majiayu000 Anti-Pattern Analysis Skill
Version: 0.17.0
Purpose
Structured guidance for identifying anti-patterns across languages, architectures, and testing. Supports code reviews, refactoring planning, technical debt assessment, and reverse-PRD workflows.
When to Invoke
- Code review sessions
- Refactoring planning
- Technical debt assessment
- Architecture review
- Reverse-PRD extraction (document as tech debt/NFRs)
Anti-Pattern Categories
1. Design/OOP
| Pattern | Description | Severity |
|---|---|---|
| God Object | Class >500 lines or >10 methods | High |
| Singleton Abuse | Overuse creating global state | Medium |
| Anemic Domain Model | Data classes with no behavior | Medium |
| Circular Dependency | Classes depend on each other | High |
| Yo-yo Problem | Deep inheritance requiring navigation | Medium |
2. Code Smells
| Pattern | Description | Severity |
|---|---|---|
| Long Method | >20-30 lines | Medium |
| Deep Nesting | >3 levels indentation | Medium |
| Magic Numbers | Unexplained literals | Low |
| Feature Envy | Method uses another class's data excessively | Medium |
| Shotgun Surgery | One change requires many edits | High |
| Divergent Change | One class changed for many reasons | High |
3. Architecture
| Pattern | Description | Severity |
|---|---|---|
| Big Ball of Mud | No discernible architecture | Critical |
| Distributed Monolith | Microservices with tight coupling | High |
| Copy-Paste Programming | Duplicated code blocks | High |
| Spaghetti Code | Tangled control flow | High |
| Lasagna Code | Too many abstraction layers | Medium |
4. Database
| Pattern | Description | Severity |
|---|---|---|
| N+1 Queries | Loop executing individual queries | Critical |
| **SELECT *** | Fetching all columns unnecessarily | Medium |
| EAV Schema | Flexible but unqueryable | High |
| No Indexes | Missing indexes on queried columns | High |
| God Table | Table with too many columns | High |
5. Testing
| Pattern | Description | Severity |
|---|---|---|
| Flaky Tests | Non-deterministic results | Critical |
| Test Interdependence | Tests depend on order | High |
| Over-Mocking | Mocking makes test meaningless | High |
| Testing Implementation | Tests break on refactor | Medium |
| Happy Path Only | No edge case testing | Medium |
6. Security
| Pattern | Description | Severity |
|---|---|---|
| Hardcoded Secrets | Credentials in source | Critical |
| SQL Concatenation | String-built queries | Critical |
| Trust All Input | No input validation | Critical |
| Rolling Own Crypto | Custom cryptography | Critical |
| Missing Authentication | Endpoints without auth | Critical |
Severity Levels
| Level | Impact | Action |
|---|---|---|
| Critical | Security/data loss risk | Must fix before merge |
| High | Major tech debt | Fix in same PR/sprint |
| Medium | Code smell | Create follow-up issue |
| Low | Minor improvement | Optional |
Quick Review Checklist
Design: No God objects, no circular deps, SRP followed Code: Methods <30 lines, nesting <4 levels, no magic numbers Database: No N+1, no SELECT *, indexes exist, parameterized queries Testing: Deterministic, independent, reasonable mocking Security: No hardcoded secrets, input validated, auth on endpoints
Refactoring Guidance
- God Object → Extract to focused services
- Long Method → Extract method
- N+1 Query → Eager loading/prefetch
- Magic Numbers → Named constants
- Deep Nesting → Guard clauses
Reverse-PRD Integration
Detection maps to NFR categories:
- Security → NFR-SEC-*
- Performance → NFR-PERF-*
- Maintainability → NFR-MAINT-*
- Testing → NFR-QUAL-*
- Architecture → NFR-ARCH-*
End of Skill Document