integrated-reasoning
Warn
Audited by Snyk on Feb 16, 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 advanced patterns explicitly instruct the agent to perform WebSearch/WebFetch on public sites (e.g., "WebSearch: '[technology] community feedback 2025'", "WebSearch: '[technology] reddit hackernews discussion'", and "WebFetch: https://react.dev/blog" in references/integrated-reasoning-advanced-patterns.md), which means the agent will ingest untrusted, user-generated or public third‑party content as part of its workflow and could be exposed to indirect prompt injection.
Audit Metadata