visual-style

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

• Third-party content exposure detected (medium risk: 0.30). The required Extract mode can ingest outsider-authored free text indirectly by fetching and analyzing arbitrary user-supplied URLs/pages to extract design information and generate style_prompt_full (e.g., “Analyze a website” where the URL content is processed at runtime).

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

• Potentially malicious external URL detected (high risk: 1.00). The "Extract from Website" extractor (references/extractors/from-website.md) explicitly fetches a user-provided website at runtime and uses the fetched HTML/CSS to build the style_prompt_full that is then used verbatim (example URL in the docs: https://stripe.com), so external content can directly control prompts.