timesfm-forecasting

Pass

Audited by Gen Agent Trust Hub on May 15, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill facilitates the download of pre-trained model weights from official Google Research repositories on HuggingFace (e.g., google/timesfm-2.5-200m-pytorch). These downloads are essential for the skill's functionality and utilize a trusted service provider.
  • [COMMAND_EXECUTION]: The script scripts/check_system.py executes hardcoded system commands like sysctl and vm_stat on macOS to retrieve hardware specifications such as total and available RAM. These operations are restricted to non-sensitive system metadata used for environmental verification.
  • [SAFE]: Dependency verification in scripts/check_system.py uses dynamic imports with hardcoded package names ('timesfm', 'torch') to check for library presence, posing no risk of unauthorized code execution.
  • [SAFE]: The skill implements secure data handling by casting user-supplied CSV data to float32 arrays before inputting them into the numerical forecasting engine, effectively neutralizing any potential instructions or logic embedded in input files.
Audit Metadata
Risk Level
SAFE
Analyzed
May 15, 2026, 02:48 AM