": "Jevmeter",
+ "owner": "ChetasLua",
+ "category": "interfaces",
+ "repoUrl": "https://github.com/ChetasLua/jevmeter",
+ "evidenceUrl": "https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/ChetasLua/jevmeter/blob/cbf8e117b5b8835e3294c3a8ee652c7dfa737a9a/jevmeter/score.py"
+ },
+ "language": "Python",
+ "license": "MIT",
+ "starsAtCapture": 81,
+ "summary": {
+ "en": "Creates edited videos with score meters by asking Jev to rate transcript sentences against selected rubrics.",
+ "zh": "把视频转成带评分仪表的视频成片：Jev 按选定规则给字幕句子评分，再由渲染器叠加显示。",
+ "ja": "選択したルーブリックに対してトランスクリプト文を評価するよう JEV に依頼することで、スコア メーター付きの編集済みビデオを作成します。"
+ },
+ "decision": {
+ "en": "Scores transcript sentences with preset questions and scales for meters placed on the video timeline.",
+ "zh": "逐句读取转录文本，按预设问题和尺度返回分数，供时间轴上的仪表使用。",
+ "ja": "事前に設定された質問を含むトランスクリプト文をスコア化し、ビデオ タイムライン上に配置されたメーターをスケールします。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "prism-liquidity-agent",
+ "name": "Prism Liquidity Agent",
+ "owner": "irfndi",
+ "category": "domain-tools",
+ "repoUrl": "https://github.com/irfndi/prism-liquidity-agent",
+ "evidenceUrl": "https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts"
+ },
+ "language": "TypeScript",
+ "license": "MIT",
+ "starsAtCapture": 79,
+ "summary": {
+ "en": "An autonomous liquidity agent that uses JEV inside its rebalancing decision service.",
+ "zh": "在再平衡决策服务中使用 JEV 的自动化流动性 Agent。",
+ "ja": "リバランス判断サービスで JEV を使う自律流動性 Agent。"
+ },
+ "decision": {
+ "en": "Evaluates bounded liquidity actions from pool, risk, and position state.",
+ "zh": "根据池子、风险与持仓状态评估有限流动性动作。",
+ "ja": "プール、リスク、ポジション状態から限定された流動性操作を評価します。"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "windtunnel-jev",
+ "name": "WindTunnel · JEV Benchmarks",
+ "owner": "nekuda-ai",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/nekuda-ai/WindTunnel",
+ "evidenceUrl": "https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs",
+ "language": "HTML",
+ "license": "Apache-2.0",
+ "starsAtCapture": 79,
"summary": {
- "en": "A JEV decision layer for Grok Bot with usage gates and skill templates.",
- "zh": "为 Grok Bot 提供用量闸门与 Skill 模板的 JEV 决策层。",
- "ja": "使用量ゲートと Skill テンプレートを備えた Grok Bot 向け JEV 判断レイヤー。"
+ "en": "A browser-agent benchmark harness with frozen JEV experiments across accessibility and WebMCP interfaces.",
+ "zh": "浏览器 Agent 基准测试工具，包含基于无障碍树与 WebMCP 接口的冻结 JEV 实验。",
+ "ja": "アクセシビリティと WebMCP を対象に固定条件の JEV 実験を行うブラウザー Agent ベンチマークです。"
},
"decision": {
- "en": "Routes bounded bot decisions before more expensive generation runs.",
- "zh": "在更昂贵的生成前路由有限的 Bot 决策。",
- "ja": "高価な生成前に限定された Bot 判断をルーティングします。"
+ "en": "Combines JEV action selection with a bounded browser executor and records drift, cost, and failure evidence.",
+ "zh": "把 JEV 动作选择与有限浏览器执行器结合，并记录漂移、成本与失败证据。",
+ "ja": "JEV の操作選択を有限ブラウザー実行器と組み合わせ、ドリフト、費用、失敗の証拠を記録します。"
+ },
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/nekuda-ai/WindTunnel/blob/5ca8644e23826ebb30108e7bad240b61043bfe67/experiments/jev/run.mjs"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4842,7 +5038,7 @@
},
"language": "Python",
"license": "NOASSERTION",
- "starsAtCapture": 73,
+ "starsAtCapture": 78,
"summary": {
"en": "A curated JEV resource list with runnable typed-decision examples.",
"zh": "带可运行类型化决策示例的 JEV 精选资源列表。",
@@ -4857,57 +5053,57 @@
"runtimeVerified": false
},
{
- "id": "captaincore:captaincore",
- "name": "Captaincore",
- "owner": "CaptainCore",
- "category": "safety-review",
- "repoUrl": "https://github.com/CaptainCore/captaincore",
- "evidenceUrl": "https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go",
+ "id": "grok-bot-jev",
+ "name": "Grok Bot JEV",
+ "owner": "Bodila51",
+ "category": "routing-optimization",
+ "repoUrl": "https://github.com/Bodila51/grok-bot-jev",
+ "evidenceUrl": "https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go"
+ "url": "https://github.com/Bodila51/grok-bot-jev/blob/1583e09928c138aeac0aa89818c67ea41f08e807/README.md"
},
- "language": "Go",
+ "language": "Python",
"license": "MIT",
- "starsAtCapture": 71,
+ "starsAtCapture": 77,
"summary": {
- "en": "Jev commands in the WordPress toolkit CaptainCore answer structured questions and prioritize malware scanner findings for review.",
- "zh": "WordPress 运维工具 CaptainCore 的 Jev 命令，可询问结构化问题，并给恶意代码扫描结果排复核优先级。",
