はじめに
年始に時間があったので、ClaudeCodeと Agent Skillsに触れたのでメモ。
参考手順
インストール環境
- Windows11 Pro 64bit
事前準備
- Windows版Nodeのインストール
- v24.12.0
- https://nodejs.org/ja/download
- PowerShellの権限変更
- 管理者で起動し、以下ポリシーを変更。
Set-ExecutionPolicy RemoteSigned
- 管理者で起動し、以下ポリシーを変更。
ClaudeCodeのインストール
- PowerShellで以下のコマンドを実行。
npm install -g @anthropic-ai/claude-code - バージョン確認
claude --version 2.0.76 (Claude Code)
ClaudeCodeの起動と初期設定
- PowerShellで、
Claudeコマンドを実行する - EnterKey を押下
Welcome to Claude Code v2.0.76
…………………………………………………………………………………………………………………………………………………………
* █████▓▓░
* ███▓░ ░░
░░░░░░ ███▓░
░░░ ░░░░░░░░░░ ███▓░
░░░░░░░░░░░░░░░░░░░ * ██▓░░ ▓
░▓▓███▓▓░
* ░░░░
░░░░░░░░
░░░░░░░░░░░░░░░░
█████████ *
██▄█████▄██ *
█████████ *
…………………█ █ █ █………………………………………………………………………………………………………………
Let's get started.
Choose the text style that looks best with your terminal
To change this later, run /theme
❯ 1. Dark mode ✔
2. Light mode
3. Dark mode (colorblind-friendly)
4. Light mode (colorblind-friendly)
5. Dark mode (ANSI colors only)
6. Light mode (ANSI colors only)
╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌╌
1 function greet() {
2 - console.log("Hello, World!");
2 + console.log("Hello, Claude!");
3 }
- Claudeのサブスクリプション設定。1を選択して、Enterを押す
Claude Code can be used with your Claude subscription or billed based on API usage through your Console account. Select login method: ❯ 1. Claude account with subscription · Pro, Max, Team, or Enterprise 2. Anthropic Console account · API usage billing
- ブラウザが起動するので、承認する
- ターミナルに戻り、以下が表示されれば成功
Logged in as hoge@fuga Login successful. Press Enter to continue…
- ターミナルの初期設定。Enterを押す。
Use Claude Code's terminal setup? For the optimal coding experience, enable the recommended settings for your terminal: Shift+Enter for newlines ❯ 1. Yes, use recommended settings 2. No, maybe later with /terminal-setup Enter to confirm · Esc to skip
- 以下のように表示されれば設定完了。
╭─── Claude Code v2.0.76 /model to try Opus 4.5 ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── > Try "write a test for <filepath>" ───────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────── ? for shortcuts
プロジェクト初期化
/initで初期化。CLAUDE.mdファイルを生成する。
Opened changes in Visual Studio Code ⧉ Save file to continue… Do you want to make this edit to CLAUDE.md? ❯ 1. Yes 2. Yes, allow all edits during this session (shift+tab) 3. Type here to tell Claude what to do differently
はじめての Agent Skills の作成
- ターミナルに以下を入力
> 動くAgent Slillsのサンプルを作成してください。
SLILL.mdなど作成される。- 最終成果物は以下。
Agent Skillsの動作サンプルを作成しました!以下の内容が含まれています:
作成したファイル
📚 ドキュメント
- README.md - Agent Skillsの完全な日本語チュートリアル
- CLAUDE.md - Claude Codeのための技術ガイド(更新済み)
🎯 Agent Skills
1. Simple Commit Helper(基本例)
.claude/skills/simple-commit-helper/SKILL.md
- 単一ファイル構造のシンプルなSkill
- Gitコミットメッセージを生成
- 使用方法: "Help me write a commit message" とClaude Codeに尋ねる
2. Anthropic API Reviewer(高度な例)
.claude/skills/anthropic-api-reviewer/
├── SKILL.md # メインSkill定義
├── REFERENCE.md # 詳細なAPIリファレンス
└── scripts/
└── find_api_usage.py # API使用検出スクリプト
- マルチファイル構造
- ツール制限(allowed-tools)の実装例
- Pythonヘルパースクリプト付き
- 使用方法: "Review my Anthropic API code" とClaude Codeに尋ねる
使い方
1. Claude Codeを再起動してSkillsを読み込む
2. 利用可能なSkillsを確認:
"What skills are available?"
