Skill Creation Guide

skill-creation-guide

Complete guide for creating new Claude Code skills. Templates, best practices, and activation patterns. Use when: creating new skills, documenting workflows, packaging expertise.

Metadata

• Version: 1.0.0
• Category: documentation
• Source: workspace

Tags

skills, templates, creation, guide, best-practices

MCP Dependencies

None specified

Inputs

• skillName (string) (required): Name for the new skill
• purpose (string) (required): What the skill should do
• category (string) (optional): Category: code-quality, git-workflow, database, api, documentation, testing, devops, productivity, analysis

Chain of Thought

1. Define skill purpose clearly. 2. Identify trigger conditions. 3. Write concise description. 4. Create input parameters. 5. Implement code logic. 6. Test activation.

Workflow

No workflow defined

Anti-Hallucination Rules

None specified

Verification Checklist

None specified

Usage

// Execute via MCP Gateway:
gateway_execute_skill({ name: "skill-creation-guide", inputs: { ... } })

// Or via REST API:
// POST /api/code/skills/skill-creation-guide/execute
// Body: { "inputs": { ... } }

Code

// Skill Creation Guide
const skillName = inputs.skillName;
const purpose = inputs.purpose;
const category = inputs.category || 'productivity';

console.log(\`## Skill Creation Guide: \${skillName}

### Purpose
\${purpose}

---

## Skill Anatomy

### 1. SKILL.md Structure (Claude Code Native)
\`\`\`markdown
---
name: \${skillName}
description: [Concise, trigger-focused description]
allowed-tools: [Read, Write, Edit, Bash, Grep, Glob]
---

# \${skillName}

## When to Use
[Specific trigger conditions]

## Instructions
[Step-by-step procedure]

## Examples
[Concrete examples]
\`\`\`

### 2. MCP Gateway Skill (JSON + Code)
\`\`\`json
{
  "name": "\${skillName}",
  "description": "[Description that triggers activation]",
  "category": "\${category}",
  "inputs": [
    {
      "name": "param1",
      "type": "string",
      "description": "What this parameter does",
      "required": true
    }
  ],
  "tags": ["tag1", "tag2"],
  "code": "// Implementation here"
}
\`\`\`

---

## Best Practices

### Description Writing (CRITICAL)
The description determines when skill activates:

❌ Bad: "Helps with code"
✅ Good: "Reviews TypeScript code for security vulnerabilities, 
   performance issues, and maintainability. Use when: code review 
   requested, PR review, security audit."

### Trigger Phrases
Include in description:
- "Use when: [specific situations]"
- "Activates for: [task types]"
- Keywords users will naturally use

### Input Parameters
- Required: Core inputs needed for skill
- Optional: Customization options with defaults
- Types: string, number, boolean, object, array

---

## Skill Categories

| Category | Use For |
|----------|---------|
| code-quality | Reviews, refactoring, linting |
| git-workflow | Commits, PRs, branching |
| database | Queries, migrations, schemas |
| api | Integrations, endpoints, webhooks |
| documentation | Docs, ADRs, READMEs |
| testing | Test generation, coverage |
| devops | CI/CD, deployment, infra |
| productivity | Workflows, automation |
| analysis | Research, exploration |

---

## Create Skill Command

\`\`\`typescript
gateway_create_skill({
  name: "\${skillName}",
  description: "\${purpose}. Use when: [triggers]",
  category: "\${category}",
  tags: ["relevant", "tags"],
  inputs: [/* parameters */],
  chainOfThought: "1. Step one. 2. Step two...",
  code: \\\`// Your implementation\\\`
});
\`\`\`

---

## Testing Activation

After creating, test with:
1. Direct invocation: \`gateway_execute_skill("\${skillName}", inputs)\`
2. Natural language: Ask Claude something matching description
3. Check if skill loads in context
\`);

Created: Tue Dec 23 2025 00:34:29 GMT+0800 (Singapore Standard Time) Updated: Tue Dec 23 2025 00:34:29 GMT+0800 (Singapore Standard Time)