npm detections from LPM Firewall.

The lifecycle hook creates a complete automatic chain from npm installation to writes in multiple AI-agent skill directories. This meets the policy for a blocking install-hook control-sur...

The package contains an automatic install hook connected to active credential harvesting and outbound transmission. This is concrete malicious data exfiltration.

The automatic lifecycle chain modifies consumer configuration, agent instructions, and Git behavior. This meets the policy threshold for an install-time AI-agent control-surface attack.

The package performs a concrete, automatic postinstall chain that changes consumer lifecycle policy and installs agent-control artifacts. This meets the install-control-surface blocking p...

The automatic lifecycle chain modifies consumer package-manager configuration and agent instruction/capability locations. This meets the install-control-surface blocking policy despite no...

This is an active, concealed remote redirect chain that forwards browser URL parameters to a remotely selected destination. The package has no visible package-aligned functionality beyond...

The automatic postinstall hook mutates a foreign, broad AI-agent control surface by installing command hooks. This meets the install-control-surface block policy.

An automatic postinstall invokes opaque setup that targets multiple AI-agent configuration surfaces. This is concrete unconsented install-time agent control-surface mutation.

The package creates a concrete default path from an existing OpenAI credential to a ZubrIQ endpoint and sends that credential in an Authorization header. This is credential exfiltration b...

The package contains an import-triggered, detached remote payload loader that executes server-authorized JavaScript. This is concrete remote code execution, not ordinary item validation.

The automatic lifecycle hook overwrites multiple AI-agent skill locations, including foreign product directories. This meets the install-control-surface block condition.

The package has a concrete automatic lifecycle hook that force-replaces skill directories for several unrelated AI-agent environments. This meets the install-control-surface blocking poli...

The automatic lifecycle hook changes an AI-agent plugin configuration outside an explicit user setup action. Its home-level configuration targeting makes the mutation a broad control-surf...

The source contains an active remote command execution path combined with security-control weakening and agent configuration mutation. This is concrete malicious behavior, not a passive i...

Active, concealed execution of an image-carried PowerShell command occurs on normal package use. The hidden detached launcher is concrete malware behavior.

The launcher deliberately disables or weakens host security controls before downloading and executing a remote binary. This is concrete malicious behavior despite requiring an explicit CL...

The package automatically configures global Claude telemetry during installation and enables collection of sensitive agent content. This is concrete install-hook abuse, not an explicit us...

The package uses an npm postinstall hook to mutate global Claude and Codex MCP control surfaces automatically. This meets the policy definition for a blocking agent-control hijack.

The automatic lifecycle hook mutates Claude and Codex user configuration to execute package code. This is a concrete unconsented AI-agent control-surface modification.

The automatic lifecycle hook performs a broad external Claude skills mutation and destructive replacement. No user-invoked setup gate is present.

The automatic lifecycle path performs broad AI-agent configuration and plugin wiring in a consumer project. This meets the policy threshold for unconsented install-time mutation of foreig...