keyword-research
Warn
Audited by Snyk on Feb 19, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 1.00). The SKILL.md Research Protocol and Community Mining sections explicitly instruct the agent to fetch and extract content from untrusted, user-generated third‑party sources (e.g., Reddit, Quora, YouTube comments, Amazon reviews, and Google SERPs/People Also Ask) and to use those findings to prioritize topics and make content decisions, so external content can materially influence the agent's actions.
Audit Metadata