review

Review for Standards Compliance

Review a notebook or Python script against all datascience, notebook structure, and rich output standards. Produces an actionable checklist of violations.

What to check

Read the target file, then evaluate it in two parts.

1. Notebook standards (if .ipynb or Jupytext .py)

Read notebook-standards/SKILL.md and tick its Post-write checklist against the file - structure, output, long runs, evidence. That checklist is the single source of truth for those boxes; do not restate or fork it here, or the two copies drift apart and a reviewer enforces the stale one.

2. Project conventions (every file)

From the datascience:datascience skill - what the notebook checklist does not cover:

• Naming: NN-initials-description.ipynb pattern
• DataFrame naming: purpose_df / purpose_lf
• Polars lazy (scan_* / LazyFrame) for large data
• Sklearn builtins preferred over custom implementations
• No imports in __init__.py
• Google docstrings with type hints
• No hardcoded paths - Path objects
• Consistent figure sizes across the file
• Rich uses standard colour names, no hex; [dim] for secondary text

Output

Report as:

## Review: <filename>

### Passing (N/M)
- [x] Item that passes

### Violations (N)
- [ ] Item that fails - **what to fix**

### Suggestions
- Optional improvements (not violations)