- "ja": "WordPress ツールキットCaptainCore の JEV コマンドは、構造化された質問に答え、マルウェア スキャナーの検出結果に優先順位を付けてレビューします。"
+ "en": "A JEV decision layer for Grok Bot with usage gates and skill templates.",
+ "zh": "为 Grok Bot 提供用量闸门与 Skill 模板的 JEV 决策层。",
+ "ja": "使用量ゲートと Skill テンプレートを備えた Grok Bot 向け JEV 判断レイヤー。"
},
"decision": {
- "en": "Sends scan rules, matched snippets, and file context to Jev to assess likely true positives and suggested handling.",
- "zh": "把扫描规则、命中片段与文件上下文交给 Jev，判断真阳性可能性和建议处理方式。",
- "ja": "スキャン ルール、一致したスニペット、およびファイル コンテキストを JEV に送信して、真陽性の可能性と推奨される処理を評価します。"
+ "en": "Routes bounded bot decisions before more expensive generation runs.",
+ "zh": "在更昂贵的生成前路由有限的 Bot 决策。",
+ "ja": "高価な生成前に限定された Bot 判断をルーティングします。"
},
"verification": "source-reviewed",
"runtimeVerified": false
},
{
- "id": "jevbench",
- "name": "JevBench",
- "owner": "fstandhartinger",
- "category": "developer-tools",
- "repoUrl": "https://github.com/fstandhartinger/jevbench",
- "evidenceUrl": "https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py",
- "language": "Python",
+ "id": "ying-kai-liao-jev-browser",
+ "name": "JEV Browser",
+ "owner": "Ying-Kai-Liao",
+ "category": "browser-computer-use",
+ "repoUrl": "https://github.com/Ying-Kai-Liao/jev-browser",
+ "evidenceUrl": "https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs"
+ },
+ "language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 71,
+ "starsAtCapture": 76,
"summary": {
- "en": "A reproducible benchmark comparing JEV with open decision models, classifiers, and rerankers.",
- "zh": "用于比较 JEV、开放决策模型、分类器与重排模型的可复现基准。",
- "ja": "JEV、オープン判断モデル、分類器、リランカーを比較する再現可能なベンチマークです。"
+ "en": "A browser automation library, CLI, and MCP server pairing LLM plans with JEV actions.",
+ "zh": "把 LLM 规划与 JEV 动作结合的浏览器自动化库、CLI 与 MCP Server。",
+ "ja": "LLM の計画と JEV の操作を組み合わせるブラウザー自動化ライブラリ、CLI、MCP サーバー。"
},
"decision": {
- "en": "Uses a dedicated TypeSafe adapter to run the same bounded tasks and preserve model, latency, token, and cost evidence.",
- "zh": "通过专用 TypeSafe Adapter 执行相同有限任务，并保留模型、延迟、Token 与成本证据。",
- "ja": "専用 TypeSafe アダプターで同じ有限タスクを実行し、モデル、遅延、トークン、費用の記録を残します。"
- },
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/fstandhartinger/jevbench/blob/75e6224ed8103bbc3485ca74820a2eaf7ce8abe0/jevbench/adapters/typesafe.py"
+ "en": "Chooses the next browser action and target from the current page observation.",
+ "zh": "根据当前页面观察选择下一步浏览器动作与目标。",
+ "ja": "現在のページ観測から次のブラウザー操作と対象を選びます。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4921,7 +5117,7 @@
"evidenceUrl": "https://github.com/mizchi/jev-lint/blob/c9846c8c9ee13a917f3af26294a4fdf1421d21b1/src/jev.ts",
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 70,
+ "starsAtCapture": 74,
"summary": {
"en": "A semantic code linter that batches repository findings through a zero-dependency JEV client.",
"zh": "语义代码 Linter，通过零依赖 JEV 客户端批量判断仓库问题。",
@@ -4941,29 +5137,29 @@
"runtimeVerified": false
},
{
- "id": "prism-liquidity-agent",
- "name": "Prism Liquidity Agent",
- "owner": "irfndi",
- "category": "domain-tools",
- "repoUrl": "https://github.com/irfndi/prism-liquidity-agent",
- "evidenceUrl": "https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts",
+ "id": "captaincore:captaincore",
+ "name": "Captaincore",
+ "owner": "CaptainCore",
+ "category": "safety-review",
+ "repoUrl": "https://github.com/CaptainCore/captaincore",
+ "evidenceUrl": "https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go",
"source": {
"type": "github",
"label": "GitHub source",
- "url": "https://github.com/irfndi/prism-liquidity-agent/blob/22c67bdbe30bab608226832256a5013ad826b707/engine/jev-service.ts"
+ "url": "https://github.com/CaptainCore/captaincore/blob/f54e0669350707b2f4f4c67a0130fdd6574bd0b5/cmd/typesafe.go"
},
- "language": "TypeScript",
+ "language": "Go",
"license": "MIT",
- "starsAtCapture": 69,
+ "starsAtCapture": 71,
"summary": {
- "en": "An autonomous liquidity agent that uses JEV inside its rebalancing decision service.",
- "zh": "在再平衡决策服务中使用 JEV 的自动化流动性 Agent。",
- "ja": "リバランス判断サービスで JEV を使う自律流動性 Agent。"
+ "en": "Jev commands in the WordPress toolkit CaptainCore answer structured questions and prioritize malware scanner findings for review.",
+ "zh": "WordPress 运维工具 CaptainCore 的 Jev 命令，可询问结构化问题，并给恶意代码扫描结果排复核优先级。",
