social-media-intelligence

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

• Third-party content exposure detected (high risk: 0.90). The skill explicitly ingests and analyzes public, user-generated social media content (e.g., SocialPost.content and Claim.first_seen_url from platforms like Twitter/Reddit/Instagram listed in Platform) and even fetches arbitrary URLs via SocialArchiver.archive_to_wayback/archive_to_archive_today, so the agent would read untrusted third‑party content.