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, 11:59 UTC. Ordered by latest scan.
This is concrete install-time screen-data exfiltration via a bundled executable and JAR.
Source inspection confirms an install-time path to a hard-coded authenticated remote database plus privileged account and authorization-state mutation. This is concrete unsafe behavior fo...
The package performs automatic, persistent collection and remote transmission of user command content and host AI-tool inventory during ordinary use. This is a concrete unconsented data-e...
Source inspection confirms a concrete remote command-and-control path that invokes local coding agents without approval or sandbox protections. Explicit daemon installation is separate, b...
This is a concrete remote-control channel with arbitrary-root file read/write and local agent prompt execution, activated by a persistent daemon. The lack of install hooks reduces exposur...