web-research
Warn
Audited by Snyk on Mar 7, 2026
Risk Level: MEDIUM
Full Analysis
MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).
- Third-party content exposure detected (high risk: 0.90). The skill's SKILL.md and README explicitly direct the agent to perform WebSearch/WebFetch and fetch details from public URLs (e.g., TechCrunch, Stack Overflow, Reddit are cited in examples) and to incorporate those third‑party findings into its decision-making and reports, so untrusted web content can be read and materially influence actions.
Audit Metadata