Audits Jupyter Notebook tutorials against Anthropic's documentation standards and technical style guidelines.
The cookbook-audit skill provides a rigorous framework for reviewing and improving Anthropic Cookbook notebooks. It automates the detection of hardcoded secrets, validates notebook structure against canonical templates, and evaluates narrative flow using the Diataxis framework. By applying a standardized rubric, it ensures that tutorials are problem-focused, action-oriented, and technically sound, helping developers transform raw code into high-quality educational content that builds user agency and transferable AI skills.
Key Features
01Standardized scoring across narrative quality, code quality, and technical accuracy
02Markdown-based review workflow to analyze code without distracting outputs
03Detailed report generation with prioritized, actionable recommendations
041 GitHub stars
05Automated secret detection and credential scanning using detect-secrets
06Structure validation for Terminal Learning Objectives (TLOs) and Enabling Learning Objectives (ELOs)
Use Cases
01Auditing technical documentation for consistent style and proper model usage patterns
02Reviewing community-submitted notebooks for the Anthropic Cookbook repository
03Refactoring existing AI tutorials to improve educational value and clarity