+ "ja": "WordPress ツールキットCaptainCore の JEV コマンドは、構造化された質問に答え、マルウェア スキャナーの検出結果に優先順位を付けてレビューします。"
},
"decision": {
- "en": "Evaluates bounded liquidity actions from pool, risk, and position state.",
- "zh": "根据池子、风险与持仓状态评估有限流动性动作。",
- "ja": "プール、リスク、ポジション状態から限定された流動性操作を評価します。"
+ "en": "Sends scan rules, matched snippets, and file context to Jev to assess likely true positives and suggested handling.",
+ "zh": "把扫描规则、命中片段与文件上下文交给 Jev，判断真阳性可能性和建议处理方式。",
+ "ja": "スキャン ルール、一致したスニペット、およびファイル コンテキストを JEV に送信して、真陽性の可能性と推奨される処理を評価します。"
},
"verification": "source-reviewed",
"runtimeVerified": false
@@ -4982,7 +5178,7 @@
},
"language": "JavaScript",
"license": "MIT",
- "starsAtCapture": 68,
+ "starsAtCapture": 71,
"summary": {
"en": "An experimental Chrome extension that asks Jev whether candidate DOM elements are ads, then highlights or removes them.",
"zh": "一个实验性 Chrome 扩展，让 Jev 判断候选 DOM 元素是否是广告，再高亮或移除。",
@@ -4996,62 +5192,6 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
- {
- "id": "ying-kai-liao-jev-browser",
- "name": "JEV Browser",
- "owner": "Ying-Kai-Liao",
- "category": "browser-computer-use",
- "repoUrl": "https://github.com/Ying-Kai-Liao/jev-browser",
- "evidenceUrl": "https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/Ying-Kai-Liao/jev-browser/blob/578cff6e701a131733d03256078bb559a45ad188/src/jev.mjs"
- },
- "language": "JavaScript",
- "license": "MIT",
- "starsAtCapture": 67,
- "summary": {
- "en": "A browser automation library, CLI, and MCP server pairing LLM plans with JEV actions.",
- "zh": "把 LLM 规划与 JEV 动作结合的浏览器自动化库、CLI 与 MCP Server。",
- "ja": "LLM の計画と JEV の操作を組み合わせるブラウザー自動化ライブラリ、CLI、MCP サーバー。"
- },
- "decision": {
- "en": "Chooses the next browser action and target from the current page observation.",
- "zh": "根据当前页面观察选择下一步浏览器动作与目标。",
- "ja": "現在のページ観測から次のブラウザー操作と対象を選びます。"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
- {
- "id": "siliconlab-openjev",
- "name": "OpenJev",
- "owner": "SiliconLabAI",
- "category": "open-models",
- "starsAtCapture": 67,
- "language": "TypeScript",
- "license": "MIT",
- "summary": {
- "en": "An open-source JEV-compatible implementation for typed decision requests.",
- "zh": "面向类型化决策请求的开源 JEV 兼容实现。",
- "ja": "型付き判断リクエスト向けのオープンソース JEV 互換実装。"
- },
- "decision": {
- "en": "Accepts bounded question schemas and returns structured answers through a compatible interface.",
- "zh": "接受有限问题 Schema，并通过兼容接口返回结构化答案。",
- "ja": "限定質問 Schema を受け取り、互換インターフェースで構造化回答を返します。"
- },
- "repoUrl": "https://github.com/SiliconLabAI/OpenJev",
- "evidenceUrl": "https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md",
- "source": {
- "type": "github",
- "label": "GitHub source",
- "url": "https://github.com/SiliconLabAI/OpenJev/blob/a08e969c37b2e4a37f95b3426f983bd94303590c/README.md"
- },
- "verification": "source-reviewed",
- "runtimeVerified": false
- },
{
"id": "nidhi-agent-router",
"name": "Agent Router",
@@ -5066,7 +5206,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 63,
+ "starsAtCapture": 68,
"summary": {
"en": "A CLI that routes tasks to coding agents, models, and reasoning effort with JEV.",
"zh": "用 JEV 把任务路由到编码 Agent、模型与推理强度的 CLI。",
@@ -5094,7 +5234,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 62,
+ "starsAtCapture": 66,
"summary": {
"en": "A context cleaner that asks JEV which history to retain, truncate, or drop.",
"zh": "让 JEV 判断哪些历史应保留、截断或丢弃的上下文清理器。",
@@ -5117,7 +5257,7 @@
"evidenceUrl": "https://github.com/yzfly/awesome-jev-zh/blob/cdb8a78cb3ac4cec36ebe305b73a4e0b4f5cba21/README.md",
"language": "HTML",
"license": "CC0-1.0",
- "starsAtCapture": 59,
+ "starsAtCapture": 60,
"summary": {
"en": "A Chinese JEV ecosystem guide with curated projects, practical tutorials, pricing, and independent caveats.",
"zh": "中文 JEV 生态指南，包含精选项目、实践教程、价格信息与独立限制说明。",
@@ -5136,6 +5276,34 @@
"verification": "source-reviewed",
"runtimeVerified": false
},
+ {
+ "id": "jevmlx",
+ "name": "jevmlx",
+ "owner": "bnsd55",
+ "category": "open-models",
+ "starsAtCapture": 59,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A local Apple Silicon decision layer that scores constrained fields from MLX logits and speaks System One HTTP.",
+ "zh": "在 Apple Silicon 本地从 MLX logits 为受限字段打分，并提供 System One HTTP 接口。",
+ "ja": "Apple Silicon 上で MLX logits から制約付きフィールドを採点し、System One HTTP を提供するローカル層。"
