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 10 Oct 2026, 06:43 UTC. Ordered by latest scan.
This is a concrete remote code execution backdoor activated through the package's normal export. Concealment, detached execution, and remote dynamic evaluation leave no package-aligned be...
This is an unconsented postinstall mutation of a broad AI-agent control surface. The automatic hook, skill, statusline, and global instruction changes meet the install-hook abuse blocking...
The package contains a concrete default-enabled data-export path to hardcoded external webhooks. It is not limited to an explicit user logging command or a caller-selected endpoint.
The complete lifecycle chain automatically alters global AI-agent configuration and instructions without interactive consent. This meets the install-hook abuse blocking policy even though...