create-image

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

• Third-party content exposure detected (high risk: 0.80). The skill's required workflow (step 3) uses generate_image in "editing mode" with an image_url and the MCP tool reference includes load_image/upload_asset, which clearly downloads and ingests arbitrary external/user-provided images (untrusted third‑party content) that the agent will read and act on to determine subsequent generation/edit actions.