Investigate malicious npm packages and AI-agent policy findings. Each report identifies the package version, supporting evidence, advisory sources, and current Firewall decision. Malware findings and policy violations are labelled separately.
Updated 13 Oct 2026, 04:29 UTC. Ordered by latest scan.
This is a concrete unconsented npm postinstall mutation of a foreign AI-agent control surface. The conditional hook/MCP path does not mitigate the unconditional skills write.
The source establishes unconsented install-time host fingerprint exfiltration and system mutation, not merely dormant capability. The telemetry endpoint is identifiable, but that does not...
Source inspection confirms a deliberately obfuscated, unconditional browser redirect that forwards query parameters after a deceptive verification screen. The absence of install hooks lim...
The install-time hook performs concrete privileged persistence and remote-access weakening, not merely a user-invoked administrative utility. No exfiltration endpoint is needed for this h...