video-outreach

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

• Third-party content exposure detected (high risk: 0.75). The skill’s runtime workflow includes “Pull their website, LinkedIn profile, product page, and recent posts” (Step 1) and “Screenshot the reference material you will show on screen,” which means it may ingest outsider-authored free text from public web/LinkedIn pages into the agent/LLM context for summarization and script creation.