From npm to PyPI: How Supply Chain Attacks Are Getting Harder to Catch

Analysis

OS-Level Sandboxing Blocked the Axios RAT Where Every Other Tool Failed

This is a concrete technical breakdown of exactly how a capability-based sandbox (using Landlock and Seatbelt) would have stopped the axios attack at multiple points including network isolation and filesystem writes. If you're running npm install in any automated pipeline, this is the architecture you should be studying.

AI Agents Massively Amplify Supply Chain Risk Compared to Traditional Package Ecosystems

AI agents install and execute tools autonomously without human review, which makes supply chain compromise fundamentally more dangerous than in traditional dev workflows. This piece connects historical npm and PyPI attacks to the emerging AI agent threat surface in a way that's actually useful for builders.

The Axios Attack Exposed the External Blind Spot in Enterprise Security Tooling

Internal agent-based security tools are blind to forgotten cloud deployments and unmonitored dev environments where supply chain attacks breed. This covers how external attack surface mapping surfaces exposed npm configs, git tokens, and unmanaged Node.js deployments before attackers find them.

Tools

Practical Commands to Detect and Remediate the Axios Supply Chain Compromise

Actionable detection guide with terminal commands for npm, yarn, and pnpm lockfiles plus system-level checks for the RAT payload across macOS, Windows, and Linux. Copy-paste this into your runbook now.

News

LiteLLM PyPI Supply Chain Attack: Malware in a 97M-Download AI Library Went Undetected by Every Security Tool

A poisoned LiteLLM version harvested SSH keys, cloud credentials, and Kubernetes configs before being caught only because the malware had a bug that crashed machines. Zero security tools flagged it, which is the real story here.