patent-portfolio-report
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes patent data fetched via vendor-specific tools (e.g.,
patsnap_fetch), which represents a surface for indirect prompt injection if external patent documents contain adversarial instructions intended to influence the agent. \n - Ingestion points: Patent full-text content retrieved through the
patsnap_fetchtool (SKILL.md). \n - Boundary markers: The skill provides structural report templates but lacks explicit instructions for the agent to ignore or isolate commands embedded within the fetched patent text. \n
- Capability inventory: Capabilities are limited to data retrieval, technical analysis, and the generation of HTML reports; no arbitrary shell execution or unauthorized network access was detected (SKILL.md). \n
- Sanitization: The instructions do not specify sanitization or escaping protocols for the retrieved content prior to its inclusion in the HTML report output.
Audit Metadata