+ },
+ "decision": {
+ "en": "Accepts POST /v1/systemone payloads, maps Noul/Choice/Score questions to one-pass schema scoring, and returns probabilities.",
+ "zh": "接受 POST /v1/systemone，把 Noul/Choice/Score 映射为单次 schema 打分并返回概率。",
+ "ja": "POST /v1/systemone を受け、Noul/Choice/Score を一回の schema 採点へ写して確率を返します。"
+ },
+ "repoUrl": "https://github.com/bnsd55/jevmlx",
+ "evidenceUrl": "https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
{
"id": "oxlint-plugin-jev",
"name": "Oxlint Plugin JEV",
@@ -5150,7 +5318,7 @@
},
"language": "TypeScript",
"license": "MIT",
- "starsAtCapture": 55,
+ "starsAtCapture": 58,
"summary": {
"en": "An Oxlint plugin that evaluates semantic lint rules with JEV.",
"zh": "用 JEV 评估语义 Lint 规则的 Oxlint 插件。",
@@ -5163,6 +5331,90 @@
},
"verification": "source-reviewed",
"runtimeVerified": false
+ },
+ {
+ "id": "open-alternative-jev",
+ "name": "Open Alternative to Jev",
+ "owner": "ikermoel",
+ "category": "open-models",
+ "starsAtCapture": 51,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "An open System One-style decider that packs typed questions into one forward pass over local open models.",
+ "zh": "开源 System One 风格决策器，把类型化问题打包进本地开放模型的一次前向。",
+ "ja": "型付き質問をローカル開放モデルの一回 forward に詰める、オープンな System One 風 Decider。"
+ },
+ "decision": {
+ "en": "Scores Choice options from label logits without generating text, with optional calibration helpers.",
+ "zh": "从标签 logits 为 Choice 选项打分且不生成文本，并提供可选校准辅助。",
+ "ja": "テキストを生成せずラベル logits から Choice を採点し、任意の較正ヘルパーも備えます。"
+ },
+ "repoUrl": "https://github.com/ikermoel/open-alternative-jev",
+ "evidenceUrl": "https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "canny-jev-hooks",
+ "name": "Canny · Jev Hooks",
+ "owner": "qkal",
+ "category": "safety-review",
+ "starsAtCapture": 50,
+ "language": "TypeScript",
+ "license": "MIT",
+ "summary": {
+ "en": "Claude Code and Codex hooks that keep an append-only ledger and optionally ask Jev Noul questions for advisory checks.",
+ "zh": "Claude Code / Codex Hooks：维护只追加账本，并可选向 Jev 询问 Noul 做建议性检查。",
+ "ja": "追記専用台帳を保ち、任意で Jev に Noul 助言チェックを頼む Claude Code / Codex hooks。"
+ },
+ "decision": {
+ "en": "Posts Noul questions to api.typesafe.ai/v1/systemone and only uses high-confidence answers to relax a stop gate.",
+ "zh": "向 api.typesafe.ai/v1/systemone 提交 Noul，仅在高置信时放宽停止门。",
+ "ja": "api.typesafe.ai/v1/systemone に Noul を送り、高信頼のときだけ停止ゲートを緩めます。"
+ },
+ "repoUrl": "https://github.com/qkal/Canny",
+ "evidenceUrl": "https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "ha-jev",
+ "name": "HA-Jev",
+ "owner": "AboveColin",
+ "category": "domain-tools",
+ "starsAtCapture": 50,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A Home Assistant integration that turns typed Jev questions about house state into sensors and automation actions.",
+ "zh": "Home Assistant 集成：把关于家庭状态的类型化 Jev 问题变成传感器与自动化动作。",
+ "ja": "家の状態への型付き Jev 質問をセンサーと自動化アクションにする Home Assistant 連携。"
+ },
+ "decision": {
+ "en": "Builds System One payloads via jevclient and updates entities from Noul, Choice, and Score answers under a token budget.",
+ "zh": "通过 jevclient 构建 System One 请求，并在额度内用 Noul/Choice/Score 更新实体。",
+ "ja": "jevclient で System One ペイロードを作り、トークン予算内で Noul/Choice/Score からエンティティを更新します。"
+ },
+ "repoUrl": "https://github.com/AboveColin/HA-Jev",
+ "evidenceUrl": "https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
}
]
}
diff --git a/scripts/curate-star50-expansion-2026-09-23.mjs b/scripts/curate-star50-expansion-2026-09-23.mjs
new file mode 100644
index 0000000..817fb7f
--- /dev/null
+++ b/scripts/curate-star50-expansion-2026-09-23.mjs
@@ -0,0 +1,273 @@
+import { readFile, writeFile } from 'node:fs/promises';
+
+const catalogueFile = new URL('../data/projects.json', import.meta.url);
+const catalogue = JSON.parse(await readFile(catalogueFile, 'utf8'));
+
+const additions = [
+ {
+ "id": "fx-typesafe-permission-reviewer",
+ "name": "fx · TypeSafe Permission Reviewer",
+ "owner": "vercel-labs",
+ "category": "safety-review",
+ "starsAtCapture": 3127,
+ "language": "Zig",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "A Zig coding agent with an optional TypeSafe Jev permission reviewer for pending tool actions.",
+ "zh": "Zig 编码 Agent，可选 TypeSafe Jev 权限审查器，用于审核待执行工具动作。",
