researching-best-practices
Warn
Audited by Snyk on Mar 9, 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). SKILL.md's Research & Discovery (Phase 2) explicitly instructs the agent to perform "Web research if needed" and names community/untrusted sources (e.g., Stack Overflow, Reddit, Twitter/X, GitHub discussions), so the agent is expected to fetch and interpret arbitrary third-party/user-generated content that can materially influence its recommendations.
Audit Metadata