book-media-generator

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

• Third-party content exposure detected (high risk: 0.78). SKILL.md directs the workflow into the “chapter images / verified infographic” paths that rely on reading outsider-authored text such as public web content (e.g., Wikimedia Commons/Gov archive material and quoted passages from web searches) and then using it in the runtime Claude verification steps (“verification is done by Claude in text”), which is then fed back into the agent context for further prompting.