sketch-design-from-reference

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

• Third-party content exposure detected (high risk: 0.85). Step 1 requires classifying and extracting from a user-provided reference URL/screenshot/code, and the workflow explicitly ingests readable page/source text (e.g., DOM/HTML, linked JS/CSS, copy, asset URLs) into the agent’s context, which is outsider-authored free text from public web content fetched at runtime.