research-parallel
Warn
Audited by Snyk on May 12, 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.md workflow explicitly assigns WebSearch/WebFetch (and optional Perplexity/Gemini) agents to fetch and ingest open web content—including community sources like StackOverflow/Reddit mentioned in the source-evaluation/decomposition examples—and then uses those fetched results to drive cross-validation and final report decisions, exposing the agent to untrusted third-party content.
Issues (1)
W011
MEDIUMThird-party content exposure detected (indirect prompt injection risk).
Audit Metadata