+ "ja": "保留中のツール操作を審査する TypeSafe Jev 権限レビュアー付き Zig コーディング Agent。"
+ },
+ "decision": {
+ "en": "Posts pending actions to System One as Choice questions and maps allow/deny probabilities into the agent permission gate.",
+ "zh": "把待执行动作作为 Choice 提交到 System One，并把允许/拒绝概率映射进 Agent 权限门。",
+ "ja": "保留中の操作を Choice として System One に送り、許可/拒否確率を Agent の権限ゲートへ写します。"
+ },
+ "repoUrl": "https://github.com/vercel-labs/fx",
+ "evidenceUrl": "https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/vercel-labs/fx/blob/e094531ec3fe23b2a1498132f2cea04df680434b/src/builtins/gateway/typesafe_permission_reviewer.zig"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "agent-beacon-jev-evaluator",
+ "name": "Agent Beacon · Jev Evaluator",
+ "owner": "asymptote-labs",
+ "category": "agent-workflows",
+ "starsAtCapture": 1103,
+ "language": "Go",
+ "license": "MIT",
+ "summary": {
+ "en": "Cross-harness agent memory that can ask Jev to score bounded, redacted session projections before promoting lessons.",
+ "zh": "跨 Harness 的 Agent 记忆系统，可用 Jev 对脱敏会话投影打分后再提升经验。",
+ "ja": "複数 Harness 向け Agent メモリ。Jev で編集済みセッション投影を採点してから教訓を昇格できます。"
+ },
+ "decision": {
+ "en": "Sends Noul rubric questions to TypeSafe System One and stores calibrated probabilities with human review gates.",
+ "zh": "向 TypeSafe System One 发送 Noul 评分问题，并把校准概率与人工复核门一起保存。",
+ "ja": "TypeSafe System One に Noul 採点質問を送り、較正確率を人手レビュー付きで保存します。"
+ },
+ "repoUrl": "https://github.com/asymptote-labs/agent-beacon",
+ "evidenceUrl": "https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/asymptote-labs/agent-beacon/blob/29c504ea4f0163f1281cc6774a0b1d3d9d9436ee/cli/beacon/internal/learning/evaluator.go"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "quackd-jev-stepper",
+ "name": "quackd · Jev Stepper",
+ "owner": "rokbenko",
+ "category": "domain-tools",
+ "starsAtCapture": 229,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "A robot orchestration CLI with an optional System One stepper that chooses among permitted discrete robot calls.",
+ "zh": "机器人编排 CLI，可选 System One 步进器在已许可的离散机器人调用中做选择。",
+ "ja": "許可済みの離散ロボット呼び出しから選ぶ任意 System One ステッパー付きロボット制御 CLI。"
+ },
+ "decision": {
+ "en": "Routes Choice and Noul questions through the TypeSafe System One client while the executor keeps safety gates.",
+ "zh": "通过 TypeSafe System One 客户端路由 Choice 与 Noul，执行器仍保留安全门。",
+ "ja": "Choice と Noul を TypeSafe System One クライアント経由で送り、実行側が安全ゲートを維持します。"
+ },
+ "repoUrl": "https://github.com/rokbenko/quackd",
+ "evidenceUrl": "https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/rokbenko/quackd/blob/1250c05fd29fc388868366678917119d63146242/quackd/agent/decision/systemone.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "ai-python-jev-evaluate",
+ "name": "AI SDK Python · Jev Evaluate",
+ "owner": "vercel-labs",
+ "category": "sdk-integrations",
+ "starsAtCapture": 185,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "Vercel's AI SDK for Python with an experimental evaluate path that asks typed questions through typesafe-ai/jev.",
+ "zh": "Vercel Python AI SDK 的实验性 evaluate 路径，可通过 typesafe-ai/jev 提问类型化问题。",
+ "ja": "typesafe-ai/jev 経由で型付き質問を送る実験的 evaluate を備えた Vercel Python AI SDK。"
+ },
+ "decision": {
+ "en": "Batches Choice, Score, and Boolean questions against shared state via the AI Gateway evaluation protocol.",
+ "zh": "通过 AI Gateway 评估协议，对共享状态批量提交 Choice、Score 与布尔问题。",
+ "ja": "AI Gateway の評価プロトコルで、共有状態に対し Choice・Score・真偽質問をまとめて送ります。"
+ },
+ "repoUrl": "https://github.com/vercel-labs/ai-python",
+ "evidenceUrl": "https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/vercel-labs/ai-python/blob/7b406c2f6428e97d8366b090f3f37d6ac0d01337/src/ai/ops/evaluation.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "fastbrowse-jev",
+ "name": "fastbrowse · Jev",
+ "owner": "agent-labs-dev",
+ "category": "browser-computer-use",
+ "starsAtCapture": 97,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A browser agent that indexes page controls and lets Jev pick the next action while an LLM plans and reads.",
+ "zh": "浏览器 Agent：索引页面控件，由 Jev 选择下一步动作，LLM 负责规划与阅读。",
+ "ja": "ページ上の操作候補を索引化し、Jev が次の操作を選び、LLM が計画・読解するブラウザ Agent。"
+ },
+ "decision": {
+ "en": "Issues Choice and Noul questions over candidate DOM actions and verifies answer claims against quoted page text.",
