ecommerce-patent-references
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill retrieves and displays citation data from the external Zhihuiya database, which introduces a potential surface for indirect prompt injection if the source data contains instructions designed to override agent behavior. \n
- Ingestion points: External data is fetched from the patent database via the
scripts/patent_references.pyscript using a POST request. \n - Boundary markers: The skill instructions lack explicit boundary markers or directions for the agent to ignore instructions that might be embedded in the retrieved patent titles or literature descriptions. \n
- Capability inventory: The skill is capable of performing network requests to vendor-managed endpoints and writing JSON data to the local filesystem for session tracking and caching. \n
- Sanitization: While the script handles JSON structure, it does not include specific sanitization or filtering of the text content within the patent references before presentation to the agent context.
Audit Metadata