customer-research
Warn
Audited by Snyk on Jul 8, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.85). This skill’s required workflow includes “Mode 2: Go Find Research” where it instructs the agent to mine online communities/review sites (e.g., Reddit/G2/forums) and extract verbatim quotes, which are outsider-authored free text that would be read at runtime and fed into the LLM context.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata