The automatic package-owned skill setup is a meaningful agent extension capability, but the inspected source does not establish a concrete attack. The manifest has no npm lifecycle hook, so this supports a warning rather than a publication block.
Static reason
High-risk behavior combination matched malicious policy.; source matched previously finalized malicious package; routed for review
Trigger
Starting the host and opening a Claude session; Codex skill registration follows app-server initialization.
Impact
These integrations can influence agent behavior through package-owned instructions, and the Claude skill can refresh automatically. The inspected chain did not establish credential theft or unrequested account changes.
Mechanism
The Claude host fetches skill files from the Lizard GitHub repository and supplies the directory as a plugin. The Codex host registers its bundled skills directory as an extra root.
Evidence
src/host/claude-host.mjssrc/host/codex-host.mjssrc/host/skills/lizard/SKILL.mdsrc/host/skills/lizard/README.mdsrc/host/skills/lizard/skills.sh.json