machine-learning

Pass

Audited by Gen Agent Trust Hub on Jun 23, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill is designed for data science and machine learning tasks on Snowflake, including data analysis, model training, and deployment. All identified patterns are standard for the intended machine learning use case.
  • [EXTERNAL_DOWNLOADS]: The skill provides guidance on setting up Snowflake External Access Integrations (EAIs) to access well-known services such as Hugging Face, Weights & Biases, and PyPI for package management and model downloads. These references are to trusted organizations and well-known technology services.
  • [COMMAND_EXECUTION]: It facilitates the execution of user-provided Python scripts on Snowflake compute pools and distributed Ray clusters. These operations are core functionalities of the Snowflake ML platform and are managed through the official snowflake-ml-python library.
  • [DATA_EXFILTRATION]: Data movement is restricted to user-defined Snowflake stages, tables, and authenticated external services. The skill emphasizes explicit user confirmation for object creation and the use of secure session management via a provided helper script (snowpark_session.py).
  • [PROMPT_INJECTION]: No malicious prompt injection or behavior override patterns were detected in the skill instructions. The skill focuses on operational guidance for data science tasks.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 23, 2026, 02:22 AM
Security Audit — agent-trust-hub — machine-learning