patent-avoidance-design
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external patent documents provided by users, which serves as an ingestion point for untrusted data.
- Ingestion points: User-uploaded patent PDFs or patent numbers.
- Boundary markers: Not explicitly defined in the prompt templates, but the highly structured TRIZ methodology (8-step functional modeling) acts as a natural constraint.
- Capability inventory: File reading (patents), SVG generation, and HTML report generation via sub-skills.
- Sanitization: Relies on the standard processing of the patent analysis tools.
- [DYNAMIC_EXECUTION]: The skill generates SVG functional models and interactive HTML reports using the
html-pptsub-skill. - Evidence: The skill instructions specify the generation of
CN113371388B_功能模型图.svgand a corresponding HTML report. - Context: This is a core functional requirement for visualizing patent analysis results and does not involve executing arbitrary user-provided code.
- [EXTERNAL_DOWNLOADS]: The skill references the Patsnap Open Platform MCP service.
- Evidence: Instructions mention requirement for account authorization and tool activation at
open.zhihuiya.com. - Context: This is a vendor-owned infrastructure reference for legitimate API and tool integration.
Audit Metadata