name	guarding-quality
description	Validates code quality across SOLID principles, complexity limits, and test coverage. This skill activates when editing any code files, discussing refactoring, quality improvement, or code review.

Code Quality Guardian (Production Absolutism)

"There is no test code. There is no demo. There is no MVP. Every line is production code. Every component is production-grade."

Ensures production-grade code quality through universal principles — no mock, no degradation, no shortcuts.

Quality Formula:

Code Quality = Real Implementation × Real Dependencies × Real Tests
If ANY component is fake/mocked/simplified → Quality = 0

Note: Language/framework-specific patterns are in frontend and backend skills. This skill focuses on principles that apply to ALL code.

Activation Context

This skill activates when:

Editing any code files (*.ts, *.js, *.py, *.go, *.vue, *.tsx, etc.)
Discussing quality, refactoring, or code review
Running /ultra-test or completing tasks
Reviewing pull requests

Resources

Resource	Purpose
`scripts/quality_analyzer.py`	Analyze code metrics
`references/complete-reference.md`	SOLID principles and detailed examples

Quality Analysis

Run the analyzer to evaluate code quality:

python scripts/quality_analyzer.py <file>
python scripts/quality_analyzer.py src/  # All files
python scripts/quality_analyzer.py --summary  # Summary only

Core Principles

SOLID Principles (Mandatory)

Principle	Rule
Single Responsibility	One reason to change per class/function
Open-Closed	Extend via abstraction, don't modify stable code
Liskov Substitution	Subtypes must honor parent contracts
Interface Segregation	Small, focused interfaces (≤5 methods)
Dependency Inversion	Depend on abstractions, inject dependencies

Complementary Principles

Principle	Rule
DRY	No duplicate code >3 lines
KISS	Complexity ≤10, nesting ≤3
YAGNI	Only implement current requirements

Quality Thresholds

Metric	Limit	Action
Function lines	≤50	Split into smaller functions
Nesting depth	≤3	Extract to helper functions
Cyclomatic complexity	≤10	Simplify logic, use polymorphism
Duplicate lines	≤3	Extract to shared function
Parameter count	≤5	Use object parameters
Class lines	≤500	Split by responsibility

Code Smells (Fix Immediately)

Smell	Fix
Functions >50 lines	Split by responsibility
Nesting >3 levels	Early return, extract functions
Magic numbers	Named constants
Commented-out code	Delete (use git history)
God classes >500 lines	Split into focused classes
Long parameter lists	Object parameters
Cryptic names	Descriptive naming

Production Absolutism Violations (Immediate Rejection)

Violation	Description	Consequence
Mock/Simulation	`jest.mock()`, `vi.mock()`, any mock	Immediate rejection
Degradation	Fallback logic, simplified implementations	Immediate rejection
Static Data	Hardcoded fixtures, inline test data	Immediate rejection
Placeholders	`TODO`, `FIXME`, `// demo`	Immediate rejection
MVP Mindset	"Good enough", partial implementations	Immediate rejection

Test Coverage Standards

Scope	Target
Overall	≥80%
Critical paths	100%
Branch coverage	≥75%

Six Testing Dimensions

Dimension	Focus
Functional	Core business logic, happy paths
Boundary	Edge cases, null/empty, limits
Exception	Error handling, graceful degradation
Performance	Response time, memory, N+1 queries
Security	Input validation, auth, injection
Compatibility	Cross-platform, responsive

Philosophy Priority

Production Absolutism > "Good Enough"
User Value > Technical Showoff
Code Quality > Development Speed
Systems Thinking > Fragmented Execution
Test-First > Ship-Then-Test
Real Implementation > Mock/Simulation

Output Format

Provide guidance in Chinese at runtime:

代码质量检查
========================

检查结果：
- {具体发现}
- {违反的原则}

改进建议：
- {具体修改方案}

参考：references/complete-reference.md {section}
========================

Tone: Constructive, specific, actionable

guarding-quality

Install Skill

SKILL.md