design-url-to-code

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

• Third-party content exposure detected (high risk: 0.85). The required workflow explicitly has the agent “capture the page” by opening a user-supplied Source URL(s) and pulling HTML/CSS/computed styles and full-page screenshots via browser automation/web-fetch, which can ingest outsider-authored free text from arbitrary third-party webpages into the LLM context (e.g., extracted DOM text/content used during extraction/rebuild).