Code Quality

Unified skill for Rust code quality and clean code development. Use scripts as primary interface.

Quick Commands

# Format check
./scripts/code-quality.sh fmt

# Lint with clippy
./scripts/code-quality.sh clippy --workspace

# Security audit
./scripts/code-quality.sh audit

# Run all quality gates
./scripts/code-quality.sh check

# Full quality gates (coverage threshold 90%)
./scripts/quality-gates.sh

Quality Gates

Check	Command	Target
Format	./scripts/code-quality.sh fmt	100% compliant
Lint	./scripts/code-quality.sh clippy --workspace	Zero warnings
Audit	cargo audit	No known vulnerabilities
Coverage	cargo llvm-cov --html	>=90%
Docs	cargo doc --no-deps	All public APIs

Rust Quality Dimensions

Dimension	Focus	Target
Structure	Files <500 LOC, clear modules	<500 LOC per file
Error Handling	Custom Error, Result, no unwrap	? operator, thiserror
Async Patterns	spawn_blocking for CPU work	No blocking in async
Testing	>=90% coverage, AAA pattern	cargo nextest + doctests
Documentation	Public APIs documented	cargo doc passes
Security	Parameterized SQL, env vars	No hardcoded secrets

Clean Code Principles

SOLID

• Single Responsibility: One reason to change per module
• Open-Closed: Extend via traits, not modification
• Liskov Substitution: Subtypes substitutable for base types
• Interface Segregation: Small, focused interfaces
• Dependency Inversion: Depend on abstractions, inject deps

DRY/KISS/YAGNI

• DRY: Extract common code, single source of truth
• KISS: Simple solutions, avoid over-engineering
• YAGNI: Build what's needed now, defer complexity

Anti-Patterns

Rust-Specific

• Excessive clone -> Use borrowing or Arc
• Unnecessary unwrap -> Use ? operator
• Deep nesting -> Extract methods
• Large functions -> Split (<50 LOC)
• Deadlocks -> Release locks before .await

General Code Smells

• Long methods -> Break into smaller
• Duplicate code -> Extract reusable functions
• God class -> Reduce responsibilities
• Feature envy -> Move methods to appropriate classes

Best Practices Checklist

DO

• Files <500 LOC, clear module hierarchy
• Custom Error enum with Result
• No unwrap() in production (tests only)
• async fn for IO, spawn_blocking for CPU

• =90% test coverage

• Public APIs documented
• SOLID principles applied
• No code duplication (DRY)

DON'T

• Violate SOLID principles
• Duplicate code "for now"
• Functions >50 lines without justification
• Skip static analysis warnings
• Over-engineer simple problems
• Create tight coupling

Code Review Output Format

# Code Quality Report

## Summary
- Score: X/100
- Critical Issues: N
- Warnings: M

## By Dimension
- Structure: [Status]
- Error Handling: [Status]
- Async Patterns: [Status]
- Testing: [Status]
- Documentation: [Status]

## Critical Issues
1. [Issue] - File:line - Fix: [Recommendation]

## Action Items
- [ ] High: Fix critical issues
- [ ] Medium: Address warnings

Dependency Monitoring

# Count duplicate dependencies (target: <100)
cargo tree -d | grep -cE "^[a-z]"

# Find unused dependencies
cargo machete && cargo shear

References

• Detailed dimensions: quality-dimensions.md (in this directory)
• ADR-036: Dependency Deduplication
• ADR-032: Disk Space Optimization