@ai-content-space/loopx@0.4.1

One or more suspicious static signals were detected.; previous stored version diff introduced dangerous source

npm install postinstall

Can alter future Codex/Claude agent behavior for the user by adding bundled skills, hook scripts, and Claude hook settings.

unconsented AI-agent skill and hook installation

On npm install, the postinstall script imports the installer and runs installSkillsForTargets with default Codex and Claude targets. That writes bundled skill directories plus managed hook scripts into user agent locations, and for Claude merges a UserPromptSubmit hook command into settings.json. This is package-related behavior, but it is activated by lifecycle install rather than an explicit user CLI command.

The package does not show data theft, destructive behavior, or remote payload loading, but it performs unconsented lifecycle mutation of AI-agent skill/hook surfaces. For an upstream firewall, that is concrete install-time agent control-surface abuse even if the content is mostly workflow-oriented. Product guard normalized a non-low false-positive publish_block request to warn-only suspicious.