aesthetic-research

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

• Third-party content exposure detected (medium risk: 0.65). Step 3–4 performs runtime web image searches (Pinterest/Dribbble/Behance/Google Images) and then Step 5–6 ingests the resulting page/image context to create per-image descriptions and URLs, which can include outsider-authored free text (e.g., captions/metadata/comments) that may be fed into the agent’s LLM context during browsing.