restaurant-photos

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

• Third-party content exposure detected (high risk: 0.85). The skill’s runtime workflow explicitly requires “capture” and “audit” of existing photos from outsider-authored sources like Google Business Profile, Yelp, Instagram, and delivery platforms; those platforms’ user-generated photo captions/metadata and any extracted page content are read as free-form text/prose by the agent during inventory/audit.