dark-social-attributor

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

• Third-party content exposure detected (medium risk: 0.30). This skill ingests user-pasted GA4 landing-page/channel exports, user-provided share-surface/form inventories, and survey answers as untrusted runtime inputs (and also runs a connector script for Wikipedia attention), so outsider-authored free text can be posted into the workflow inputs/exports that the LLM reads without selecting a specific trusted item first.