Skill Developer

Skill Developer Guide

Purpose

Comprehensive guide for creating and managing Claude Code skills, hooks, agents, and commands. This meta-skill helps you build and configure Claude Code infrastructure components.

When to Use This Skill

• Creating new skills
• Configuring skill triggers in skill-rules.json
• Setting up hooks
• Creating agents
• Adding slash commands
• Understanding skill activation patterns
• Customizing skill behavior

Quick Start

Creating a New Skill

1. Create skill directory:

   bashCopy code
   mkdir -p .claude/skills/{skill-name}
   

2. Create SKILL.md file:

   • Add frontmatter with name and description
   • Write comprehensive guidelines
   • Use progressive disclosure (main file + resources/)

3. Update skill-rules.json:

   • Add skill entry with triggers
   • Configure fileTriggers and promptTriggers
   • Set enforcement and priority

Skill Structure

Copy code
.claude/skills/
  {skill-name}/
    SKILL.md              # Main skill file
    resources/            # Optional: Additional resources
      pattern-1.md
      pattern-2.md

Skill Configuration

skill-rules.json Structure

jsonCopy code
{
  "version": "1.0",
  "description": "Skill activation triggers",
  "skills": {
    "{skill-name}": {
      "type": "domain" | "guardrail",
      "enforcement": "suggest" | "block" | "warn",
      "priority": "critical" | "high" | "medium" | "low",
      "description": "Skill description",
      "promptTriggers": {
        "keywords": ["keyword1", "keyword2"],
        "intentPatterns": ["regex pattern"]
      },
      "fileTriggers": {
        "pathPatterns": ["app/**/*.tsx"],
        "pathExclusions": ["**/*.test.tsx"],
        "contentPatterns": ["import.*from.*next"]
      }
    }
  }
}

Enforcement Types

• suggest: Skill suggestion appears but doesn't block execution
• block: Requires skill to be used before proceeding (guardrail)
• warn: Shows warning but allows proceeding

Priority Levels

• critical: Highest - Always trigger when matched
• high: Important - Trigger for most matches
• medium: Moderate - Trigger for clear matches
• low: Optional - Trigger only for explicit matches

File Triggers

Path Patterns

Use glob patterns to match file paths:

jsonCopy code
{
  "pathPatterns": [
    "app/**/*.tsx", // All .tsx files in app/
    "components/**/*.ts", // All .ts files in components/
    "**/*.tsx" // All .tsx files anywhere
  ]
}

Path Exclusions

Exclude files from triggering:

jsonCopy code
{
  "pathExclusions": [
    "**/*.test.tsx", // Test files
    "**/node_modules/**", // Dependencies
    "**/.next/**" // Build output
  ]
}

Content Patterns

Match file content with regex:

jsonCopy code
{
  "contentPatterns": [
    "from '@/components/ui/", // Shadcn imports
    "import.*from.*next", // Next.js imports
    "'use client'" // Client component directive
  ]
}

Prompt Triggers

Keywords

Simple keyword matching:

jsonCopy code
{
  "keywords": ["component", "page", "route", "frontend"]
}

Intent Patterns

Regex patterns for flexible matching:

jsonCopy code
{
  "intentPatterns": [
    "(create|add|make|build).*?component", // Create component
    "(how to|best practice).*?react", // How to questions
    "app router.*?(page|route)" // App router related
  ]
}

Skill Types

Domain Skills

• Purpose: Provide guidelines for specific domains
• Example: frontend-dev-guidelines, backend-dev-guidelines
• Enforcement: Usually "suggest"

Guardrail Skills

• Purpose: Enforce best practices and prevent mistakes
• Example: Code quality checks, security rules
• Enforcement: Usually "block" or "warn"

Best Practices

Skill Design

1. Progressive Disclosure: Main file + resources/ for detailed guides
2. Clear Examples: Include working code examples
3. Quick Reference: Add quick reference tables
4. When to Use: Clearly state when skill applies

Trigger Configuration

1. Specific Keywords: Use domain-specific terms
2. Flexible Patterns: Use regex for intent matching
3. Path Specificity: Match actual project structure
4. Avoid Over-triggering: Use exclusions appropriately

File Organization

1. Modular Structure: Split large skills into resources/
2. Clear Naming: Use descriptive skill names
3. Documentation: Document all configuration options

Common Patterns

Tech Stack Specific Skills

jsonCopy code
{
  "frontend-dev-guidelines": {
    "fileTriggers": {
      "pathPatterns": ["app/**/*.tsx", "components/**/*.tsx"],
      "contentPatterns": ["from '@/components/ui/", "import.*from.*next"]
    },
    "promptTriggers": {
      "keywords": ["component", "shadcn", "next.js"],
      "intentPatterns": ["(create|build).*?component"]
    }
  }
}

Framework Agnostic Skills

jsonCopy code
{
  "error-tracking": {
    "fileTriggers": {
      "pathPatterns": ["**/*Controller.ts", "**/*Service.ts"],
      "contentPatterns": ["Sentry\\.", "captureException"]
    },
    "promptTriggers": {
      "keywords": ["error", "sentry", "exception"],
      "intentPatterns": ["(add|implement).*?error.*?handling"]
    }
  }
}

Integration Checklist

When adding a new skill:

• Create skill directory and SKILL.md
• Write comprehensive guidelines
• Add to skill-rules.json
• Configure fileTriggers (pathPatterns, exclusions, contentPatterns)
• Configure promptTriggers (keywords, intentPatterns)
• Set appropriate enforcement and priority
• Test skill activation
• Document customization needs

Troubleshooting

Skill Not Triggering

1. Check pathPatterns match actual file paths
2. Verify keywords are spelled correctly
3. Test intentPatterns regex patterns
4. Check for pathExclusions blocking triggers

Over-triggering

1. Add more specific pathPatterns
2. Use pathExclusions to filter out files
3. Make intentPatterns more specific
4. Lower priority level

Skill File Not Found

1. Verify skill directory exists: .claude/skills/{skill-name}/
2. Check SKILL.md file exists
3. Verify skill name matches skill-rules.json entry

Related Skills

• frontend-dev-guidelines: Frontend development patterns
• backend-dev-guidelines: Backend development patterns

Skill Status: Meta-skill for skill development and management