skill-development
Skill Development Rules
When & Why to Use This Skill
This Claude skill establishes a standardized framework for developing, documenting, and packaging agentic capabilities. It provides clear architectural guidelines for the .claude/skills/ directory, ensuring that all skills are concise, maintainable, and properly integrated with the Model Context Protocol (MCP) for optimal agent performance and reliability.
Use Cases
- Standardizing Skill Documentation: Implementing a consistent SKILL.md structure across a development team to ensure all agent capabilities are well-documented and easy to maintain.
- Secure MCP Integration: Guiding the creation of MCP wrapper skills by enforcing restricted tool access (e.g., limiting to Bash and Read) and proper script referencing.
- Development Quality Control: Using the 'DO' and 'DON'T' checklists to prevent common errors such as duplicate content, missing triggers in skill-rules.json, or bloated documentation.
- Technical Onboarding: Serving as a primary reference for developers to understand the specific requirements and best practices for building within a Claude-based agentic ecosystem.
| name | skill-development |
|---|---|
| description | Skill Development Rules |
| user-invocable | false |
Skill Development Rules
When working with files in .claude/skills/:
SKILL.md Structure
DO
- Keep SKILL.md concise (< 200 lines)
- Include clear "When to Use" section
- Provide copy-paste bash commands
- Reference scripts/ for MCP operations
- Add triggers to skill-rules.json
DON'T
- Include implementation details in SKILL.md
- Duplicate content across skills
- Create skills without corresponding trigger in skill-rules.json
- Use allowed-tools that aren't needed
MCP Wrapper Skills
For skills that wrap MCP scripts:
- Use
allowed-tools: [Bash, Read]to restrict capabilities - Point to the script in scripts/ directory
- Include parameter documentation