sketch

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly enables grounding via Google Image Search and reference-based editing using externally sourced images (see "Grounding with Google Image Search" in references/api-integration.md and the SKILL.md grounding guidance), which means the agent will fetch and incorporate untrusted public third-party image content that can influence generation decisions.