DevSecOps

The Cache That Bites Back: GitHub Actions Cache Poisoning Attacks

How attackers turn GitHub Actions' shared build cache into a supply chain weapon — real cases, attack mechanics, detection logic, and mitigations.

GitHub's VS Code Extension Breach: What We Know, What We Don't, and How to Defend

GitHub says an employee device was compromised through a poisoned third-party VS Code extension and internal repositories were exfiltrated. Here is the fact-checked breakdown for defenders.

LiteLLM CVE-2026-59822: Your AI Gateway Is a Cloud Control Plane

An exploited LiteLLM MCP authentication bypass, unsafe defaults, code-executing guardrails, and unrestricted pass-through routes show why AI gateways must be isolated and operated as Tier-1 infrastructure.

GitLab CVE-2026-85706: Patch the File Read, Then Rotate What It Exposed

CISA lists GitLab CVE-2026-85706 as exploited. Self-managed administrators must upgrade to 19.3.2, 19.2.6, or 19.1.8 or later, hunt the repository commits API, and treat readable secrets as potentially compromised.

JFrog Artifactory CVEs: Your Artifact Repository Is a Build Boundary

CISA KEV activity and a fresh critical Artifactory authentication bypass show why package repositories need incident-grade monitoring, not just routine patching.

TeamPCP Arrests in Australia: The Supply Chain Lesson Is Bigger Than Two Suspects

Australian and U.S. authorities have charged alleged TeamPCP operators after a software supply chain campaign that authorities say hit more than 1,000 organizations. The defensive lesson is about tokens, publishing rights, and update speed.

NGINX's Configuration-Dependent Vulnerabilities: Why Version Scanning Is Not Enough

NGINX 1.30.4 and 1.31.3 fixed three new memory-safety flaws, but exposure depends on map, slice, SSI, proxy, and buffering configuration. Here is how to audit the real path.

DuneSlide: How a Prompt Injection Became Full RCE in Cursor IDE

Cato AI Labs found two 9.8 CVSS flaws in Cursor's terminal sandbox — CVE-2026-50548 and CVE-2026-50549 — that let a poisoned MCP response or search result silently escape to full remote code execution. Neither requires a click.

STRIDE Threat Modeling: A Practical Guide Nobody Needs a Tool to Start

STRIDE has been Microsoft's threat modeling framework since 1999 and still works. Here's how to run a real session with a whiteboard, a data flow diagram, and 30 minutes — no expensive tooling required.

Dependency Confusion: Your Internal Package Name Is a Public Attack Surface

Publish a package with the same name as a company's private one, give it a higher version number, and package managers will happily install the attacker's code instead. Here's how it still works in 2026.

Minimum Package Age: The Supply Chain Control That Buys Defenders Time

Fast takedowns do not protect systems that auto-install malicious packages or extensions in the first minutes after release. Minimum package age turns time into a practical supply chain defense.

Quasar Linux QLNX: A Developer Workstation RAT Built for Supply Chain Access

Trend Micro documented QLNX, a Linux RAT that combines credential harvesting, LD_PRELOAD persistence, PAM backdoors, and rootkit behavior. The real risk is not one infected host - it is the supply chain access behind it.

GitHub Finally Puts a Human in the Loop: npm Staged Publishing Explained

npm packages no longer publish instantly. GitHub's staged publishing forces a 2FA-gated human approval before any version hits the registry — here's what it means and how to enable it.

GitHub Secrets Management Crisis: 65% of AI Companies Leaked Credentials

65% of Forbes AI 50 companies leaked secrets on GitHub with 94-day median remediation time. Blue team guide to detect, prevent, and respond to repository leaks.