npm security reports, with evidence.

Active import-time collection and export of cloud and local secrets is concrete credential exfiltration. The package should be blocked.

The active entry point automatically harvests sensitive cloud and local secret material and sends it to an unrelated external endpoint. This is concrete malicious credential and data exfi...

The package contains active automatic collection and external transmission of cloud credentials and local secrets. This is concrete credential and data exfiltration behavior.

This is active, import-triggered secret harvesting and exfiltration to an unrelated fixed endpoint. The behavior is concrete and requires blocking publication.

The package uses postinstall to alter a broad AI-agent control surface and enable export of Claude session content to its endpoint. This is concrete unconsented install-time agent configu...

This is an unconsented postinstall mutation of a broad AI-agent configuration that enables export of sensitive Claude Code content. The opt-out does not make the automatic default behavio...

This is an automatic install-time remote-code-execution chain that persists and executes an unverified latest-release binary. It is concrete attack behavior rather than an inert downloader.

This is unconsented postinstall mutation of a broad AI-agent control surface that enables export of sensitive interaction telemetry. The behavior is active by default and is sufficient to...

This is concrete automatic credential harvesting and data exfiltration during installation, concealed by heavy obfuscation. It is unrelated to a legitimate dotenv preflight check.

The active constructor-to-detached-payload chain establishes concealed code execution. Opaque encrypted payload storage and in-place decryption make this malware behavior rather than ordi...

The browser entrypoint contains deliberately obfuscated remote-configured redirect logic rather than a legitimate package function. Its active forwarding of visitor URL parameters to a re...

The automatic global postinstall path modifies external agent configuration and disables a named safety preload for its child process. Those concrete behaviors meet the install-control-su...

Automatic postinstall execution of a remotely selected binary followed by AI-tool onboarding is a concrete unconsented install-time control-surface risk. The opt-out is not an affirmative...

The package performs automatic installation-time execution of a remotely supplied binary and invokes AI-tool onboarding by default. That is a concrete unconsented AI-agent control-surface...

An install-time hook executes an opaque, remotely selected binary command advertised as connecting AI tools. This is an unconsented lifecycle path to a broad AI-agent control surface.

The package has an automatic postinstall path that mutates foreign AI-agent dependency code. This meets the install-time foreign control-surface mutation block condition.

The package uses postinstall to inject agent-directed guidance into consumer projects without an explicit user command. This is concrete install-hook abuse affecting AI-agent behavior.

The automatic lifecycle hook performs concrete cross-platform Claude Desktop configuration mutation. This meets the install-hook abuse blocking policy.

The automatic post-install mutation of foreign AI-agent runtime code meets the blocking policy. Its narrow scope reduces uncertainty but does not establish consent for the dependency rewr...

The automatic lifecycle hook executes an opaque downloaded or bundled binary specifically to alter Codex and Claude Code integration. This is a concrete install-time AI-agent control-surf...

The active import-time server and concealed remote credentialed client are incompatible with the minimal package identity and create concrete unconsented forwarding of user input. The lac...