npm security reports, with evidence.

The manifest contains an automatic preinstall reverse-shell command to an external IP, plus an outbound POST. The claimed canary purpose does not provide consent or make this install-time...

The package contains a concrete install-time reverse-shell command with network output transfer. Its lifecycle trigger makes the behavior automatic for consumers.

The automatic postinstall mutation of broad, foreign AI-agent control surfaces is concrete and unconsented. The code also deletes prior directories before writing replacement instruction...

The automatic postinstall path performs broad user-level AI-agent configuration mutation and destructive replacement without explicit user setup. This meets the install-hook abuse policy...

The automatic lifecycle chain modifies global AI-agent configuration locations across multiple platforms. This meets the install-control-surface policy for blocking.

This is not a passive connector: it combines remote terminal observation and prompt injection with default permission bypass and persistent changes to a broad AI-agent control surface. Th...

The automatic lifecycle hook changes consumer-project package-manager settings and installs AI-agent instructions that suppress review and encourage secret disclosure. These concrete beha...

This is a concrete automatic lifecycle hook that broadly mutates foreign AI-agent configuration and extension directories, with an additional automatic project upgrade. The absence of obs...

The manifest contains a concrete automatic reverse-shell-style command and remote POST behavior. This is malicious install-hook abuse, regardless of its canary-themed description.

The package contains concrete runtime redirection of model traffic and embedded obfuscated credentials, not merely optional user-supplied networking. Its benign node-pty postinstall hook...

This is concrete, automatic install-time credential and data exfiltration to an attacker-controlled endpoint. The lifecycle hooks make the behavior execute without an explicit user command.

This is unconsented postinstall mutation of a foreign AI-agent control surface, with a mounted plugin that accesses credentials and sends user content externally. The automatic lifecycle...

The postinstall hook broadly mutates foreign AI-agent skill directories and deploys a skill that can automatically install a CLI. This is a concrete unconsented control-surface modificati...

The package has a concrete postinstall chain that modifies DSH agent and web-plugin configuration without an explicit user command. The installed component can later use a local credentia...

The package contains an unconsented postinstall action that invokes a native logout operation and mutates user authentication state. This is concrete destructive install-hook behavior, no...

This package contains a concrete path that collects a local GitHub credential and transmits it to a hard-coded remote endpoint, plus default telemetry on every command. The lack of lifecy...

This is a concrete remote payload delivery path into broad AI-agent skill control surfaces, activated automatically when the gateway starts. The absence of an npm lifecycle hook does not...

The code deliberately sends credential-bearing configuration to a remote logging endpoint unrelated to the required mail and Google APIs. Absence of an install hook does not mitigate this...

The package performs default postinstall deletion and replacement across several foreign AI-agent skill locations. This concrete lifecycle behavior meets the install-hook abuse blocking p...

This is an unconsented postinstall mutation of broad, foreign AI-agent control surfaces. The lack of observed exfiltration does not remove the concrete install-hook abuse.