Anthropic-Cybersecurity-Skills

Anthropic Cybersecurity Skills La plus grande bibliothèque open-source de compétences en cybersécurité pour agents IA !Hermes Agent...

defending-code-reference-harness

Эталонный стенд для защиты кода Эталонная реализация для автономного обнаружения и устранения уязвимостей с помощью Claude, основанная на нашем опыте партнёрства с командами безопасности в нескольких организациях с момента запуска Claude Mythos Preview. Подробный разбор этих выводов и лучших...

Anthropic-Credited-CVEs

ANTHROPIC CVE TRACKER Overview Tracking vulnerabilities that credit the Anthropic research team and are possibly discovered by Project Glasswing. CURRENT CVE COUNT: 301 Fixed Anthropic Findings w/o CVE: 128 Findings Withdrawn by Anthropic: 243 Distributions | ---|--- Anthropic Research Tracking...

synaptic-sentinel

SYNAPTIC Sentinel مراقب أمان للبرمجة بالإحساس vibe-coding. مجموعة أدوات أمان وكيلية agentic بترخيص Apache-2.0 للمشاريع التي يكتب معظم كودها بواسطة الذكاء الاصطناعي. خمسة كشافة حتمية deterministic scouts تلتقط المشاكل النحوية؛ طبقة الدماغ Brain Layer BYOK Anthropic تقرر ما هو مهم حقًا وكيفية إصلاح...

vLLM <= 0.23.0 - Anthropic Router Heap Address Information Leak

vLLM = 0.23.0 incompletely fixes CVE-2026-22778. The original fix added sanitizemessage to the OpenAI router but the Anthropic-compatible router /v1/messages echoes strexc directly. id: CVE-2026-54236 info: name: vLLM = 0.23.0 - Anthropic Router Heap Address Information Leak author: kenlacroix...

MCP Inspector < 0.14.0 UnauthenticatedRemote Code Execution

The MCP inspector is a developer tool for testing and debugging MCP servers. Versions of MCP Inspector below 0.14.1 are vulnerable to remote code execution due to lack of authentication between the Inspector client and proxy, allowing unauthenticated requests to launch MCP commands over stdio. id...

CVE-2026-101998

Docker Sandboxes could fail open while masking credentials in protected proxy responses. When a response-body read returned data together with an error, affected handlers could forward unmasked bytes. Code inside an authorized sandbox could use this to recover host-managed OAuth access and refres...

CVE-2026-101998

Docker Sandboxes are vulnerable to a fail-open response masking issue where certain handlers may forward unmasked bytes when a response-body read returns data alongside an error. This vulnerability allows code running inside an authorized sandbox to potentially recover sensitive host-managed OAut...

CVE-2026-101998 Fail-open response masking in Docker Sandboxes can expose proxy-managed credentials

Docker Sandboxes could fail open while masking credentials in protected proxy responses. When a response-body read returned data together with an error, affected handlers could forward unmasked bytes. Code inside an authorized sandbox could use this to recover host-managed OAuth access and refres...

EUVD-2026-95157

Docker Sandboxes could fail open while masking credentials in protected proxy responses. When a response-body read returned data together with an error, affected handlers could forward unmasked bytes. Code inside an authorized sandbox could use this to recover host-managed OAuth access and refres...

CVE-2026-21852-PoC

Démo Éducative sur les Vulnérabilités de Claude Code Ce projet illustre trois vulnérabilités divulguées par Check Point Research dans l'outil CLI Claude Code d'Anthropic. Il est conçu pour les étudiants et chercheurs en cybersécurité afin de comprendre comment fonctionnent les attaques de la chaî...

PT-2026-108131

Name of the Vulnerable Software and Affected Versions Docker Sandboxes affected versions not specified Description Docker Sandboxes may fail open when masking credentials in protected proxy responses. This occurs when a response-body read returns data simultaneously with an error, causing affecte...

The Pentagon Hopes to Speed Up ‘Kill Chain’ AI Buys With 5-Minute Videos

The US government’s Tradewinds initiative has made it easier to throw millions of dollars at “nontraditional” defense contractors, including OpenAI, Anthropic, and Google...

Anthropic Expands Claude Access for Vetted Cyber Teams as Glasswing Finds 129,000 Flaws

Anthropic on Tuesday said it's expanding a program that allows vetted cybersecurity professionals to test its advanced artificial intelligence AI models with reduced safeguards and blocking classifiers, as the company claimed its Project Glasswing initiative uncovered at least 129,000 verified...

cve-2025-65099

Доказательство концепции для CVE-2025-65099 Более подробная информация приведена в опубликованной статье. Техника yarnPath Уязвимое окружение для yarnPath можно инициализировать следующим образом: docker run --rm -v "$PWD":/poc -it --workdir /poc ubuntu bash -c 'apt-get update -qq &&...

PT-2026-107521

@AnthropicAI I just submitted my application and added some of my CVE's that are public: CVE ID or IDs CVE-2026-49257, CVE-2026-76084, CVE-2026-76817, CVE-2026-63192, CVE-2026-63194, CVE-2026-54657, CVE-2026-54683, CVE-2026-55414 . Fingers crossed...

python313-langchain-anthropic-1.7.5-1.1 on GA media (moderate)

python313-langchain-anthropic-1.7.5-1.1 on GA media Announcement ID: openSUSE-SU-2026:11946-1 Rating: moderate Cross-References: CVE-2026-55443 Affected Products: openSUSE Tumbleweed An update that solves one vulnerability can now be installed. Description: These are all security issues fixed in...

CVE-2025-54794-Hijacking-Claude-AI-with-a-Prompt-Injection-The-Jailbreak-That-Talked-Back

🧠 CVE-2025-54794 : Détournement de Claude AI par injection de prompt – Le jailbreak qui a répondu Par Aditya Bhatt | Spécialiste en sécurité offensive | Opérateur Red Team | Accro au VAPT ⚔️ Introduction : Quand votre IA peut être piratée avec des mots À une époque où les modèles de langage sont...

Malicious code in anthropic-sdk (PyPI)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 500869e36b3f76202b63e8db9087adcfc42c8281a27a4402a21285aabde7a511 The package publishes as anthropic-sdk and re-exports the official anthropic client's symbols from anthropic import ; re-exports of...

MAL-2026-17472 Malicious code in anthropic-sdk (PyPI)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 500869e36b3f76202b63e8db9087adcfc42c8281a27a4402a21285aabde7a511 The package publishes as anthropic-sdk and re-exports the official anthropic client's symbols from anthropic import ; re-exports of...