trending-ad-hook-spotter

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

• Third-party content exposure detected (high risk: 0.95). SKILL.md Phase 1 requires scraping public social content at runtime (Twitter/X, Reddit, LinkedIn, Hacker News) and then using that collected text to generate trend summaries and ad hooks, so outsider-authored free text is ingested into the agent’s LLM context via the scraper outputs.