Home/🛡️ cybersecurite🛡️🛡️ CYBERSECURITECritical flaw: Claude Code executes hidden GitHub repo malware without verificationMozilla researchers (0DIN platform) showed that a compromised GitHub repo can infect a machine as soon as an AI tool like Claude Code runs its setup script.Published 2sem·1 sourceImportant Lire en françaisListen≈ 29sSpeed0.8×1×1.2×1.5×The factThe malware, loaded dynamically via a DNS query at runtime, remains undetectable to scanners and the AI agent itself.This attack grants full control of the target machine to attackers, bypassing traditional security mechanisms.🔗Click the link to read the original article:Dev.to↗🛡️🧭Explore this topic🏢Mozilla🏢Anthropic (Claude Code)🏢GitHub#malware#GitHub#DNS exfiltration#IA coding#sécurité logicielle#compromissionToday's news on Factae→↗ Share the newsAuto-synthesis from 1 media source · identified on June 29, 2026← Back to home