npm security reports, with evidence.

The automatic lifecycle hook modifies existing agent configurations and installs a package-controlled MCP command without an explicit user setup action. This meets the install-hook abuse...

This is an unconsented install-time mutation of a foreign AI-agent control surface with a wildcard command hook. It meets the install-hook abuse blocking policy even without proving exter...

The package automatically redirects Codex traffic to an unofficial external service and obscures bundled provider credentials. This is concrete unauthorized data exfiltration behavior, de...

This is an unconsented postinstall mutation of broad, foreign AI-agent control surfaces, with destructive replacement and persistent plugin activation. The automatic self-update path incr...

The automatic lifecycle hook mutates a foreign AI-agent control surface and multiple persistent user locations. This is concrete unconsented install-time control-surface behavior, despite...

The automatic postinstall writes package-controlled instructions into a global AI-agent command directory and refreshes them on upgrade. This meets the install-hook abuse policy regardles...

The automatic lifecycle hook changes consumer-project agent instructions rather than limiting setup to an explicit user command. This meets the install-control-surface blocking policy.

The package performs automatic postinstall mutation of both Codex and Claude skill directories, which is a foreign and broad AI-agent control surface. It also installs Python dependencies...

The runtime cleanup contains a concrete broad destructive action that is not limited to package-owned processes. The scanner-addressed comment reinforces malicious intent, while the posti...

The automatic postinstall reaches persistent global Codex and Claude guidance writes with no interactive approval. This matches the install-control-surface block policy despite no evidenc...

The package contains a concrete automatic data-reporting path that captures S3 request content and uploads it remotely. Lack of an install hook limits activation to runtime, but does not...

The package covertly attaches reporting to S3 operations and sends captured request data to a remote endpoint. This is not necessary for the requested storage operation and lacks an expli...

The package contains a concrete runtime data-exfiltration path from an AI-agent control store to a remote service. Its lack of an install hook limits the trigger but does not remove the u...

The package automatically sends a broad project archive to a remote endpoint from a development command, with inadequate secret exclusions. The absence of an install hook does not remove...

The package uses an automatic npm lifecycle hook to overwrite instruction-bearing skill directories for multiple AI-agent hosts. The global-install guard reduces scope but does not provid...

The automatic postinstall hook alters global Claude and Codex MCP configuration, which is a concrete unconsented AI-agent control-surface write. This meets the install-hook abuse blocking...

The package performs unconsented postinstall mutation of foreign, broad AI-agent configuration files. That concrete automatic control-surface change meets the publish-block policy even wi...

This is concrete automatic install-time mutation of foreign, global AI-agent control surfaces and the consuming project. The lack of observed network exfiltration does not remove that ins...

The package embeds and executes concealed code at import time while actively resisting inspection. That is a concrete arbitrary-code execution surface, not transparent package functionality.

The automatic lifecycle hook modifies a broad, foreign AI-agent control surface and installs code triggered by normal agent activity. The setup-state guard does not provide contemporaneou...

The package uses an automatic postinstall hook to mutate multiple foreign AI-agent control surfaces and install event-triggered commands. This is concrete unconsented install-time control...