Malicious code in microsoft-applicationinsights-common (npm)
Two malicious npm packages published by the micresoft account (typosquatting "microsoft") are part of a coordinated supply chain attack sharing identical infrastructure with packages published by the superbase account. Each package bundles a 4.5 MB statically-linked, UPX-packed ELF binary at .claude/settings and a companion .claude/settings.json that registers the binary as a Claude Code SessionStart hook, causing it to execute every time Claude Code opens the compromised project directory. On initial install, the same binary is executed via a preinstall lifecycle hook. The binary connects to a C2 server at 207.90.194.2:443 and harvests environment variables, $HOME directory contents, and /proc/ filesystem entries to exfiltrate developer credentials and system state.
microsoft-applicationinsights-common impersonates @microsoft/applicationinsights-common, the official Microsoft Application Insights Common JavaScript Library. The package copies the legitimate library's full distribution including compiled browser bundles, TypeScript declarations, and metadata, then injects a preinstall hook that executes the malicious .claude/settings binary on install.
-= Per source details. Do not edit below this line.=-
This malicious package is part the IronWorm campaign. This campaign executes a malicious binary payload during installation via a preinstall hook. The payload is a Rust-built infostealer that targets developer environments, scanning for and harvesting credentials related to cloud providers, object storage, databases, source-control, package registries, and AI developer tools. It also targets cryptocurrency wallets, specifically injecting a malicious JavaScript hook into the Exodus desktop wallet to capture passwords and recovery phrases. Furthermore, the malware exhibits worm-like behavior by stealing GitHub and NPM credentials to push malicious updates to the victim's repositories and publish trojanized packages, and it uses an eBPF-based kernel rootkit to hide its processes and network connections on Linux systems.
왜 이 VPI인가 (설명가능 · 실험적)
VPI 산정 기준
| 영향도(기본값(정보 없음)) | 55.00 |
| 악용 신호(추가 악용신호 없음) | ×1.00 |
| VPI | 55.00 |
VPI 공식 vpi-v1 기준