signal-detection-pipeline

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

• Third-party content exposure detected (high risk: 0.85). The required workflow explicitly runs outsider-sourced scraping/monitoring skills (e.g., luma-event-attendees, reddit-scraper, linkedin-post-research, linkedin-commenter-extractor) that fetch and ingest public/authored free text from third parties at runtime into the agent’s context for lead extraction and enrichment.