ccf-visual-composer

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

• Third-party content exposure detected (medium risk: 0.30). The runtime workflow for architecture diagrams extracts and ingests “supplied paper/method content” from user-authorized project sources to build the content-derived GPT Image 2 prompt, and that prompt text is then sent to an external image-generation model, creating a path for outsider free text included in the supplied method/content to influence the model output.