악성 코드 다이제스트 월간 요약: 5월

Welcome to the May edition of the Xygeni Malicious Code Digest. This month, our security research team continued tracking the evolving landscape of software supply chain threats across public package registries, developer tooling, AI-assisted workflows, and modern CI/CD environments. Throughout May, Xygeni identified and validated hundreds of malicious packages targeting npm, PyPI, VS Code, and OpenVSX ecosystems.

But May was not only about volume. It was also a month where attackers increasingly adapted their tradecraft to AI-native development environments, developer trust assumptions, and automated pipelines.

Among the most notable investigations published by the Xygeni Security Team this month:

• 줄스잭커, a multi-stage npm malware campaign impersonating Google’s Jules AI agent to steal repositories, abuse CI/CD environments, and even target the sandbox analyzing it. The campaign introduced encrypted host-gated loaders, forged google-labs-jules[bot] commits, cloud metadata theft, and Kubernetes reconnaissance against analysis infrastructure itself.
• 팬텀봇, a malicious npm campaign abusing fake wallet, blockchain, and Web3 tooling to steal credentials, compromise developer environments, and target crypto-related workflows through dependency abuse and staged payload delivery.
• A large-scale malicious npm ecosystem abusing fake TypeScript utilities, AI-themed packages, internal tooling impersonation, frontend libraries, cloud helpers, and builder frameworks to fingerprint environments, exfiltrate credentials, deploy persistence, and compromise developer systems.
• Continued waves of malicious packages leveraging dependency confusion, typosquatting, postinstall abuse, AI-tool impersonation, version inflation, and repeated namespace abuse designed to bypass developer trust heuristics and compromise modern SDLC workflows before detection.
• The discovery of cross-platform malware campaigns such as the @jaggle/resizeobserves clipboard hijacker, which impersonated a legitimate package while deploying Python-based persistence and crypto-wallet clipboard substitution across Linux, macOS, and Windows systems.

Throughout the month, we also observed a growing convergence between AI tooling, open-source ecosystems, and software supply chain attacks. Threat actors increasingly targeted AI-assisted development workflows, agent environments, copilots, and cloud-native build pipelines as part of broader credential theft and repository compromise operations.

These attacks are no longer isolated cases of simple typosquatting. They involve coordinated publishing campaigns, automated version bursts, cloud token theft, CI/CD abuse, encrypted payload staging, sandbox evasion, and persistent compromise techniques specifically designed to blend into trusted developer ecosystems.

This monthly update is part of Xygeni’s ongoing malware and supply chain threat research initiative, where our team continuously validates malicious packages, investigates emerging attack patterns, and publishes actionable intelligence to help security, AppSec, and DevSecOps teams stay ahead of evolving supply chain risk.

For full context across every malicious package, malware campaign, and threat analysis published this month, explore the complete May Malicious Code Digest and related research from the Xygeni Security Team.