SHADOWFIREWALL
معماري دفاع تكيفي
الملف المهني
ينشئ ShadowFirewall جدران حماية سلوكية قائمة على نماذج الأنماط العدائية. بخلفية في الذكاء الاصطناعي والأنظمة المضمنة، يصمم دفاعات تفاعلية تتعلم من حركة المرور.
المهارات الأساسية
جدران حماية تكيفية؛ نمذجة السلوك العدائي؛ كشف الشذوذ؛ التقسيم الدقيق؛ أمن OT/SCADA
أبرز الإنجازات
صمم أول جدار حماية سلوكي لمصنع ذكي أوروبي.
مقالات بقلم SHADOWFIREWALL
When an AI Action Can Read the Runner, Secrets Stop Being Secret
A GitHub Actions warning shows how a file-reading tool inside an agentic workflow can become a quiet path to CI/CD environment data.
When a Package Becomes a Proxy: The Claude Code MCP Token Trap
A malicious npm package was used in a demonstrated attack path that rerouted Claude Code integrations and put OAuth bearer tokens in the crosshairs.
The Seams Are the Battlefield: Multi-Cloud’s Hidden Control Problem
The real risk in multi-cloud is not having too many tools, but failing to govern routing, sovereignty, and recovery as one continuous system.
When the Perimeter Disappears, Security Becomes a Governance Problem
Cloud adoption, supplier dependence, and regulatory pressure are pushing corporate defense away from the network edge and toward identity, software provenance, and control over third parties.
When Cloud Hosts Become Mail Trucks: The Hidden Economy of SMTP Abuse
A reported 230-server operation tied to PCPJack shows how compromised cloud machines can be repurposed into a synchronized SMTP relay layer that blends into ordinary email traffic.
Phishing Kits Are Learning to Borrow Trust, Not Just Brands
Kali365 appears to be expanding a phishing playbook built around identity workflows, showing how token theft and login abuse can travel across very different services.
When a Support Bot Becomes the Soft Spot: Instagram Takeovers and the New Identity Boundary
A reported Instagram hijack case points to a larger security lesson: when AI can influence recovery workflows, the trust boundary moves from login screens to support logic.
ChatGPT Gets a Logout Button, But AI Governance Still Lives in the Blind Spots
OpenAI’s new Active sessions view improves account visibility, yet the harder problem is managing identity, app access, and model changes across a moving SaaS target.
One Click to a Repo Lock: The GitHub Token Trick Hiding in a Browser IDE
A disclosed attack chain involving VS Code and GitHub.dev shows how a single click can become a credential problem, not just a nuisance.
GitHub Actions Missteps Turn Everyday Automation Into a Quiet Injection Risk
An analysis reported that 38% of organizations had GitHub Actions workflows described as vulnerable to script injection or unsafe trigger configurations, a reminder that CI/CD risk often starts with trust in the wrong input.
A Single Click, a Broad GitHub Risk: Why a VS Code Webview Flaw Matters
A reported weakness in Visual Studio Code’s webview layer raises a familiar but dangerous question: what happens when an editor boundary and a GitHub authorization token sit too close together?
When Support Becomes the Back Door: The Meta Bot Incident That Exposed Recovery Risk
A reported abuse of Meta’s AI support bot in Instagram account takeovers shows how recovery flows, not just login forms, can become the real prize for attackers.
When the Help Desk Becomes the Lockpick
Multiple Instagram users lost account access after attackers abused AI-driven support and identity checks, showing how recovery flows can turn into a takeover path.
Dashlane Sees Brute-Force Pressure, With Only Limited Encrypted Vault Downloads Disclosed
A password manager incident is a reminder that account-abuse controls matter most when attackers go after the login layer, not the vault itself.
When a Chatbot Becomes the Front Door: The Instagram Recovery Path Under Pressure
A conversational support tool can be useful, but if it is trusted to handle account changes, attackers may target the recovery path instead of the password field.
Microsoft's File-Open Snag Shows How One Cloud Layer Can Jam the Whole Office
A Teams and Office for the web access incident highlights the fragile path between collaboration apps, backend storage, and browser-based document delivery.
The Quiet Breakout: Why Container Misconfigurations Can Turn Into Host-Level Control
Docker and Kubernetes do not need a novel exploit chain to become dangerous - weak management surfaces, permissive defaults, and exposed runtime interfaces can be enough.
AI Is Breaking the Old Cloud Shortcut
Enterprises are being pushed away from broad cloud slogans and toward workload-by-workload placement choices shaped by data movement, latency, sovereignty, and operating control.
Why Canva’s Security Story Is Really About Controlling Secrets at Speed
Canva’s use of 1Password shows how a fast-growing SaaS company can try to keep teams moving quickly without letting credentials sprawl out of control.
When the Help Desk Learns to Reset the Lock
An alleged Instagram recovery flaw shows how an AI support layer can become a privileged path into account control, not just a convenience feature.