NewSkill

Prompt file imported from klintravis/CopilotCustomizer (.github/prompts/NewSkill.prompt.md). Fill in {{skillName}}, {{purpose}}, {{targetPath}}, {{includeExamples}}, {{includeScripts}} before use. Copyright stays with the author.

New Agent Skill Generator

Creates Agent Skills (agentskills.io open standard) that work across VS Code, GitHub Copilot CLI, Claude, Cursor, and other AI platforms.

Parameters

• Skill Name [REQUIRED]: {{skillName}}
• Purpose [OPTIONAL]: {{purpose}}
• Target Path [OPTIONAL]: {{targetPath}}
• Include Examples [OPTIONAL]: {{includeExamples}}
• Include Scripts [OPTIONAL]: {{includeScripts}}

Usage Examples

# Basic skill generation
/NewSkill SKILL_NAME: "webapp-testing", PURPOSE: "Automated Playwright testing"

# With scripts and resources
/NewSkill SKILL_NAME: "api-debugging", PURPOSE: "Debug REST API issues", INCLUDE_SCRIPTS: true

# Target specific repository
/NewSkill SKILL_NAME: "deployment-automation", PURPOSE: "Deploy to AWS with validation", TARGET_PATH: "/path/to/project"

What Gets Created

TARGET_PATH/.github/skills/
└── [SKILL_NAME]/
    ├── SKILL.md                    # Skill definition with instructions
    ├── examples/                   # Example files (if INCLUDE_EXAMPLES=true)
    │   └── example-01.md
    ├── scripts/                    # Helper scripts (if INCLUDE_SCRIPTS=true)
    │   └── helper.sh
    └── resources/                  # Documentation and templates
        └── README.md

Workflow

1. Validate Parameters

   • Verify SKILL_NAME format (lowercase-with-hyphens)
   • Confirm PURPOSE is specific and clear
   • Check TARGET_PATH exists

2. Analyze Context

   • Review existing skills in repository
   • Identify tech stack and frameworks
   • Determine appropriate skill scope

3. Generate Skill Structure

   • Create .github/skills/[SKILL_NAME]/ directory
   • Generate SKILL.md with:
     • YAML frontmatter (name, description)
     • Purpose and when-to-use sections
     • Step-by-step instructions
     • Examples and resources references

4. Add Supporting Files (if requested)

   • Create examples/ with sample files
   • Generate scripts/ with helper utilities
   • Add resources/ with documentation

5. Integration

   • Update AGENTS.md with skill reference
   • Suggest related agents/instructions integration
   • Provide usage examples

6. Validation

   • Verify YAML frontmatter format
   • Check name length (≤64 chars)
   • Confirm description length (≤1024 chars)
   • Test skill discovery and loading

Refinement Commands

Skill Generation Guidelines

Name Format

✅ Good: webapp-testing, api-debugging, github-actions-debug ❌ Bad: WebAppTesting, api_debugging, TESTING

Instruction Reference

See SkillAuthoring.instructions.md for complete template and structure guidelines.

Description Format

description: |
  [What it does] for [target domain]. 
  Use when [scenario]. 
  Provides [key capabilities].

Example: "Automates Playwright test creation and execution for web applications. Use when building or debugging end-to-end tests."

Content Structure

See SkillAuthoring.instructions.md for complete template and structure guidelines.

Tech Stack Awareness

Generate skills based on detected technologies:

Skill Templates & Examples

See SkillAuthoring.instructions.md for comprehensive templates covering:

• Testing skills (unit/integration/e2e)
• Debugging skills (domain-specific troubleshooting)
• Code generation skills (components/patterns)
• Deployment skills (automation/validation)

Cross-Platform Verification

After generation, verify skill works across platforms:

1. VS Code

   • Enable chat.useAgentSkills setting
   • Confirm skill appears in discovery
   • Test automatic loading on relevant prompts

2. GitHub Copilot CLI

   • Verify skill is recognized
   • Test capability invocation

3. Portability Check

   • No VS Code-specific tool references
   • Relative paths only (no absolute paths)
   • Platform-agnostic instructions

Integration with Existing Assets

• Link to Instructions: Reference related instruction files for standards and patterns
• Reference in Agents: Update agent descriptions to mention new skill capabilities
• Combine with Prompts: Leverage skills in prompt workflows for enhanced functionality
• Update AGENTS.md: Document skill in architecture inventory

Output Format

After generation, provide:

1. Summary

   • Skill name and purpose
   • Location: .github/skills/[name]/
   • Files created

2. Usage Guidance

   The skill will automatically activate when you:
   - [Trigger scenario 1]
   - [Trigger scenario 2]
   - [Trigger scenario 3]
   

3. Integration Points

   • Related agents to update
   • Instructions to reference
   • AGENTS.md updates needed

4. Verification Steps

   • Enable chat.useAgentSkills in VS Code
   • Test skill discovery
   • Verify automatic loading
   • Check resource access

Best Practices

1. Focused Scope: One skill = one specialized capability
2. Clear Activation: Description triggers on specific prompts
3. Cross-Platform: Avoid platform-specific dependencies
4. Progressive Disclosure: Keep resources in separate files
5. Maintainable: Clear structure, well-documented

See SkillAuthoring.instructions.md for detailed quality criteria and validation steps.

Anti-Patterns to Avoid

❌ Generic skills: "Helper utilities" (too broad) ❌ VS Code-only: Skills that require terminal/file tools ❌ Instruction duplication: Just copying coding standards ❌ Absolute paths: Use relative paths only ❌ Massive SKILL.md: Move examples/scripts to separate files

Troubleshooting

Standards Compliance

Format: Agent Skills (agentskills.io) VS Code Support: v1.109+ (GA - enabled by default) File Structure: .github/skills/[name]/SKILL.md Frontmatter: YAML with required fields (name, description)

Generated: 2026-01-11 | Framework: CopilotCustomizer Purpose: Cross-platform AI skill generation