MCP HubMCP Hub
スキル一覧に戻る

create-skill

danielmiessler
更新日 Today
117 閲覧
14
1
14
GitHubで表示
メタaiautomationdesign

について

このスキルは、Kaiの個人AIインフラストラクチャ内で新しいスキルを作成、更新、構造化するための体系的なフレームワークを開発者に提供します。ユーザーがスキルの目的、起動トリガー、依存関係を定義することを支援し、Anthropicの標準とPAI固有のパターンの両方に準拠します。このワークフローは、単純な単一機能スキルからより複雑な実装まで、適切な構造を決定するのに役立ちます。

クイックインストール

Claude Code

推奨
プラグインコマンド推奨
/plugin add https://github.com/danielmiessler/PAIPlugin
Git クローン代替
git clone https://github.com/danielmiessler/PAIPlugin.git ~/.claude/skills/create-skill

このコマンドをClaude Codeにコピー&ペーストしてスキルをインストールします

ドキュメント

Create Skill - Skill Creation Framework

When to Activate This Skill

  • "Create a new skill for X"
  • "Build a skill that does Y"
  • "Add a skill for Z"
  • "Update/improve existing skill"
  • "Structure a skill properly"
  • User wants to extend Kai's capabilities

Core Skill Creation Workflow

Step 1: Understand the Purpose

Ask these questions:

  • What does this skill do? (Clear, specific purpose)
  • When should it activate? (Trigger conditions)
  • What tools/commands does it use? (Dependencies)
  • Is it simple or complex? (Determines structure)

Step 2: Choose Skill Type

Simple Skill (SKILL.md only):

  • Single focused capability
  • Minimal dependencies
  • Quick reference suffices
  • Examples: fabric-patterns, youtube-extraction

Complex Skill (SKILL.md + CLAUDE.md + supporting files):

  • Multi-step workflows
  • Extensive context needed
  • Multiple sub-components
  • Examples: development, website, consulting

Step 3: Create Directory Structure

# Simple skill
${PAI_DIR}/skills/[skill-name]/
└── SKILL.md

# Complex skill
${PAI_DIR}/skills/[skill-name]/
├── SKILL.md           # Quick reference
├── CLAUDE.md          # Full context
└── [subdirectories]/  # Supporting resources

Step 4: Write SKILL.md (Required)

Use this structure:

---
name: skill-name
description: Clear description of what skill does and when to use it. Should match activation triggers.
---

# Skill Name

## When to Activate This Skill
- Trigger condition 1
- Trigger condition 2
- User phrase examples

## [Main Content Sections]
- Core workflow
- Key commands
- Examples
- Best practices

## Supplementary Resources
For detailed context: `read ${PAI_DIR}/skills/[skill-name]/CLAUDE.md`

Step 5: Write CLAUDE.md (If Complex)

Include:

  • Comprehensive methodology
  • Detailed workflows
  • Component documentation
  • Advanced usage patterns
  • Integration instructions
  • Troubleshooting guides

Step 6: Add to Global Context

Update ${PAI_DIR}/global/KAI.md available_skills section to include the new skill so it shows up in the system prompt.

Step 7: Test the Skill

  1. Trigger it with natural language
  2. Verify it loads correctly
  3. Check all references work
  4. Validate against examples

Skill Naming Conventions

  • Lowercase with hyphens: create-skill, web-scraping
  • Descriptive, not generic: fabric-patterns not text-processing
  • Action or domain focused: ai-image-generation, chrome-devtools

Description Best Practices

Your description should:

  • Clearly state what the skill does
  • Include trigger phrases (e.g., "USE WHEN user says...")
  • Mention key tools/methods used
  • Be concise but complete (1-3 sentences)

Good examples:

  • "Multi-source comprehensive research using perplexity-researcher, claude-researcher, and gemini-researcher agents. Launches up to 10 parallel research agents for fast results. USE WHEN user says 'do research', 'research X', 'find information about'..."
  • "Chrome DevTools MCP for web application debugging, visual testing, and browser automation. The ONLY acceptable way to debug web apps - NEVER use curl, fetch, or wget."

Templates Available

  • simple-skill-template.md - For straightforward capabilities
  • complex-skill-template.md - For multi-component skills
  • skill-with-agents-template.md - For skills using sub-agents

Supplementary Resources

For complete guide with examples: read ${PAI_DIR}/skills/create-skill/CLAUDE.md For templates: ls ${PAI_DIR}/skills/create-skill/templates/

Key Principles

  1. Progressive disclosure: SKILL.md = quick reference, CLAUDE.md = deep dive
  2. Clear activation triggers: User should know when skill applies
  3. Executable instructions: Imperative/infinitive form (verb-first)
  4. Context inheritance: Skills inherit global context automatically
  5. No duplication: Reference global context, don't duplicate it
  6. Self-contained: Skill should work independently
  7. Discoverable: Description enables Kai to match user intent

GitHub リポジトリ

danielmiessler/PAIPlugin
パス: skills/create-skill

関連スキル

content-collections

メタ

This skill provides a production-tested setup for Content Collections, a TypeScript-first tool that transforms Markdown/MDX files into type-safe data collections with Zod validation. Use it when building blogs, documentation sites, or content-heavy Vite + React applications to ensure type safety and automatic content validation. It covers everything from Vite plugin configuration and MDX compilation to deployment optimization and schema validation.

スキルを見る

creating-opencode-plugins

メタ

This skill provides the structure and API specifications for creating OpenCode plugins that hook into 25+ event types like commands, files, and LSP operations. It offers implementation patterns for JavaScript/TypeScript modules that intercept and extend the AI assistant's lifecycle. Use it when you need to build event-driven plugins for monitoring, custom handling, or extending OpenCode's capabilities.

スキルを見る

evaluating-llms-harness

テスト

This Claude Skill runs the lm-evaluation-harness to benchmark LLMs across 60+ standardized academic tasks like MMLU and GSM8K. It's designed for developers to compare model quality, track training progress, or report academic results. The tool supports various backends including HuggingFace and vLLM models.

スキルを見る

sglang

メタ

SGLang is a high-performance LLM serving framework that specializes in fast, structured generation for JSON, regex, and agentic workflows using its RadixAttention prefix caching. It delivers significantly faster inference, especially for tasks with repeated prefixes, making it ideal for complex, structured outputs and multi-turn conversations. Choose SGLang over alternatives like vLLM when you need constrained decoding or are building applications with extensive prefix sharing.

スキルを見る