+ "zh": "对候选 DOM 动作发出 Choice/Noul 问题，并用页面原文引用校验回答主张。",
+ "ja": "候補 DOM 操作に Choice/Noul を投げ、ページ原文の引用で回答主張を検証します。"
+ },
+ "repoUrl": "https://github.com/agent-labs-dev/fastbrowse",
+ "evidenceUrl": "https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/agent-labs-dev/fastbrowse/blob/a088eb8b6d83a9643531acd8f5c0c1b0b276953e/src/fastbrowse/jev.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "jevmlx",
+ "name": "jevmlx",
+ "owner": "bnsd55",
+ "category": "open-models",
+ "starsAtCapture": 59,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A local Apple Silicon decision layer that scores constrained fields from MLX logits and speaks System One HTTP.",
+ "zh": "在 Apple Silicon 本地从 MLX logits 为受限字段打分，并提供 System One HTTP 接口。",
+ "ja": "Apple Silicon 上で MLX logits から制約付きフィールドを採点し、System One HTTP を提供するローカル層。"
+ },
+ "decision": {
+ "en": "Accepts POST /v1/systemone payloads, maps Noul/Choice/Score questions to one-pass schema scoring, and returns probabilities.",
+ "zh": "接受 POST /v1/systemone，把 Noul/Choice/Score 映射为单次 schema 打分并返回概率。",
+ "ja": "POST /v1/systemone を受け、Noul/Choice/Score を一回の schema 採点へ写して確率を返します。"
+ },
+ "repoUrl": "https://github.com/bnsd55/jevmlx",
+ "evidenceUrl": "https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/bnsd55/jevmlx/blob/9d3621516826a5de4480c9fdab0aa6d59e93e113/jevmlx/serve.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "open-alternative-jev",
+ "name": "Open Alternative to Jev",
+ "owner": "ikermoel",
+ "category": "open-models",
+ "starsAtCapture": 51,
+ "language": "Python",
+ "license": "Apache-2.0",
+ "summary": {
+ "en": "An open System One-style decider that packs typed questions into one forward pass over local open models.",
+ "zh": "开源 System One 风格决策器，把类型化问题打包进本地开放模型的一次前向。",
+ "ja": "型付き質問をローカル開放モデルの一回 forward に詰める、オープンな System One 風 Decider。"
+ },
+ "decision": {
+ "en": "Scores Choice options from label logits without generating text, with optional calibration helpers.",
+ "zh": "从标签 logits 为 Choice 选项打分且不生成文本，并提供可选校准辅助。",
+ "ja": "テキストを生成せずラベル logits から Choice を採点し、任意の較正ヘルパーも備えます。"
+ },
+ "repoUrl": "https://github.com/ikermoel/open-alternative-jev",
+ "evidenceUrl": "https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/ikermoel/open-alternative-jev/blob/4a85df1831b537343c9e133b5150fa3a3b1ce98e/so1/decider.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "canny-jev-hooks",
+ "name": "Canny · Jev Hooks",
+ "owner": "qkal",
+ "category": "safety-review",
+ "starsAtCapture": 50,
+ "language": "TypeScript",
+ "license": "MIT",
+ "summary": {
+ "en": "Claude Code and Codex hooks that keep an append-only ledger and optionally ask Jev Noul questions for advisory checks.",
+ "zh": "Claude Code / Codex Hooks：维护只追加账本，并可选向 Jev 询问 Noul 做建议性检查。",
+ "ja": "追記専用台帳を保ち、任意で Jev に Noul 助言チェックを頼む Claude Code / Codex hooks。"
+ },
+ "decision": {
+ "en": "Posts Noul questions to api.typesafe.ai/v1/systemone and only uses high-confidence answers to relax a stop gate.",
+ "zh": "向 api.typesafe.ai/v1/systemone 提交 Noul，仅在高置信时放宽停止门。",
+ "ja": "api.typesafe.ai/v1/systemone に Noul を送り、高信頼のときだけ停止ゲートを緩めます。"
+ },
+ "repoUrl": "https://github.com/qkal/Canny",
+ "evidenceUrl": "https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/qkal/Canny/blob/f2c5e53779445d60dc4a09d2dbced2308fccb820/src/jev.ts"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ },
+ {
+ "id": "ha-jev",
+ "name": "HA-Jev",
+ "owner": "AboveColin",
+ "category": "domain-tools",
+ "starsAtCapture": 50,
+ "language": "Python",
+ "license": "MIT",
+ "summary": {
+ "en": "A Home Assistant integration that turns typed Jev questions about house state into sensors and automation actions.",
+ "zh": "Home Assistant 集成：把关于家庭状态的类型化 Jev 问题变成传感器与自动化动作。",
+ "ja": "家の状態への型付き Jev 質問をセンサーと自動化アクションにする Home Assistant 連携。"
+ },
+ "decision": {
+ "en": "Builds System One payloads via jevclient and updates entities from Noul, Choice, and Score answers under a token budget.",
+ "zh": "通过 jevclient 构建 System One 请求，并在额度内用 Noul/Choice/Score 更新实体。",
+ "ja": "jevclient で System One ペイロードを作り、トークン予算内で Noul/Choice/Score からエンティティを更新します。"
+ },
+ "repoUrl": "https://github.com/AboveColin/HA-Jev",
+ "evidenceUrl": "https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py",
+ "source": {
+ "type": "github",
+ "label": "GitHub source",
