create-description

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

• Third-party content exposure detected (high risk: 1.00). The skill's mandatory Step 2 ("Research the Niche") directs the agent to use WebSearch and WebFetch on top public URLs to extract description examples and keywords from top creators' pages (SKILL.md — "Use WebSearch" / "Use WebFetch on the top 1-2 most relevant URLs"), meaning it ingests untrusted, user-generated web content that directly informs generation and tool decisions.

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

• Potentially malicious external URL detected (high risk: 0.80). The skill requires running WebSearch/WebFetch on the "top 1-2 most relevant URLs" (i.e., dynamic external URLs returned by WebSearch) at runtime and explicitly instructs extracting page content to drive prompts and generation, so those fetched external pages (WebFetch on top search result URLs) directly control the agent's output and are a required dependency.