Build Claude Code SkillCreate comprehensive Claude Code Skills through elicitation-driven development## InstructionsThis command orchestrates four specialized agents to build production-ready Claude Code Skills from user requirements. Follow this structured workflow:### Phase 1: Requirements ElicitationAgent: skill-elicitation-agent1. Activate Elicitation Agent - Launch the skill-elicitation-agent using the Task tool - Provide context: "The user wants to create a new Claude Code Skill" - Include any initial requirements or ideas the user has shared2. Elicitation Questions - The agent will ask 3-5 targeted questions to understand: - Purpose and scope of the skill - Complexity and structure requirements - Tool permissions needed - Context and references - Success criteria3. Specification Creation - Agent will create a comprehensive skill specification document - Includes: metadata, structure, instructions outline, code needs, examples, dependencies - Uses progressive disclosure strategy - Validates completeness before proceeding4. User Approval - Present the specification to the user - Confirm understanding and agreement - Make adjustments if needed - Get explicit approval to proceed to generation### Phase 2: Skill GenerationAgent: skill-generator-agent1. Activate Generator Agent - Launch the skill-generator-agent using the Task tool - Pass the approved specification document - Specify target location (personal, project, or plugin skill)2. Directory Structure Creation - Agent creates appropriate directory structure: skill-name/ Ôö£ÔöÇÔöÇ SKILL.md (required) Ôö£ÔöÇÔöÇ reference.md (if needed) Ôö£ÔöÇÔöÇ examples.md (if needed) Ôö£ÔöÇÔöÇ scripts/ (if needed) ÔööÔöÇÔöÇ templates/ (if needed) 3. SKILL.md Generation - Creates main skill file with proper frontmatter - Includes clear instructions and examples - Follows progressive disclosure principles - Links to supporting files4. Supporting Files Generation - Creates reference documentation - Generates example files - Writes scripts with proper error handling - Creates reusable templates5. Dependency Documentation - Documents all required packages - Provides installation instructions - Notes version requirements6. Quality Checks - Validates YAML frontmatter - Checks file structure - Verifies code syntax - Tests progressive disclosure### Phase 3: Validation and TestingAgent: skill-validator-agent1. Activate Validator Agent - Launch the skill-validator-agent using the Task tool - Provide path to generated skill - Request comprehensive validation2. YAML Validation - Check frontmatter syntax - Verify required fields - Validate optional fields - Test YAML parsing3. Description Analysis - Assess discoverability - Check trigger keywords - Verify clarity and completeness - Compare to best practices4. Structure Validation - Verify file organization - Check all references - Test script execution - Validate permissions5. Code Testing - Syntax validation - Security checks - Dependency verification - Error handling tests6. Integration Testing - Test skill loading - Verify triggering - Check execution flow - Validate outputs7. Validation Report - Generate comprehensive report - Score each category - List issues by severity - Provide actionable fixes8. Issue Resolution - If issues found, work with generator agent to fix - Re-validate after fixes - Iterate until validation passes### Phase 4: Documentation EnhancementAgent: skill-documenter-agent1. Activate Documenter Agent - Launch the skill-documenter-agent using the Task tool - Provide skill path and specification - Request comprehensive documentation2. SKILL.md Enhancement - Refine instructions for clarity - Add comprehensive examples - Include best practices - Create troubleshooting section3. Reference Documentation - Create detailed technical reference (if needed) - Document API and configuration - Provide advanced patterns - Include performance tuning tips4. Example Collection - Generate beginner to advanced examples - Include troubleshooting examples - Show integration patterns - Provide case studies5. README Creation - Create skill directory README (if distributing) - Document installation - Provide quick start - Link to full documentation6. Documentation Quality Check - Verify clarity and completeness - Test all code examples - Check organization - Validate accuracy### Phase 5: Final Delivery1. Generate Summary - List all files created - Document location (personal/project/plugin) - Provide usage instructions - Include test scenarios2. Installation Verification - Confirm skill is in correct location - Verify file permissions - Check dependencies documented - Test skill loading3. Usage Guide - Explain how to trigger the skill - Provide example prompts - Show expected behavior - Link to documentation4. Next Steps - Suggest testing approach - Recommend improvements - Explain maintenance - Note future enhancements## Agent Coordination### Sequential FlowUser Request Ôåôskill-elicitation-agent (Requirements) ÔåôUser Approval Ôåôskill-generator-agent (Creation) Ôåôskill-validator-agent (Testing) ÔåôFixes if needed (loop back to generator) Ôåôskill-documenter-agent (Enhancement) ÔåôFinal Delivery### Agent CommunicationBetween Agents:- Elicitation ÔåÆ Generator: Pass specification document- Generator ÔåÆ Validator: Pass skill location and files- Validator ÔåÆ Generator: Pass validation issues (if any)- Generator ÔåÆ Documenter: Pass skill for enhancement- Documenter ÔåÆ User: Final documentationWith User:- Get approval after elicitation- Confirm location preference (personal/project)- Review validation results- Approve final deliverable## Best Practices1. Always Start with Elicitation - Don't skip the question phase - Understand deeply before building - Get user approval on specification2. Use All Four Agents - Each has specialized expertise - Complete flow ensures quality - Don't shortcut the process3. Iterate Based on Validation - Fix issues immediately - Re-validate after changes - Don't proceed with failures4. Test Comprehensively - Manual trigger testing - Script execution testing - Integration testing - Real-world scenario testing5. Document Thoroughly - Clear instructions - Comprehensive examples - Troubleshooting guides - Best practices## Location Options### Personal Skills (~/.claude/skills/)Use for:- Individual workflows- Experimental skills- Personal preferences- Private tools### Project Skills (.claude/skills/)Use for:- Team-shared workflows- Project-specific expertise- Version-controlled skills- Collaborative tools### Plugin Skills (Plugin directory structure)Use for:- Distributable skills- Marketplace deployment- Public sharing- Bundled capabilities## Output FormatProvide the user with a comprehensive summary:­ƒÄë Skill Created Successfully!­ƒôï Skill: [Skill Name]­ƒôü Location: [path]­ƒôØ Type: [simple/multi-file/tool-restricted/code-execution]­ƒôä Files Created:Ô£à SKILL.md - Main skill instructionsÔ£à reference.md - Technical reference (if created)Ô£à examples.md - Comprehensive examples (if created)Ô£à scripts/[name].py - Helper scripts (if created)Ô£à README.md - Installation guide (if created)­ƒº¬ Validation: Ô£à PASSED (Score: X/10)­ƒôª Dependencies:[List if any, or "None"]­ƒÄ» Usage:Trigger this skill by saying:- "[Example trigger 1]"- "[Example trigger 2]"Or explicitly: "Use [skill-name] to [task]"Ô£¿ Test with:[Specific test scenario]­ƒôÜ Documentation:- See SKILL.md for instructions- See examples.md for comprehensive examples- See reference.md for technical details­ƒÜÇ Next Steps:1. Test the skill with provided scenarios2. Refine based on usage3. Share with team (if project skill)4. Consider adding more examples over time## TroubleshootingAgent Not Found: Ensure .claude/agents/skill-builder/ exists with all agent filesPermission Errors: Check file permissions with chmod +x scripts/*.pyYAML Errors: Validator will catch and report theseAgent Confusion: Use explicit agent names: "Launch skill-elicitation-agent"## Examples### Example 1: Simple Instruction-Only SkillRequest: "Create a skill for writing conventional commit messages"Flow:1. Elicitation asks about commit style, projects, examples2. Generates simple single-file SKILL.md3. Validates structure and description4. Documents with examples and best practices### Example 2: Multi-File Skill with ScriptsRequest: "Create a skill for PDF form filling"Flow:1. Elicitation asks about PDF types, operations, dependencies2. Generates SKILL.md + scripts/fill_form.py + FORMS.md3. Validates code execution and file structure4. Documents with comprehensive examples and API reference### Example 3: Tool-Restricted Read-Only SkillRequest: "Create a skill for security code analysis"Flow:1. Elicitation determines read-only requirement2. Generates SKILL.md with allowed-tools: Read, Grep, Glob3. Validates tool restrictions4. Documents security patterns and analysis techniques## Remember- Quality over speed - Take time for thorough elicitation- User involvement - Get approval at key stages- Comprehensive validation - Test everything- Excellent documentation - Make skills easy to use- Iterative improvement - Skills can evolve over timeThis command ensures every skill is production-ready, well-documented, and follows Claude Code best practices!