marketing-content
Warn
Audited by Snyk on Mar 13, 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 Mode explicitly instructs the agent to use agent-browser (or WebFetch/WebSearch) to open and scrape public URLs including Google search results, AnswerThePublic, competitor blogs, Reddit, and Quora as part of its required workflow, meaning it ingests untrusted, user-generated third-party content that can directly influence recommendations and tool usage.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata