generic-fto-report

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests technical descriptions and features from user-provided .docx files and interpolates them into prompts for the AI07 (CC GPT) API to perform infringement analysis.
  • Ingestion points: scripts/run_generic_fto_report.py parses document content via the parse_docx function.
  • Boundary markers: Absent. User-controlled text is directly concatenated into the prompt string within the call_ai07 function in scripts/run_generic_fto_report.py without using delimiters like XML tags or triple backticks.
  • Capability inventory: The skill has file system access (read/write) and network access to the vendor's API.
  • Sanitization: Text is truncated to a specific length (e.g., [:220]), but no escaping or filtering of instruction-like phrases is performed.
  • [EXTERNAL_DOWNLOADS]: The skill makes network requests to connect.zhihuiya.com to fetch patent information and analysis results. This is the official API endpoint for the service provided by the skill author.
  • Evidence: scripts/zhihuiya_api.py defines the base URL as https://connect.zhihuiya.com and utilizes it for P070, P002, P018, and AI07 API interactions.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 01:22 AM