cookbook-audit

To audit a Jupyter notebook against a style guide, place the file in the Skill path and run the validation script using nbconvert to generate a structured Markdown quality report. This reviews introduction, prerequisites, core content, and conclusion.

A notebook quality assessment rubric evaluates the introduction, prerequisites, core chapters, and conclusion. It checks for hardcoded keys, suppressed pip install outputs, explanatory code text, and proper model constant definitions to generate actionable improvement suggestions.

Yes, you can automatically check notebooks for hardcoded secrets and formatting issues. The automated review probes for hardcoded keys, verifies pip install output suppression, and validates code explanation text against the style guide criteria.

You generate a YAML metadata block from a notebook review by running the validation script. It processes the notebook against the rubric to produce concise YAML output containing dependencies, components, and toxicity evaluation ready for ingestion.

Yes, you need uv and nbconvert to run the notebook audit script. The process executes via uv run with the nbconvert package to validate the notebook and output the Markdown summary report.