ads-quick

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

• Third-party content exposure detected (high risk: 1.00). The skill's required "Step 1: Fetch and Analyze the Homepage" instructs the agent to use WebFetch to retrieve and interpret an arbitrary public website's homepage (third-party content) as part of its scoring and platform recommendations, which could allow indirect prompt injection from untrusted web content.

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

• Potentially malicious external URL detected (high risk: 1.00). The skill's runtime explicitly calls WebFetch(url) to retrieve the user-provided homepage and injects that fetched content into the model's analysis (i.e., the user-supplied URL passed to WebFetch controls the agent's prompts), making the external URL a required runtime dependency.