generate-image

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

• Third-party content exposure detected (high risk: 0.80). The skill explicitly supports web grounding and image search grounding (see scripts/generate.py --grounding / --image-grounding flags and the SKILL.md "Google Search Grounding" / references/capability-patterns), which cause the agent to fetch public web/image search results and include them as context for generation, so untrusted third‑party content can influence model behavior.