+ "url": "https://github.com/AboveColin/HA-Jev/blob/c3d4881121e4d02eacf272d9340447235c6ecc41/custom_components/jev/coordinator.py"
+ },
+ "verification": "source-reviewed",
+ "runtimeVerified": false
+ }
+];
+
+const byRepo = new Map(catalogue.projects.map((project) => [project.repoUrl.toLowerCase(), project]));
+for (const project of additions) byRepo.set(project.repoUrl.toLowerCase(), project);
+
+catalogue.projects = [...byRepo.values()].sort(
+ (a, b) => b.starsAtCapture - a.starsAtCapture || a.name.localeCompare(b.name),
+);
+catalogue.capturedAt = '2026-09-23';
+catalogue.source.batch = '2026-09-23-star50-cross-search-and-source-review';
+catalogue.source.policy =
+ 'Project-level deduplication. Public JEV-related GitHub repositories at or above 50 stars are included only after fixed-commit source review. Discovery includes GitHub repository search, competitor catalogues (AbdelStark/awesome-typesafe-jev, heyjunpenn/awesome-jev), and direct source inspection. Runtime behavior is not independently verified.';
+
+await writeFile(catalogueFile, `${JSON.stringify(catalogue, null, 2)}\n`);
+console.log(`Added or refreshed ${additions.length} projects; catalogue now has ${catalogue.projects.length}.`);
diff --git a/scripts/sync-readmes.mjs b/scripts/sync-readmes.mjs
index 405cab2..9efe56e 100644
--- a/scripts/sync-readmes.mjs
+++ b/scripts/sync-readmes.mjs
@@ -37,9 +37,9 @@ const configs = [
const { scenarios } = JSON.parse(await readFile(new URL('../data/scenarios.json', import.meta.url), 'utf8'));
const byId = new Map(projects.map(project => [project.id, project]));
const ui = {
- en: { quick: 'Find a project', scene: 'By use case', type: 'By project type', online: 'Search online', agent: 'Search with your agent', featured: 'Featured project gallery', all: 'All projects', stats: `${projects.length} projects · ${categoryCounts.size} project types · Updated ${catalogue.capturedAt}`, evidence: 'Source-reviewed; not independently run. Stars belong to the whole repository, not its JEV integration.', intro: 'Start with a task, browse a project type, or ask your agent to recommend useful implementations.', refs: 'Projects to learn from', learn: 'What to reference', flow: 'Where JEV fits', scope: 'These projects illustrate parts of this pattern, not a single ready-made application. Source review does not establish runtime behavior, performance, or BeatAPI compatibility.', free: 'Try the free JEV API', freeBody: 'For your own integration, use `jev-1.13-free` with `POST /v1/systemone`: input and output cost $0, even on a zero balance. Use the default auto key group; before your first top-up, the account limit is one successful request per minute. Third-party projects may need configuration or code changes.', contribute: 'Contribute', data: 'Catalogue data', back: 'Back to discovery' },
- zh: { quick: '快速查找', scene: '按使用场景', type: '按项目类型', online: '在线搜索', agent: '用 Agent 搜索', featured: '精选项目 Gallery', all: '全部项目', stats: `${projects.length} 个项目 · ${categoryCounts.size} 种项目类型 · 更新于 ${catalogue.capturedAt}`, evidence: '已核对源码，未独立运行验证。Star 属于整个仓库，不代表其中 JEV 集成的热度。', intro: '带着需求找场景，按类型浏览项目，或让 Agent 帮你选择参考实现。', refs: '可以参考的项目', learn: '值得参考什么', flow: 'JEV 在哪一步', scope: '这些项目分别展示该模式的一部分，并非一套开箱即用的完整应用。源码核对不代表运行、性能或 BeatAPI 接入兼容性已经验证。', free: '试用免费 JEV API', freeBody: '自行接入时，用 `jev-1.13-free` 调用 `POST /v1/systemone`：输入、输出均为 $0，零余额可用。Key 使用默认 auto 分组；首次充值前，账户每分钟可成功请求 1 次。第三方项目可能需要调整配置或代码。', contribute: '参与贡献', data: '目录数据', back: '返回检索入口' },
- ja: { quick: 'プロジェクトを探す', scene: '用途から', type: 'プロジェクトの種類から', online: 'サイトで検索', agent: 'Agent で検索', featured: '注目プロジェクト', all: '全プロジェクト', stats: `${projects.length} プロジェクト · ${categoryCounts.size} 種類 · 更新 ${catalogue.capturedAt}`, evidence: 'ソース確認済み、独立した実行検証は未実施。Star はリポジトリ全体の数であり、JEV 連携部分の評価ではありません。', intro: '用途や種類から探すか、Agent に参考実装を選んでもらえます。', refs: '参考プロジェクト', learn: '参考にする部分', flow: 'JEV が担当する段階', scope: '各プロジェクトはこのパターンの一部を示すもので、単一の完成アプリではありません。ソース確認は、動作・性能・BeatAPI 互換性の検証を意味しません。', free: '無料 JEV API を試す', freeBody: '自分で連携する場合は、`POST /v1/systemone` に `jev-1.13-free` を指定します。入力・出力ともに $0、残高 0 でも利用可能です。キーはデフォルトの auto グループを使います。初回チャージ前はアカウントごとに 1 分あたり成功 1 回まで。外部プロジェクトでは設定やコードの変更が必要な場合があります。', contribute: '貢献する', data: 'カタログデータ', back: '検索入口に戻る' },
