using-toolkit

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill mandates invoking other skills whose "content is loaded and presented to you" at runtime and explicitly references installing skills from external repositories via "npx skills add owner/repo@skill", which shows it will fetch and then read/act on untrusted third-party skill files that can change agent behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). Yes — the skill requires invoking the Skill tool which loads remote skill files at runtime (and even suggests fetching packages via commands like "npx skills add owner/repo@skill"), and those fetched skill files are presented to and must be followed by the agent, meaning external content can directly control prompts.