social-graph-ranker

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

• Third-party content exposure detected (high risk: 0.78). The skill’s runtime workflow ingests OUTSIDER-authored free text via the “Harvest Your Graph” step—e.g., LinkedIn CSV export or browser-use scraping of connections/followers (public/profile-derived text from other people) and via lead-intelligence’s enrichment agents (Exa + X API results that include third-party profile/content snippets), which are then used to generate personalized outreach drafts in the LLM context.