3. Skillsを試す:
- Commit Helper: "Help me create a commit message"
- API Reviewer: "Review the hello_world.py file"
主な特徴
- ✅ 2つの完全に動作するAgent Skillsサンプル
- ✅ 基本から高度な構造まで実例を提供
- ✅ 日本語の詳細なドキュメント
- ✅ ツール制限とマルチファイル構造の実装例
- ✅ 実用的なユーティリティスクリプト
- ✅ テスト済み(find_api_usage.pyスクリプトが正常に動作)
これで、Agent Skillsの作成方法を学ぶための完全なチュートリアル環境が整いました!
作成された Skills の確認
> 使えるスキルを教えて下さい ● このリポジトリに作成したAgent Skillsは以下の2つです: 1. simple-commit-helper(シンプル・コミット・ヘルパー) 説明: ステージされたGit変更から明確で簡潔なコミットメッセージを生成します いつ使えるか: - コミットメッセージを作成したい時 - コミット内容をレビューしたい時 使用例: "コミットメッセージを書いて" "Help me create a commit message" "何をコミットするか確認したい" ファイル: .claude/skills/simple-commit-helper/SKILL.md --- 2. anthropic-api-reviewer(Anthropic APIレビュアー) 説明: Anthropic Claude APIを使用するPythonコードをレビューし、ベストプラクティス、エラーハンドリング、最適化をチェックします いつ使えるか: - Anthropic APIを使うコードをレビューしたい時 - APIの使い方が正しいか確認したい時 - ベストプラクティスに従っているかチェックしたい時 使用例: "hello_world.pyをレビューして" "Review my Anthropic API code" "このファイルのAPI使用をチェックして" ファイル: - .claude/skills/anthropic-api-reviewer/SKILL.md - .claude/skills/anthropic-api-reviewer/REFERENCE.md - .claude/skills/anthropic-api-reviewer/scripts/find_api_usage.py --- ⚠️ 重要な注意事項 これらのSkillsを使用するには: 1. Claude Codeを再起動する必要があります 2. 再起動後、Claudeが自動的にSkillsを読み込みます 3. 上記の使用例のような質問をすると、Claudeが該当するSkillの使用を提案します 確認方法 Claude Codeを再起動後、以下のように尋ねてください: "What skills are available?" すると、利用可能なすべてのSkillsがリストアップされます。
Agent Slills の利用
事前準備(Git)
- Git のブランチ、ステージング確認(まだ何もやってない)
> git status
On branch master
No commits yet
Untracked files:
(use "git add <file>..." to include in what will be committed)
.claude/
CLAUDE.md
README.md
hello_world.py
quickstart.py
- 1ファイルのみステージングする
> git add ./CLAUDE.md warning: in the working copy of 'CLAUDE.md', LF will be replaced by CRLF the next time Git touches it
コミットメッセージを考える Skills を使用する
> コミットメッセージを考えてください
● コミットメッセージを考えます。まず、リポジトリの状態を確認させてください。
● Bash(git status)
⎿ On branch master
No commits yet
… +11 lines (ctrl+o to expand)
● 現在の状態を確認しました。CLAUDE.mdだけがステージされていますが、このプロジェクト全体を見ると、他のファイルも一緒にコミットする方が適切だと思います 。
まず、ステージされたCLAUDE.mdの変更内容を確認します:
● Bash(git diff --staged)
⎿ diff --git a/CLAUDE.md b/CLAUDE.md
new file mode 100644
index 0000000..c856e72
… +78 lines (ctrl+o to expand)
● 提案
現在はCLAUDE.mdだけがステージされていますが、2つの選択肢があります:
オプション1: すべてのファイルを含む初期コミット(推奨)
すべてのファイルをステージしてまとめてコミットする場合:
git add .
コミットメッセージ:
Add Agent Skills tutorial with working examples
- Add CLAUDE.md with repository guidance for Claude Code
- Add README.md with comprehensive Japanese tutorial
- Include hello_world.py demonstrating basic Anthropic API usage
- Create two working Agent Skills examples:
- simple-commit-helper: Basic Skill for commit message generation
- anthropic-api-reviewer: Advanced Skill with multi-file structure
- Include API best practices reference and utility scripts
This provides a complete tutorial for learning Agent Skills development
with Claude Code.
