visual-style

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

• Third-party content exposure detected (high risk: 0.85). The skill’s Mode 1 (Extract) and Mode 3 (Reference) explicitly use WebFetch to scrape user-supplied URLs at runtime, and the scraped page text is then analyzed and fed into the agent’s LLM context—this is outsider-authored public web content.

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

• Potentially malicious external URL detected (high risk: 0.80). This skill fetches user-provided URLs (scraped at runtime via WebFetch/Read) and directly uses the fetched page/image content to synthesize the "Visual Brand Style" and produce the Reference Prompts used by /image-gen, so remote content can directly control the agent's prompts.