ecommerce-zhihuiya-patent-image-search

Pass

Audited by Gen Agent Trust Hub on Sep 11, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data retrieved from a patent database and displays it to the agent.
  • Ingestion points: External patent information, including titles, abstracts, and descriptions, are retrieved via the /zhihuiya/patentImageSearch API and processed by scripts/zhihuiya_patent_image_search.py.
  • Boundary markers: The skill lacks explicit instructions to ignore embedded commands within the retrieved patent data, although it does append a mandatory legal disclaimer to the final output.
  • Capability inventory: The skill maintains capabilities for network requests (via urllib) and local file writing for session logging and caching.
  • Sanitization: There is no evidence of sanitization or escaping of the retrieved patent text before it is interpolated into the agent's response.
  • [COMMAND_EXECUTION]: The skill requires the execution of Python scripts (scripts/zhihuiya_patent_image_search.py and scripts/upload_image.py) to function.
  • The scripts perform network operations to interact with the Nexscope API at NEXSCOPE_PROXY_BASE and skill-api.nexscope.com.
  • scripts/zhihuiya_patent_image_search.py creates a local directory named nexscope in the current working directory or user home folder to manage cache files and session logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 11, 2026, 07:19 AM
Security Audit — agent-trust-hub — ecommerce-zhihuiya-patent-image-search