npm detections from LPM Firewall.

This is concrete unconsented install-time remote code execution with persistence. The harmless runtime stub does not mitigate the lifecycle attack.

The package contains a concrete remote download-and-shell execution path plus silent privileged self-updating. Its install hooks are cosmetic, but runtime behavior is sufficient to block...

This is an unconsented install-time transmission of local system identity data to an external endpoint. The behavior is concrete and unrelated to the package's runtime code.

This is an unconsented postinstall mutation of a foreign AI-agent control surface, including enabled command definitions. Conflict protection does not remove the risk because a new worksp...

The package performs automatic install-time host identification and sends it to an external endpoint. This is concrete data exfiltration with no legitimate implementation in the package.

The package performs unconsented install-time host fingerprinting and external transmission. Its empty runtime code does not justify that behavior.

The package performs unconsented install-time host reconnaissance and exfiltrates it to a fixed external endpoint. Its only runtime file is empty, leaving no package function that justifi...

The automatic SVG import, deliberate obfuscation, and dynamic code execution are incompatible with a passive SVG package. Lack of install hooks reduces install-time risk but does not remo...

Source inspection confirms a concrete concealed capture-and-AI-solving payload with proctoring evasion and persistence. The benign postinstall hook does not reduce the risk of the explici...

Source inspection confirms a concrete stealth, persistence, capture, and AI-assisted cheating workflow. Its credential-handling code and bundled credential material compound the risk, so...

This is unconsented postinstall mutation of broad, foreign AI-agent control surfaces, including workspace trust. The lifecycle behavior is concrete and sufficient to block even though the...

The postinstall lifecycle performs broad global Claude and Codex control-surface writes and registers continued execution on Codex turn completion. This meets the policy threshold for mal...

This is a concrete unconsented postinstall mutation of a consumer AI-agent control surface, combined with remote vendor-authorized content deployment. Signature verification does not remo...

The package uses an automatic npm lifecycle hook to write and prune skills in both Codex and Claude user directories. This is unconsented postinstall mutation of a broad AI-agent control...

The package performs unconsented postinstall writes and deletions in both Codex and Claude user skill locations. This meets the install-control-surface policy for blocking.

This package performs an automatic postinstall write to global Codex and Claude skill directories and prunes their contents. That concrete lifecycle control-surface mutation meets the blo...

The package automatically mutates the global Claude skills directory and installs instructions that take effect across Claude responses. This is unconsented install-time control-surface m...

The package has a concrete automatic postinstall path that injects and prunes skills in two unrelated agent homes. This meets the install-time AI-agent control-surface abuse policy.

The confirmed postinstall hook modifies both Codex and Claude configuration trees without a user-invoked setup action. This meets the policy threshold for malicious install-hook abuse eve...

This package uses an automatic install hook to persist a GitHub Actions workflow in a consumer repository and transmit repository metadata externally. The behavior is unconsented install-...