conversion-copywriting
Warn
Audited by Snyk on Feb 13, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). The skill explicitly instructs mining and ingesting user-generated VOC from open/public sources—e.g., "Review mining (Amazon, G2, Capterra, Trustpilot)", "Social media comments and discussions", "Forum posts and Reddit threads"—and includes prompts where the agent is expected to read/interpret pasted customer reviews/VOC data, exposing it to untrusted third-party content.
Audit Metadata