create-image-fal

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

• Third-party content exposure detected (medium risk: 0.30). At runtime, outsider-authored free text can be injected via the user-supplied --payload (e.g., prompt/image_urls) passed from scripts/gen_image.py into fal_generate()/media_proxy._fal_run() which POSTs the payload to the FAL proxy.

MEDIUM W009: Direct money access capability detected (payment gateways, crypto, banking).

• Direct money access detected (high risk: 1.00). The skill explicitly routes paid image-generation calls through a fal/media proxy that "bills the Ads agent" (mentions billing in the run example and Contract). The bundled media_proxy.py and the statement that paid calls "bill the Ads agent" indicate the skill will execute charges against an account (i.e., perform financial billing), which is direct financial execution capability.