reputation-sentiment
Warn
Audited by Snyk on Jul 22, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.95). Yes—Phase 1 instructs WebSearch and WebFetch to extract full review text from public third-party platforms (e.g., Google/Yelp/Trustpilot/Reddit) written by outsiders, which becomes readable prose in the LLM context for Phase 2-4 classification and quoting.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata