skillspector

أداة تسهل عليك فهم أي مادة/موضوع للتثبيت: " I need you to go to this repo: MstrPxl/lesson-deck-builder And install it here as a skill + any required file(s) for the skill to function properly. Provide me with the skill card for installation. " تقدر تستعمل هذي المهارة من انفيديا عشان تتأكد من أمان أي مهارة تحملها.

أداة تسهل عليك فهم أي مادة/موضوع للتثبيت: " I need you to go to this repo: MstrPxl/lesson-deck-builder And install it here as a skill + any required file(s) for the skill to function properly. Provide me with the skill card for installation. " تقدر تستعمل هذي المهارة من انفيديا عشان تتأكد من أمان أي مهارة تحملها.

SkillSpector: Security scanner for AI agent skills (22k stars) What it does All text below is quoted directly from the project README. >Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, and security risks before installing agent skills. Features: - Multi-format input: Scan Git repos, URLs, zip files, directories, or single files - 71 vulnerability patterns across 17 categories: prompt injection, data exfiltration, privilege escalation, supply chain, excessive agency, output handling, system prompt leakage, memory poisoning, tool misuse, rogue agent, anti-refusal, trigger abuse, dangerous code (AST), taint tracking, YARA signatures, MCP least privilege, and MCP tool poisoning - Two-stage analysis: Fast static analysis + optional LLM semantic evaluation - Live vulnerability lookups: SC4 queries OSV.dev for real-time CVE data with automatic offline fallback - Multiple output formats: Terminal, JSON, Markdown, and SARIF reports - Risk scoring: 0-100 score with severity labels and clear recommendations - Baseline / false-positive suppression: Accept known findings via a glob-rule or fingerprint baseline so re-scans surface only new issues >uv tool install git+NVIDIA/skillspector Repository NVIDIA/skillspector Repository stats |Metric|Value| |:-|:-| |Stars|22,008| |Forks|1.9k| |Last release|v2.13.1 (2026-10-23)| |Languages|Python 98.6%, TypeScript 0.8%, JavaScript 0.3%| |Contributors|100| |AI contributor accounts (Claude / Codex / Cursor / Copilot)|@claude (91 commits), @cursoragent (2 commits), @Copilot (1 commits)| |AI co-authored commits|Claude: 104, Cursor: 2| Main contributor |Field|Value| |:-|:-| |GitHub|@yashrajp22| |Commits to this repo|127 (top contributor)| |Followers|5| ────────────────────────────── 📌 How many great GitHub repos have you discovered and completely forgotten about? Don't let another useful tool disappear into your Reddit feed. Get the community's best repos, organized by category, plus key takeaways from real developer discussions, delivered to your inbox twice a week. 📬 Never lose track of great repos again. Subscribe free.

أداة لفحص ملفات Skills المستخدمة مع وكلاء AI للبرمجة وتقوم بالكشف عن الثغرات و malicious patterns ، المخاطر الأمنية ، تسريب البيانات ، prompt injection وغيرها NVIDIA/SkillSpector #الأمن_السيبراني #AI

