profile

CybersecurityOS 🛡️

Node2Shell, stolen AI keys, and poisoned IDEs


This week, attackers went after the things developers trust by default: a compression library, a hosting platform holding your secrets, and the extensions inside your IDE. Here's what matters.


CISA: Node2Shell is being exploited

CISA confirmed on Tuesday that hackers are exploiting a critical remote code execution flaw in Brotli compression, known as Node2Shell, Broccoli2Shell, and HTTP2Shell. Find every service that depends on Brotli, directly or through a framework, and apply the patches or workarounds now.

Read more at BleepingComputer →

Tools I use: I use Make to pipe vulnerability feeds into enrichment and a phone alert, so critical ones like this reach me without another cron job. That's my own setup, not part of the story. (Affiliate link: it supports CyberSHIELD at no cost to you.)


Northflank breach: customers' AI API keys abused

A breach at cloud hosting platform Northflank led to unauthorized use of customers' AI API keys, and the company is telling customers to rotate all credentials and secrets stored there. If you're a customer, rotate everything, not just the AI keys. Everyone else: how fast could you rotate every secret your provider holds?

Read more at BleepingComputer →


A trojanized Claude Code extension targets Cursor users

Dark Reading reports the AI-powered IDE Cursor had two security incidents on the same day, including a trojanized Claude Code extension. Your IDE holds code, tokens, and cloud credentials, so treat extensions like any third-party software: verified publishers and a team allowlist.

Read more at Dark Reading →


VOLTA infostealer uses LLMs to rank its victims

Cisco Talos details VOLTA, an infostealer that grabs credentials, sessions, and crypto wallets, then uses large language models to profile victims and rank the most valuable loot. It also deletes itself. Short session lifetimes and fast revocation limit what a stolen session is worth.

Read more at Cisco Talos →


CVSS 10.0 flaw in LXD and MicroCloud

A flaw in Canonical's LXD and MicroCloud let an attacker with access to the HTTPS API and a trusted client's public certificate log in as that client without its private key, gaining access to the cluster and its VMs. Update, and keep management APIs off the open internet.

Read more at The Hacker News →


An AI model chained bugs into working exploits

An unreleased Anthropic model found flaws in Chrome, Safari, Android, Windows, and Linux and chained them into proof-of-concept exploits in a lab. None were used in real attacks, and vendors patched them all. Expect more fixes, arriving faster.

Read more at The Hacker News →

Tools I use: Building a security career? A simple site for your write-ups does a lot of work for you. I use Carrd for quick one-page sites. My own recommendation, unrelated to this story. (Affiliate link: it supports CyberSHIELD at no cost to you.)


Also this week: Operation Serengeti's 1,462 arrests · Ryan Fetterman on "blue collar" detection engineering · Ask HN: building a CVE-tracking agent responsibly

Want the full breakdown? Read this week's OS Weekly on the blog. Hit reply and tell me which of these hit your team. See you next Sunday.

Michael

CybersecurityOS 🛡️

Engineering security from first principles. I'm d0uble 3L, I write CybersecurityOS, where I break down secure-by-design architecture, DevSecOps, cloud security, and emerging-tech risk into practical frameworks for engineers, leaders, and teams. Weekly perspectives, clarity over complexity.

Share this page