product-visual-generator

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

• Third-party content exposure detected (high risk: 0.90). The skill's Phase 1 explicitly scrapes arbitrary public brand websites (SKILL.md and scripts/generate-visuals.py's scrape_product_images) to produce products.json which Claude then reads in Phase 2 to write prompts, so untrusted third‑party page content can materially influence prompt generation and downstream actions.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill accepts and scrapes arbitrary user-supplied brand URLs at runtime (e.g., "https://brand.com"), downloads product images from those sites, uploads them to Fal storage, and passes the resulting image URLs into the model (fal-ai/nano-banana-2/edit), which injects externally fetched content into the model context and can directly influence generation outputs.