review-korean-patent-claims-ip
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface as it processes untrusted user-provided patent applications. 1. Ingestion points: File reading instructions in the Five-step workflow (Step 1 and Step 3). 2. Boundary markers: Use of an evidence ledger and preservation of original Korean source text alongside translations. 3. Capability inventory: Integration with external patent search and briefing tools via MCP connectors. 4. Sanitization: Explicit instructions in the HTML report requirements section to escape all user, patent, and connector-derived text to prevent active content injection.
- [EXTERNAL_DOWNLOADS]: The skill utilizes the
advanced_patent_searchandpatent_briefingMCP connectors hosted on the author's official marketplace (open.patsnap.com). These resources are legitimate vendor-provided extensions required for the skill's documented functionality.
Audit Metadata