ai-graphic-design

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

• Third-party content exposure detected (high risk: 0.70). The skill's required workflow and prompt-engineering guidance (see SKILL.md Section 3.2: Midjourney --sref [url] / --cref [url], Section 5 and 3.4 on training Style IDs and LoRAs with brand images, and the Phase 2/3 workflow) explicitly instruct ingesting external image URLs and user/third-party image assets which the agent would use to drive generation decisions, so untrusted third‑party content can materially influence tool behavior.