ai-image-generation

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

• Third-party content exposure detected (high risk: 0.80). The skill's i2i routes (e.g., Schema fields image_urls / images in the Nano Banana / GPT Image 2 / FLUX Kontext examples) and the enable_web_search option cause the RunComfy model server to fetch arbitrary public HTTPS image/URL references (called out in "How it works" and the route examples), meaning untrusted third‑party content is ingested and can materially influence generation (image-based or web-grounded prompt injection).

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

• Potentially malicious external URL detected (high risk: 0.90). The skill's install instructions instruct running a remote installer ("curl -fsSL https://runcomfy.com/install.sh | sh"), which would execute fetched code and is a required dependency for the runcomfy CLI used by the skill at runtime.