NYC

china-stock-analysis

Warn

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION] (MEDIUM): The skill executes shell commands to run scripts in the 'scripts/' directory with parameters derived directly from user input (e.g., '--code', '--scope', '--industry'). This creates a risk of command injection. Additionally, the scripts themselves are missing from the provided files, making it impossible to verify their safety.
  • [EXTERNAL_DOWNLOADS] (MEDIUM): The skill prompts the user to install 'akshare', a third-party library from a non-whitelisted source, along with 'pandas' and 'numpy'.
  • [PROMPT_INJECTION] (LOW): The skill demonstrates a surface for indirect prompt injection.
  • Ingestion points: User input for stock codes, screening criteria, and external financial data retrieved via the 'akshare' API.
  • Boundary markers: No delimiters or 'ignore' instructions are used to isolate untrusted data.
  • Capability inventory: Execution of multiple Python scripts through shell commands with variable interpolation.
  • Sanitization: No sanitization or validation logic is defined for user-supplied strings before they are incorporated into executable commands.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 17, 2026, 04:38 PM