オプション2: CLAUDE.mdだけをコミット
現在ステージされているCLAUDE.mdだけをコミットする場合:
コミットメッセージ:
Add CLAUDE.md with repository guidance
- Document environment setup and dependencies
- Describe Python API examples and Agent Skills structure
- Include guidance for Agent Skills development
- Provide API usage patterns and best practices
---
おすすめ: オプション1をお勧めします。これは初期コミットとして、プロジェクト全体の構造と目的を一度に記録できます。
どちらのオプションを選びますか?それとも、別の方法でファイルを分けてコミットしたいですか?
- オプション1にするが、コミットメッセージを日本語にしたい旨を追加。
- すると、勝手に
git add .を実行する- さらに日本語のコミットメッセージを作成する
- すると、勝手に
Bash command
git commit -m "Agent Skillsチュートリアルと動作サンプルを追加
- CLAUDE.mdを追加(Claude Code用のリポジトリガイダンス)
- README.mdを追加(包括的な日本語チュートリアル)
- hello_world.pyを追加(基本的なAnthropic API使用例)
- 2つの動作するAgent Skillsサンプルを作成:
- simple-commit-helper: コミットメッセージ生成用の基本Skill
- anthropic-api-reviewer: マルチファイル構造の高度なSkill
- APIベストプラクティスリファレンスとユーティリティスクリプトを含む
Claude CodeでAgent Skills開発を学ぶための完全なチュートリアルを提供
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>"
Commit with Japanese message
Do you want to proceed?
❯ 1. Yes
2. Yes, and don't ask again for similar commands in C:\vscode\agent_skills_tutorial
3. Type here to tell Claude what to do differently
- コミット完了メッセージが表示される
● ✅ コミットが完了しました!
コミット内容
- コミットハッシュ: 37dfe56
- ブランチ: master (初期コミット)
- ファイル数: 8ファイル
- 変更行数: 839行追加
コミットされたファイル
1. .claude/skills/anthropic-api-reviewer/REFERENCE.md
2. .claude/skills/anthropic-api-reviewer/SKILL.md
3. .claude/skills/anthropic-api-reviewer/scripts/find_api_usage.py
4. .claude/skills/simple-commit-helper/SKILL.md
5. CLAUDE.md
6. README.md
7. hello_world.py
8. quickstart.py
コミットメッセージ
Agent Skillsチュートリアルと動作サンプルを追加
- CLAUDE.mdを追加(Claude Code用のリポジトリガイダンス)
- README.mdを追加(包括的な日本語チュートリアル)
- hello_world.pyを追加(基本的なAnthropic API使用例)
- 2つの動作するAgent Skillsサンプルを作成:
- simple-commit-helper: コミットメッセージ生成用の基本Skill
- anthropic-api-reviewer: マルチファイル構造の高度なSkill
- APIベストプラクティスリファレンスとユーティリティスクリプトを含む
Claude CodeでAgent Skills開発を学ぶための完全なチュートリアルを提供
🤖 Generated with Claude Code
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
これでAgent Skillsチュートリアルプロジェクトの初期コミットが完了しました!
コミット内容確認
> git log
Date: Fri Jan 2 11:51:40 2026 +0900
Agent Skillsチュートリアルと動作サンプルを追加
- CLAUDE.mdを追加(Claude Code用のリポジトリガイダンス)
- README.mdを追加(包括的な日本語チュートリアル)
- hello_world.pyを追加(基本的なAnthropic API使用例)
- 2つの動作するAgent Skillsサンプルを作成:
- simple-commit-helper: コミットメッセージ生成用の基本Skill
- anthropic-api-reviewer: マルチファイル構造の高度なSkill
- APIベストプラクティスリファレンスとユーティリティスクリプトを含む
Claude CodeでAgent Skills開発を学ぶための完全なチュートリアルを提供
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
動かしてみた所感
やはり自分の手で動かすのは大事だと感じました。いきなり、 Agen Skills に取り組もうとしましたが、その前に ClaudeCode が使えないとそもそも前に進めない。順番に理解しないとダメですね。
ブログ執筆の Skills や、原稿執筆の Skills ができたら便利そうなので、次は自分の Skills 作成をやってみようと思います。