---

name: code-review description: Reviews code changes for bugs, style issues, and best practices. Use when looking for errors or improving code quality.

Code Review Skill

When reviewing code, systematically check the following areas:

1. Correctness & Logic

• Boundary Conditions: Are edge cases (empty lists, None values, zero division) handled?
• Error Handling: Are exceptions caught specifically? (No bare except:).
• Type Safety: Do function signatures match their usage?

2. Architecture & Design

• Single Responsibility: Does the function/class do one thing?
• Coupling: Are dependencies manageable? (e.g., UI code shouldn't import low-level processors directly if possible).
• Resource Management: Are files and GPU memory released properly? (Look for try...finally or context managers).

3. Style & Conventions

• Naming: Do names reflect intent? (e.g., pnm_adjacency vs adjacency).
• Docstrings: Are complex logic blocks explained?
• Imports: Are imports grouped and sorted?

4. Performance (Specific to CT/Volume Data)

• Vectorization: Is numpy/cupy used instead of explicit loops?
• Memory: Are large arrays copied unnecessarily?
• GPU Usage: Is generic try-import used for optional GPU backends?

Output Format

Provide feedback in a structured list:

• Severity: [Critical/Major/Minor]
• File: [Filename]
• Issue: [Description]
• Suggestion: [Proposed Fix]