+ en: { quick: 'Find a project', scene: 'By use case', type: 'By project type', online: 'Search online', agent: 'Search with your agent', featured: 'Featured project gallery', all: 'All projects', stats: `${projects.length} projects · ${categoryCounts.size} project types · Updated ${catalogue.capturedAt}`, evidence: 'Source-reviewed; not independently run. Stars belong to the whole repository, not its JEV integration.', intro: 'Start with a task, browse a project type, or ask your agent to recommend useful implementations.', refs: 'Projects to learn from', learn: 'What to reference', flow: 'Where JEV fits', scope: 'These projects illustrate parts of this pattern, not a single ready-made application. Source review does not establish runtime behavior, performance, or BeatAPI compatibility.', freePrompt: 'Want to use JEV in your own app?', freeCta: 'Get started with the free JEV API on BeatAPI →', free: 'Try the free JEV API', freeBody: 'For your own integration, use `jev-1.13-free` with `POST /v1/systemone`: input and output cost $0, even on a zero balance. Use the default auto key group; before your first top-up, the account limit is one successful request per minute. Third-party projects may need configuration or code changes.', contribute: 'Contribute', data: 'Catalogue data', back: 'Back to discovery' },
+ zh: { quick: '快速查找', scene: '按使用场景', type: '按项目类型', online: '在线搜索', agent: '用 Agent 搜索', featured: '精选项目 Gallery', all: '全部项目', stats: `${projects.length} 个项目 · ${categoryCounts.size} 种项目类型 · 更新于 ${catalogue.capturedAt}`, evidence: '已核对源码，未独立运行验证。Star 属于整个仓库，不代表其中 JEV 集成的热度。', intro: '带着需求找场景，按类型浏览项目，或让 Agent 帮你选择参考实现。', refs: '可以参考的项目', learn: '值得参考什么', flow: 'JEV 在哪一步', scope: '这些项目分别展示该模式的一部分，并非一套开箱即用的完整应用。源码核对不代表运行、性能或 BeatAPI 接入兼容性已经验证。', freePrompt: '想在自己的项目中接入 JEV？', freeCta: '在 BeatAPI 免费接入 JEV API →', free: '试用免费 JEV API', freeBody: '自行接入时，用 `jev-1.13-free` 调用 `POST /v1/systemone`：输入、输出均为 $0，零余额可用。Key 使用默认 auto 分组；首次充值前，账户每分钟可成功请求 1 次。第三方项目可能需要调整配置或代码。', contribute: '参与贡献', data: '目录数据', back: '返回检索入口' },
+ ja: { quick: 'プロジェクトを探す', scene: '用途から', type: 'プロジェクトの種類から', online: 'サイトで検索', agent: 'Agent で検索', featured: '注目プロジェクト', all: '全プロジェクト', stats: `${projects.length} プロジェクト · ${categoryCounts.size} 種類 · 更新 ${catalogue.capturedAt}`, evidence: 'ソース確認済み、独立した実行検証は未実施。Star はリポジトリ全体の数であり、JEV 連携部分の評価ではありません。', intro: '用途や種類から探すか、Agent に参考実装を選んでもらえます。', refs: '参考プロジェクト', learn: '参考にする部分', flow: 'JEV が担当する段階', scope: '各プロジェクトはこのパターンの一部を示すもので、単一の完成アプリではありません。ソース確認は、動作・性能・BeatAPI 互換性の検証を意味しません。', freePrompt: '自分のアプリに JEV を組み込みたいですか？', freeCta: 'BeatAPI で無料の JEV API を使い始める →', free: '無料 JEV API を試す', freeBody: '自分で連携する場合は、`POST /v1/systemone` に `jev-1.13-free` を指定します。入力・出力ともに $0、残高 0 でも利用可能です。キーはデフォルトの auto グループを使います。初回チャージ前はアカウントごとに 1 分あたり成功 1 回まで。外部プロジェクトでは設定やコードの変更が必要な場合があります。', contribute: '貢献する', data: 'カタログデータ', back: '検索入口に戻る' },
};
for (const config of configs) {
@@ -66,10 +66,18 @@ for (const config of configs) {
}[lang];
const readme = `
+
+
+
+
+
Awesome JEV
${config.strong}
+${t.freePrompt} ${t.freeCta}
+
${t.quick} · ${t.featured} · ${t.agent} · ${t.all} · ${t.online}
English · 简体中文 · 日本語
diff --git a/scripts/validate.test.mjs b/scripts/validate.test.mjs
index 765c1e8..e4dc348 100644
--- a/scripts/validate.test.mjs
+++ b/scripts/validate.test.mjs
@@ -9,7 +9,7 @@ const catalogue = JSON.parse(
test('catalogue metadata is explicit', () => {
assert.equal(catalogue.schemaVersion, 3);
assert.match(catalogue.capturedAt, /^\d{4}-\d{2}-\d{2}$/);
- assert.equal(catalogue.projects.length, 183);
+ assert.equal(catalogue.projects.length, 192);
const belowFloor = catalogue.projects.filter((project) => project.starsAtCapture < 50).map((project) => project.id);
assert.deepEqual(belowFloor, []);
assert.doesNotMatch(catalogue.source.policy, /exception/i);
@@ -42,7 +42,7 @@ test('every project keeps repository and fixed-commit evidence links', () => {
test('catalogue covers multiple practical patterns', () => {
assert.ok(new Set(catalogue.projects.map((project) => project.category)).size >= 9);
- assert.equal(catalogue.projects.filter((project) => project.starsAtCapture >= 1_000).length, 49);
+ assert.equal(catalogue.projects.filter((project) => project.starsAtCapture >= 1_000).length, 51);
for (const required of [
'jev-ultrafast',
'laya',