export-ml-notebook

Pass

Audited by Gen Agent Trust Hub on Oct 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs file conversions, policy management, and site generation tasks using the skore_skills CLI tool.
  • [EXTERNAL_DOWNLOADS]: The skill manages the installation of standard data science and documentation packages, including jupytext, nbformat, nbconvert, ipywidgets, and mkdocs-material, through a controlled package management utility.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled Python scripts to generate derived notebook and HTML files, creating an ingestion surface.
  • Ingestion points: Python source files (.py) located in data_analysis/, experiments/, and audit/ directories.
  • Boundary markers: The skill does not implement specific delimiters or 'ignore instructions' warnings for the content of the processed scripts.
  • Capability inventory: Includes tool-assisted file writing, CLI command execution, and package installation.
  • Sanitization: No explicit sanitization or filtering of the script contents is described beyond the standard conversion process.
  • [DYNAMIC_EXECUTION]: During specific conversion modes, the skill dynamically injects a setup cell containing %matplotlib inline into the notebook to facilitate graphical output rendering.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 15, 2026, 02:42 AM