Automated Content Generation Pipeline Skill

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

• Third-party content exposure detected (high risk: 0.98). The runtime workflow scrapes outsider-authored free text from public platforms via Apify (e.g., TikTok/Instagram/Reddit post captions/titles/selftext) and then injects that text into the Claude prompt in extractHooks() via VIRAL POSTS: ${JSON.stringify(viralPosts...)}.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). The skill calls Apify actors at runtime (e.g., apify/tiktok-hashtag-scraper and apify/reddit-scraper) and uses start URLs such as https://www.reddit.com/r/Entrepreneur/ to scrape posts which are then injected verbatim into the Claude prompt (VIRAL POSTS: ...), so externally fetched content directly controls the model's instructions.