npm AI-agent policy reports, with evidence.

Source confirms an automatic postinstall write to global Claude Code configuration and executable command registration. That broad user-wide control-surface change meets the blocking poli...

This is a concrete, unconsented postinstall mutation of broad foreign AI-agent control surfaces, meeting the firewall block boundary despite the lack of observed exfiltration.

The automatic lifecycle hook performs unconsented mutation of foreign global Claude/Codex agent control surfaces and registers executable lifecycle behavior. This meets the install-time A...

This is concrete, unconsented install-time mutation of a shared/foreign AI-agent credential and configuration surface, coupled with automatic account provisioning and provider redirection...

The postinstall performs concrete destructive mutation of credentials/settings associated with Anthropic/Claude and automatically replaces credentials through a vendor network request. Th...

The package performs unconsented postinstall mutation of a foreign/broad Claude-compatible AI-agent control surface, including deleting credentials and replacing them with its own provide...

The package performs concrete, unconsented postinstall writes to an existing AI-agent prompt/control surface. This meets the firewall block criterion regardless of the absence of observed...

The package performs an automatic, unpinned remote-code execution chain during npm postinstall and uses it to register AI-agent hooks. This meets the firewall block boundary for unconsent...

The shipped source establishes an automatic remote-payload execution chain and unconsented Claude Code hook registration. No source-level integrity check constrains the executed script or...

Source confirms a concrete install-time credential/config replacement chain, not merely suspicious primitives. This meets the blocking policy for unconsented postinstall mutation of a for...

Source confirms an npm postinstall path that modifies Windows Terminal profiles for Claude/Codex and establishes a persistent scheduled task. This meets the install-time AI-agent control-...

The install-time, unprompted mutation of Windows Terminal profiles used to control Claude/Codex, combined with elevated persistence, is a concrete malicious chain. The locally configured...

The package performs concrete, automatic Windows persistence and foreign Terminal profile mutation from npm postinstall. This exceeds a normal CLI install hook and meets the install-time...

The unconsented postinstall modifies a foreign AI-assistant launch surface and installs persistent elevated execution. This is a concrete install-time control-surface and persistence chai...

This is a concrete, unconsented postinstall mutation of foreign AI-agent control surfaces with persistent hooks and relay behavior. The existing-directory guard does not make the installa...

This is unconsented install-time mutation of a foreign AI-agent control surface, meeting the block policy even though the hook is guarded and package-branded. The source also establishes...

The concrete postinstall behavior meets the firewall block criterion for foreign/broad AI-agent control-surface mutation. No separate network or credential-exfiltration behavior was found.

The lifecycle hook performs broad, persistent Claude/Codex interception and scheduled execution on global install. Under the firewall policy, this concrete unconsented postinstall control...

This is concrete unconsented postinstall mutation of broad host AI-agent control surfaces, including third-party plugin installation. No exfiltration was found, but the install-time contr...

This is concrete, unconsented postinstall mutation of project and user-wide AI-agent control surfaces. It meets the firewall block policy despite no observed exfiltration or remote payloa...

The lifecycle script concretely modifies global Claude Code and Codex skill roots on installation, meeting the firewall block condition. The later hook trust flow does not negate the earl...