NVIDIA's SkillSpector Scans AI Skills Before You Install Out of roughly 31,000 AI skills tested, about 1 in every 20 appeared to be harmful or dangerous. SkillSpector is an open source scanner from NVIDIA that checks a skill before you install it. It looks for 71 risky patterns, like sending your data to outside websites, then gives a score out of 100 and says whether to install. You install it with one command, and you can connect it to Claude Code so Claude can scan for you. It runs basic checks on its own, and an AI model adds more. Repo: NVIDIA/SkillSpector Voiceover is AI-generated (ElevenLabs). Visuals made in Remotion. NVIDIA, Claude, Claude Code and GitHub are trademarks of their owners, named here only to describe the products. This channel is not affiliated with them. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [de-DE] NVIDIAs SkillSpector scannt KI-Skills vor der Installation Von rund 31.000 getesteten KI-Fähigkeiten erwies sich etwa jede 20. als schädlich oder gefährlich. SkillSpector ist ein Open-Source-Scanner von NVIDIA, der Fähigkeiten vor der Installation überprüft. Er sucht nach 71 Risikomustern, wie z. B. dem Senden von Daten an externe Websites, vergibt eine Punktzahl von 100 und gibt eine Installationsempfehlung. Die Installation erfolgt mit einem einzigen Befehl. SkillSpector kann mit Claude Code verbunden werden, sodass Claude automatisch Scans durchführt. Er führt selbstständig grundlegende Prüfungen durch, die durch ein KI-Modell ergänzt werden. Repository: NVIDIA/SkillSpector Die Sprachausgabe wurde von einer KI generiert (ElevenLabs). Die Visualisierungen wurden mit Remotion erstellt. NVIDIA, Claude, Claude Code und GitHub sind Marken ihrer jeweiligen Inhaber und werden hier nur zur Produktbeschreibung genannt. Dieser Kanal steht in keiner Verbindung zu diesen Unternehmen. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [es-US] SkillSpector de NVIDIA analiza las habilidades de IA antes de que las instales De aproximadamente 31.000 habilidades de IA analizadas, alrededor de 1 de cada 20 resultó ser dañina o peligrosa. SkillSpector es un escáner de código abierto de NVIDIA que verifica una habilidad antes de instalarla. Busca 71 patrones de riesgo, como enviar tus datos a sitios web externos, asigna una puntuación de 0 a 100 e indica si se debe instalar. Se instala con un solo comando y se puede conectar a Claude Code para que este realice el análisis. Realiza comprobaciones básicas automáticamente, y un modelo de IA añade más. Repositorio: NVIDIA/SkillSpector La voz en off es generada por IA (ElevenLabs). Los efectos visuales se crearon con Remotion. NVIDIA, Claude, Claude Code y GitHub son marcas comerciales de sus respectivos propietarios; se mencionan aquí únicamente para describir los productos. Este canal no está afiliado a ellos. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [fr-FR] SkillSpector de NVIDIA scanne les compétences IA avant installation Sur environ 31 000 compétences d'IA testées, près d'une sur vingt s'est révélée nuisible ou dangereuse. SkillSpector est un scanner open source de NVIDIA qui vérifie une compétence avant son installation. Il recherche 71 comportements à risque, comme l'envoi de vos données à des sites web externes, puis attribue un score sur 100 et indique si l'installation est recommandée. Son installation se fait en une seule commande, et vous pouvez le connecter à Claude Code pour que ce dernier effectue des analyses. SkillSpector effectue des vérifications de base automatiquement, et un modèle d'IA en complète les analyses. Dépôt : NVIDIA/SkillSpector Voix off générée par IA (ElevenLabs). Visuels réalisés avec Remotion. NVIDIA, Claude, Claude Code et GitHub sont des marques déposées de leurs propriétaires respectifs, citées ici uniquement à titre descriptif. Cette chaîne n'est affiliée à aucune d'entre elles. