product-feature-patent-finder
Pass
Audited by Gen Agent Trust Hub on Aug 27, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted content from the internet through web research tools.\n
- Ingestion points: Data retrieved via
web.searchandweb_fetchfrom product documentation, official websites, and technical whitepapers.\n - Boundary markers: No specific delimiters or instructions are used to isolate external technical data from the agent's reasoning instructions.\n
- Capability inventory: The skill possesses file-writing capabilities (
files.begin_write,files.append) and network search tools.\n - Sanitization: The skill relies on AI-based extraction of technical dimensions but does not specify technical sanitization or validation of the fetched web data.\n- [DYNAMIC_EXECUTION]: The skill generates interactive HTML reports containing embedded JavaScript code.\n
- Evidence: The instructions in Step 3 of SKILL.md direct the agent to implement a "pure JS" dimension filter within the self-contained HTML report to allow users to toggle visibility of patent entries.\n- [EXTERNAL_DOWNLOADS]: The skill references and links to legitimate vendor-owned platforms for patent analysis.\n
- Evidence: Links to
eureka.zhihuiya.comfor patent viewing and references toopen.zhihuiya.comfor tool configuration are official resources provided by the skill author, Patsnap (Zhihuiya), and are appropriate for the skill's functional purpose.
Audit Metadata