content-research
Warn
Audited by Snyk on Feb 15, 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 explicitly instructs the agent to use WebSearch and scrape public third‑party sources (e.g., Twitter/X, LinkedIn, Reddit, Quora, news sites, top‑ranking articles and podcast transcripts) and to extract quotes and questions from them, which are untrusted user-generated/open web contents the agent is expected to read and interpret as part of its workflow.
Audit Metadata