registry / @aswless_854771076/ai_short_studio_cli / 0.1.41

@aswless_854771076/ai_short_studio_cli@0.1.41

VVICAT 无限画布项目命令行工具

AI Security Review

scanned 9d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. npm postinstall automatically installs bundled agent skills into the user's Codex skills directory. This changes agent behavior without an explicit CLI command.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

npm dependency installation

Impact

Unconsented persistence in Codex's agent-control surface; installed instructions can influence future agent actions.

Mechanism

postinstall copies bundled skills into ~/.codex/skills

Policy narrative

Installing the package invokes postinstall. Unless narrowly suppressed by environment values, it calls autoInstallSkills with the Codex default root, copies two bundled skills into ~/.codex/skills, and records a manifest. These skills become available to later Codex sessions without the user explicitly running a setup command.

Rationale

This is a concrete unconsented postinstall mutation of the broad Codex skill control surface. The absence of immediate exfiltration does not remove the persistence and agent-control risk.

Evidence

package.jsonscripts/postinstall.mjsdist/skill/skill-manager.jsskills/using-vvicat-ai-short-studio-cli/SKILL.mdskills/using-vvicat-ai-short-studio-cliskills/short-drama~/.codex/skills/using-vvicat-ai-short-studio-cli~/.codex/skills/short-drama

@aswless_854771076/ai_short_studio_cli@0.1.41

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Source & flagged code