patent-family-explorer

Pass

Audited by Gen Agent Trust Hub on Jun 27, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [PROMPT_INJECTION]: The skill is subject to indirect prompt injection through data ingested from the external patent database.
  • Ingestion points: Patent family details and legal status updates are retrieved from the Zhihuiya API in scripts/patent_family_explorer.py.
  • Boundary markers: The retrieved data is interpolated into the final markdown report without explicit delimiters or instructions to the agent to ignore potential commands embedded in the patent titles or abstracts.
  • Capability inventory: The skill is configured with Bash, Read, and Write tool permissions in SKILL.md, which could be targeted by an injection attack.
  • Sanitization: The script performs JSON parsing but does not sanitize or escape the content of the API response before presenting it to the agent.
  • [EXTERNAL_DOWNLOADS]: The skill performs network requests to external endpoints for its primary functionality.
  • It uses the NEXSCOPE_PROXY_BASE environment variable to define the target host for API calls to the Zhihuiya patent database.
  • These requests are used to fetch family data and legal status, which is the intended purpose of the skill.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 27, 2026, 11:06 AM
Security Audit — agent-trust-hub — patent-family-explorer