npm detections from LPM Firewall.

The package automatically modifies consumer-project AI-agent configuration through postinstall. The local-only implementation does not remove the concrete control-surface hijack.

The automatic lifecycle hook overwrites skills in several unrelated agent environments, including Codex and Claude. This is concrete install-hook abuse even though the hook does not make...

The compiled source establishes an import-time default exporter that posts trace data to an unrelated hard-coded HTTP address. This is concrete data-exfiltration behavior, not ordinary pa...

This release performs an unconsented install-time deletion of another named AI coding binary and forms a remote executable delivery chain. Although no credential theft is visible in the J...

The package has no install hook, but its normal runtime path automatically exports conversation data and enables cloud-relayed control of local agent sessions. Those capabilities are acti...

This is a concrete postinstall mutation of foreign AI-agent control surfaces, followed by an automatic package-install path. The behavior meets the install-hook abuse blocking policy desp...

This is an unconsented postinstall mutation of broad AI-agent control surfaces in the consumer project. The lack of network theft does not remove the concrete install-time control-hijack...

This is an unconsented postinstall mutation path for a foreign AI-agent control surface, backed by a native payload containing an explicit safety-bypass command. The remote download path...

This is a concrete automatic install-hook chain that executes an opaque native installer against AI-agent control surfaces. The binary's embedded unsafe agent commands make the resulting...

This is a concrete automatic install hook that retrieves and executes an unauthenticated native payload, then requests agent-hook installation. That combination is malicious install-hook...

This is unconsented postinstall execution that installs AI-agent hooks through an opaque native binary, with a remote-download fallback. It meets the install-control-surface block criterion.

The published runtime automatically exfiltrates agent transcripts to a remote endpoint while suppressing failures and providing no opt-out. This is concrete data exfiltration, not merely...

The automatic postinstall performs broad cross-vendor AI-agent control-surface writes and configuration changes. This meets the install-hook abuse blocking boundary despite the absence of...

The package contains active, default-configured external logging routes with hard-coded webhook credentials and sends rich application records to them. The absence of an install hook does...

The package has a concrete postinstall path that automatically persists integrations across three AI-agent environments on global installation. This meets the install-hook abuse policy ev...

This is an unconsented postinstall mutation of broad AI-agent control surfaces, preceded by automatic executable acquisition and execution. It meets the install-control-surface block poli...

The package uses an automatic npm lifecycle hook to copy package-controlled skill content into shared and third-party AI-agent directories. This meets the install-time AI-agent control-su...

This is an automatic postinstall mutation of user-level Claude configuration that establishes ongoing command execution. The behavior meets the install-control-surface block policy despit...

The package contains a concrete runtime path that exports license keys, host metadata, and exception text to a third-party endpoint. The benign local build-only install hook does not offs...

The inspected source establishes an automatic npm lifecycle chain that writes third-party agent configuration and rules, producing persistent automatic package execution. This meets the i...

This is concrete install-hook abuse: an automatic npm lifecycle script modifies global Claude settings and installs recurring command hooks without user confirmation. The absence of netwo...