poster

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

• Third-party content exposure detected (medium risk: 0.30). In scripts/imagegen.py the outsider-authored --prompt free text is directly sent to the AI Hive generation/image endpoint via cmd_image() → client.generate_image(..., prompt=args.prompt, ...) (and similarly --system/--prompt in cmd_chat()), making it runtime-ingested user input.