What does the cookbook-audit skill specifically evaluate?

It evaluates notebooks across four dimensions: Narrative Quality, Code Quality, Technical Accuracy, and Actionability, providing a standardized score out of 20.

How does it handle large Jupyter Notebook files?

The skill generates a markdown version of the notebook in a temporary folder, removing extraneous outputs to provide a cleaner context for the AI to review.

What are the technical requirements for running an audit?

The skill requires a style_guide.md and a validation script (validate_notebook.py) to be present in the environment to perform the full automated and manual workflow.

Is this skill only for Anthropic Cookbooks?

While it is optimized for the Anthropic style guide, the underlying principles of problem-focused introductions and clean setup patterns are applicable to any high-quality technical tutorial.

Does it check for security vulnerabilities?

Yes, it integrates with automated scripts to run detect-secrets, catching hardcoded API keys and credentials before they are committed.

Cookbook Quality Audit

Name: Cookbook Quality Audit
Author: Microck

byMicrock

•

학습 및 문서화

Audits and scores Jupyter notebooks against specialized style guides and technical standards for AI documentation.

This skill provides a comprehensive framework for reviewing and improving AI-focused Jupyter notebooks, specifically optimized for the Anthropic Cookbook style. It automates technical checks—including secret detection and model constant validation—while guiding Claude through a qualitative assessment of pedagogical quality, narrative structure, and code best practices. By converting notebooks to clean markdown and applying a structured rubric, it helps developers and technical writers ensure their educational content is secure, actionable, and adheres to the Diataxis framework.

주요 기능

01Checklist-driven evaluation of setup patterns and model usage

02Markdown conversion for efficient, context-aware code review

03Qualitative narrative scoring based on pedagogical best practices

04Automated technical validation and secret scanning using detect-secrets

05Structured audit reports with executive summaries and prioritized fixes

0681 GitHub stars

사용 사례

01Ensuring internal data science notebooks follow consistent style guidelines

02Pre-publication security scanning for hardcoded API keys in tutorials

03Reviewing community contributions to technical documentation repositories

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add microck/ordinary-claude-skills cookbook-audit

For use in Claude.ai and ChatGPT

Download Skill