wechat-article

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION] (LOW): The skill instructs the agent to execute a Python script using a user-provided file path as an argument. This is a potential command injection vector if the agent does not properly sanitize the filename.
  • [EXTERNAL_DOWNLOADS] (SAFE): The skill requires standard Python packages (markdown, beautifulsoup4, pygments) from PyPI, which is a trusted registry.
  • [INDIRECT_PROMPT_INJECTION] (LOW): The skill processes untrusted Markdown data from the user. 1. Ingestion point: input.md file contents. 2. Boundary markers: Not specified in instructions. 3. Capability: Local script execution via subprocess. 4. Sanitization: None described in the provided files for the markdown content itself.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 06:07 PM