generate-svg

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly accepts and ingests arbitrary public URLs for reference images and source images (see SKILL.md flags --references and --image and examples like "--references https://example.com/ref1.png" and "--image https://example.com/photo.jpg"), and those untrusted third-party images are used by the model to drive SVG generation/vectorization, so they can materially influence behavior and enable indirect prompt-injection.