Claude Code Skills Marketplace

A curated collection of skills for Claude Code - the AI-powered development assistant.

Claude Code Skills are specialized prompts and patterns that enhance code review, development assistance, and productivity.

Available Skills

Spring AOP Prompt

Version: 1.0.0 | Author: zml

Spring AOP detection and guidance for code review. Helps identify:

• AOP proxy behavior issues
• Self-invocation problems
• Transaction boundary issues
• Event/RPC/messaging integration patterns

Auto-triggers: @Transactional, @Aspect, @Async, @Cacheable

---

Safe Java Code

Version: 1.2.0 | Author: zml

Comprehensive Java code safety checker for both general applications and high-load financial systems. Detects:

• Distributed systems: Lock/transaction ordering, SAGA pattern, cache consistency
• Concurrency: synchronized blocks, ReentrantLock, ThreadLocal, ConcurrentHashMap
• Race conditions: Check-Then-Act, double-checked locking, unsafe publication
• Database safety: Unchecked updates, optimistic locking, N+1 queries
• Async execution: Thread pool shutdown, exception handling, context passing
• Security vulnerabilities: SQL injection, IDOR, input validation, resource leaks
• Basic code issues: NPE, array bounds, exception handling

Auto-triggers: synchronized, @Transactional, ExecutorService, redisLock, .get(

---

Skill Structure

Each skill follows this standard structure:

skill-name/
├── SKILL.md                     # Core skill logic (main entry point)
├── skill.yaml                   # Metadata and auto-trigger configuration
├── README.md                    # User-facing documentation
├── CLAUDE.md                    # Skill-specific guidance (optional)
└── details/                     # Progressive loading modules
    ├── module-name.md
    └── comprehensive-checklist.md

Using Skills

Skills are auto-loaded by Claude Code when:

1. File patterns match - e.g., **/service/**/*.java
2. Code triggers are found - e.g., @Transactional annotation
3. Annotation triggers are present - e.g., @Async
4. Manual invocation - Use the /skill command

Progressive Loading

Skills use progressive loading to minimize token usage:

SKILL.md (always loaded, ~90 lines)
├── details/module-1.md (loaded when triggers match)
├── details/module-2.md (loaded when triggers match)
└── details/comprehensive-checklist.md (loaded for full review)

Contributing

To add a new skill:

1. Create the directory: mkdir -p skills/new-skill/{details,references}
2. Create SKILL.md: Main entry point with core detection logic
3. Create skill.yaml: Metadata including auto-trigger and progressive loading config
4. Create detail modules: Organized by category under details/
5. Create README.md: User-facing documentation

See CLAUDE.md for detailed guidance.

License

See individual skill directories for license information.