nano-banana-pro-product-image

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

• Third-party content exposure detected (medium risk: 0.30). The runtime path scripts/imagegen.py accepts outsider-provided --prompt (and optional reference images) and sends it in the JSON body to the AI Hive endpoint (client.generate_image(... "prompt": prompt ...)), meaning the agent/system ingests free text authored by the workflow user at execution time.