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [hi] NVIDIA का SkillSpector इंस्टॉल करने से पहले AI स्किल्स को स्कैन करता है लगभग 31,000 AI स्किल्स के परीक्षण में से, लगभग हर 20 में से 1 स्किल हानिकारक या खतरनाक पाई गई। SkillSpector, NVIDIA का एक ओपन सोर्स स्कैनर है जो किसी स्किल को इंस्टॉल करने से पहले उसकी जांच करता है। यह 71 जोखिम भरे पैटर्न की पहचान करता है, जैसे कि आपका डेटा बाहरी वेबसाइटों पर भेजना, फिर 100 में से स्कोर देता है और बताता है कि इसे इंस्टॉल करना है या नहीं। इसे एक कमांड से इंस्टॉल किया जा सकता है, और आप इसे Claude Code से कनेक्ट कर सकते हैं ताकि Claude आपके लिए स्कैन कर सके। यह बुनियादी जांच अपने आप करता है, और एक AI मॉडल इसमें अतिरिक्त जांच जोड़ता है। रिपॉजिटरी: NVIDIA/SkillSpector वॉइसओवर AI द्वारा जनरेट किया गया है (ElevenLabs)। विजुअल्स Remotion में बनाए गए हैं। NVIDIA, Claude, Claude Code और GitHub उनके स्वामियों के ट्रेडमार्क हैं, जिनका उल्लेख यहां केवल उत्पादों का वर्णन करने के लिए किया गया है। यह चैनल उनसे संबद्ध नहीं है। #शॉर्ट्स #क्लाउडकोड #एआईसिक्योरिटी #ओपनसोर्स #एनवीडिया #एजेंटस्किल्स [id] SkillSpector dari NVIDIA Memindai Skill AI Sebelum Anda Menginstalnya Dari sekitar 31.000 keterampilan AI yang diuji, sekitar 1 dari 20 tampak berbahaya atau berisiko. SkillSpector adalah pemindai sumber terbuka dari NVIDIA yang memeriksa keterampilan sebelum Anda menginstalnya. Ia mencari 71 pola berisiko, seperti mengirim data Anda ke situs web eksternal, kemudian memberikan skor dari 100 dan mengatakan apakah perlu diinstal. Anda menginstalnya dengan satu perintah, dan Anda dapat menghubungkannya ke Claude Code sehingga Claude dapat memindai untuk Anda. Ia menjalankan pemeriksaan dasar sendiri, dan model AI menambahkan lebih banyak pemeriksaan. Repositori: NVIDIA/SkillSpector Narasi dihasilkan oleh AI (ElevenLabs). Visual dibuat di Remotion. NVIDIA, Claude, Claude Code, dan GitHub adalah merek dagang dari pemiliknya, yang disebutkan di sini hanya untuk menggambarkan produk. Saluran ini tidak berafiliasi dengan mereka. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [it] SkillSpector di NVIDIA analizza le competenze AI prima dell'installazione Su circa 31.000 skill basate sull'IA testate, circa 1 su 20 è risultata dannosa o pericolosa. SkillSpector è uno scanner open source di NVIDIA che controlla una skill prima della sua installazione. Ricerca 71 modelli di rischio, come l'invio dei dati a siti web esterni, assegna un punteggio da 0 a 100 e indica se è possibile installarla. L'installazione avviene con un singolo comando ed è possibile collegarlo a Claude Code in modo che Claude esegua le scansioni per conto dell'utente. Esegue autonomamente controlli di base, a cui si aggiunge un modello di IA. Repository: NVIDIA/SkillSpector La voce fuori campo è generata dall'IA (ElevenLabs). Le immagini sono state realizzate con Remotion. NVIDIA, Claude, Claude Code e GitHub sono marchi registrati dei rispettivi proprietari, citati qui solo a scopo descrittivo. Questo canale non è affiliato con loro. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills [pt-BR] SkillSpector da NVIDIA analisa habilidades de IA antes da instalação De aproximadamente 31.000 habilidades de IA testadas, cerca de 1 em cada 20 pareceu ser prejudicial ou perigosa. O SkillSpector é um scanner de código aberto da NVIDIA que verifica uma habilidade antes da instalação. Ele procura por 71 padrões de risco, como o envio de seus dados para sites externos, e então atribui uma pontuação de 0 a 100, indicando se a instalação é recomendada. A instalação é feita com um único comando, e você pode conectá-lo ao Claude Code para que o Claude realize as verificações. Ele executa verificações básicas automaticamente, e um modelo de IA adiciona outras. Repositório: NVIDIA/SkillSpector A narração é gerada por IA (ElevenLabs). Os visuais foram criados no Remotion. NVIDIA, Claude, Claude Code e GitHub são marcas registradas de seus respectivos proprietários, mencionadas aqui apenas para descrever os produtos. Este canal não possui qualquer vínculo com eles. #Shorts #ClaudeCode #AISecurity #OpenSource #NVIDIA #AgentSkills

