image-3d

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

• Third-party content exposure detected (high risk: 0.85). Yes: the required workflow accepts outsider-provided free text via the prompt argument (user input not authored by the operating user is treated as outsider content) and then injects it into the LLM-relevant context by building final_prompt in generate_3d.py and sending it as the prompt field to the external image model API.