папку: проверит 7️⃣1️⃣ шаблон уязвимостей и даст оценку риска от 0 до 100. NVIDIA/skillspector

GitHub Trending Oct 27: NVIDIA SkillSpector, Postiz (free social scheduler), Microduck robot For your AI agent: gitnova.dev (projects: NVIDIA/SkillSpector, gitroomhq/postiz-app, pollen-robotics/microduck) Three standout tremors on GitHub today, measured by GitNova (Oct 27, 2026, 10:20 UTC). 1. NVIDIA/SkillSpector: magnitude 7.7, +950 stars in 24 hours. An open-source scanner (Apache-2.0) that checks AI agent skills for Claude Code, Codex and MCP before you install them: prompt injection, data exfiltration, supply-chain risks; 71 patterns, risk score 0–100. In the 31,132-skill subset of its research data, 26.1% had vulnerabilities and 5.2% looked malicious, per the README. NVIDIA/SkillSpector 2. gitroomhq/postiz-app: magnitude 7.6, +939 stars in 24 hours. A social media scheduler with calendar, analytics and AI agent support. The self-hosted edition is free (AGPL-3.0) with unlimited channels, posts and team members; you create your own developer app on each platform, and Meta, YouTube or TikTok approval can take weeks. gitroomhq/postiz-app 3. pollen-robotics/microduck: magnitude 5.5, +613 stars in 24 hours. The software brain (Rust, Apache-2.0) of a 25 cm, 800 g robot duck: a 50 Hz control loop runs neural policies across 15 servos. It walks, rolls on wheels, picks things up and gets back up. The robot itself is sold by Pollen Robotics. pollen-robotics/microduck Magnitude 0–10 measures growth of interest, not quality or safety. Check a project before you run it. All tremors and project search: gitnova.dev 0:00 NVIDIA SkillSpector 0:29 Postiz 0:56 Microduck Narration is an AI voice (ElevenLabs). #GitHub #OpenSource #GitHubTrending [ru] Взлёты дня на GitHub: SkillSpector от NVIDIA, Postiz и утка-робот Microduck Ролик на английском — русские субтитры включаются в настройках видео (⚙ → Субтитры). Три заметных толчка на GitHub за сутки — по замерам GitNova (27 октября 2026, 13:20 МСК). 1. NVIDIA/SkillSpector — магнитуда 7,7, +950 звёзд за 24 часа. Открытый сканер (Apache-2.0), который проверяет навыки AI-агентов для Claude Code, Codex и MCP до установки: prompt injection, утечка данных, риски цепочки поставок; 71 шаблон уязвимостей, оценка риска 0–100. В выборке из 31 132 навыков из его исследовательских данных у 26,1% нашлись уязвимости, 5,2% выглядели вредоносными — по README. NVIDIA/SkillSpector 2. gitroomhq/postiz-app — магнитуда 7,6, +939 звёзд за 24 часа. Планировщик постов в соцсетях с календарём, аналитикой и поддержкой AI-агентов. Версия для своего сервера бесплатна (AGPL-3.0), без лимитов на каналы, посты и участников; приложение разработчика на каждой платформе создаёте сами, одобрение Meta, YouTube или TikTok может занять недели. gitroomhq/postiz-app 3. pollen-robotics/microduck — магнитуда 5,5, +613 звёзд за 24 часа. Программный «мозг» (Rust, Apache-2.0) утки-робота ростом 25 см и весом 800 г: цикл управления 50 Гц гоняет нейросетевые политики по 15 сервоприводам. Ходит, катается на колёсах, поднимает предметы и встаёт сама. Самого робота продаёт Pollen Robotics. pollen-robotics/microduck Магнитуда 0–10 — рост интереса, а не оценка качества или безопасности. Проверяйте проект перед запуском. Все толчки и поиск по задаче: gitnova.dev Telegram: t.me 0:00 NVIDIA SkillSpector 0:29 Postiz 0:56 Microduck Озвучка — синтезированный голос (ElevenLabs), на английском. #GitHub #OpenSource #GitHubTrending

Before you install that shiny new claude skill, maybe check what's inside it first. Skillspector scans agent skills for prompt injection, data exfiltration and other nasty stuff before they get anywhere near your machine. NVIDIA/SkillSpector

五個 GitHub 專案全是 AI 程式相關工具，今天漲星前 5 名沒例外 今天 GitHub 熱榜五強齊發！榜首 ECC 以單日 896 星奪冠，帶來 AI 代理工程系統；Agent-Reach 幫助代理突破網路牆；Karpathy 規範檔讓編碼更精準；huashu-design 一句話出設計稿；SkillSpector 則是 NVIDIA 出品的技能安全掃描器。 1｜ECC：AI 代理工程系統 affaan-m/ECC 2｜Agent-Reach：AI 代理爬蚠工具 Panniantong/Agent-Reach 3｜andrej-karpathy-skills：Claude Code 規範檔 multica-ai/andrej-karpathy-skills 4｜huashu-design：AI 設計技能套件 alchaincyf/huashu-design 5｜SkillSpector：AI 技能安全掃描器 NVIDIA/SkillSpector 資料來源：GitHub Trending（今日新增 star） 每天幫你篩一遍 GitHub 開源熱榜。 排名依 GitHub Trending 當日新增 star 數，僅供參考，不代表推薦或背書。 #GitHub #開源 #程式設計 #開源專案 #GitHubTrending #軟體開發 #ECC #JavaScript #Python #AI #ClaudeCode #設計 #Shorts

103K-Star Eyes for Your AI Agent | GitHub Shorts Trending now: 1. Panniantong/Agent-Reach (103,256 stars) — Give your AI agent eyes to see the entire internet. Read & search Twitter, Reddit, YouTub…: Panniantong/Agent-Reach 2. pbakaus/impeccable (88,867 stars) — The design language that makes your AI harness better at design.: pbakaus/impeccable 3. NVIDIA/SkillSpector (21,962 stars) — Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, securit…: NVIDIA/SkillSpector 4. pollen-robotics/microduck (12,284 stars) — A Tiny biped duck robot 🦆: pollen-robotics/microduck Follow SmartStart for daily GitHub trending. #Shorts #GitHub #OpenSource #SmartStart