はじめに
Strands AgentsのMCPサーバーが提供されていたので、Amazon Q Developer CLIに設定して、Strands Agentsのアプリを簡単に作成できるように設定してみました。
Amazon Q Developer CLI へのMCPサーバー設定
今回の環境
- OS:Ubuntu(WSL2)
- Amazon Q Developer: 1.12.2(MCP対応)
- uv:インストール済み
方法1:mcp.jsonを直接編集
公式の手順にもサンプルが置いてありますが、mcp.jsonに追加するだけで動きます。
プロジェクト固有の場合:/project/.amazonq/mcp.json
グローバル設定の場合:~/.aws/amazonq/mcp.json
{
"mcpServers": {
"strands-agents": {
"command": "uvx",
"args": [
"strands-agents-mcp-server"
]
}
}
}
方法2:Q Developer自体に依頼
Q Developerに設定自体を頼んでしまっても大丈夫でした。
q mcp addコマンドを実行してくれました。
> Strands AgentsのMCPサーバーをワークスペース設定で設定したい(uvx "strands-agents-mcp-server")
> Strands AgentsのMCPサーバーをワークスペース設定で追加しましょう。uvxを使用してStrands Agents MCPサーバーを設定します。
🛠️ Using tool: execute_bash (trusted)
⋮
● I will run the following shell command:
q mcp add --name "strands-agents" --command "uvx" --args "strands-agents-mcp-server" --scope workspace
⋮
↳ Purpose: Add Strands Agents MCP server to workspace configuration
📁 Created MCP config in '/home/xxxxxx/qdev/.amazonq/mcp.json'
To learn more about MCP safety, see https://docs.aws.amazon.com/amazonq/latest/qdeveloper-ug/command-line-mcp-security.html
✓ Added MCP server 'strands-agents' to 📄 workspace
⋮
● Completed in 1.389s
> 設定が完了しました!設定内容を確認してみましょう。
🛠️ Using tool: execute_bash (trusted)
⋮
● I will run the following shell command: q mcp list
⋮
↳ Purpose: List configured MCP servers
📄 workspace:
/home/xxxxxx/qdev/.amazonq/mcp.json
• strands-agents uvx
🌍 global:
/home/xxxxxx/.aws/amazonq/mcp.json
(empty)
チャットを再び開くと認識されます。
冒頭の部分でMCPサーバー(strands_agents)が読み込まれています。
$ q chat
✓ strands_agents loaded in 1.15 s
⢠⣶⣶⣦⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⢀⣤⣶⣿⣿⣿⣶⣦⡀⠀
⠀⠀⠀⣾⡿⢻⣿⡆⠀⠀⠀⢀⣄⡄⢀⣠⣤⣤⡀⢀⣠⣤⣤⡀⠀⠀⢀⣠⣤⣤⣤⣄⠀⠀⢀⣤⣤⣤⣤⣤⣤⡀⠀⠀⣀⣤⣤⣤⣀⠀⠀⠀⢠⣤⡀⣀⣤⣤⣄⡀⠀⠀⠀⠀⠀⠀⢠⣿⣿⠋⠀⠀⠀⠙⣿⣿⡆
⠀⠀⣼⣿⠇⠀⣿⣿⡄⠀⠀⢸⣿⣿⠛⠉⠻⣿⣿⠛⠉⠛⣿⣿⠀⠀⠘⠛⠉⠉⠻⣿⣧⠀⠈⠛⠛⠛⣻⣿⡿⠀⢀⣾⣿⠛⠉⠻⣿⣷⡀⠀⢸⣿⡟⠛⠉⢻⣿⣷⠀⠀⠀⠀⠀⠀⣼⣿⡏⠀⠀⠀⠀⠀⢸⣿⣿
⠀⢰⣿⣿⣤⣤⣼⣿⣷⠀⠀⢸⣿⣿⠀⠀⠀⣿⣿⠀⠀⠀⣿⣿⠀⠀⢀⣴⣶⣶⣶⣿⣿⠀⠀⠀⣠⣾⡿⠋⠀⠀⢸⣿⣿⠀⠀⠀⣿⣿⡇⠀⢸⣿⡇⠀⠀⢸⣿⣿⠀⠀⠀⠀⠀⠀⢹⣿⣇⠀⠀⠀⠀⠀⢸⣿⡿
⢀⣿⣿⠋⠉⠉⠉⢻⣿⣇⠀⢸⣿⣿⠀⠀⠀⣿⣿⠀⠀⠀⣿⣿⠀⠀⣿⣿⡀⠀⣠⣿⣿⠀⢀⣴⣿⣋⣀⣀⣀⡀⠘⣿⣿⣄⣀⣠⣿⣿⠃⠀⢸⣿⡇⠀⠀⢸⣿⣿⠀⠀⠀⠀⠀⠀⠈⢿⣿⣦⣀⣀⣀⣴⣿⡿⠃
⠚⠛⠋⠀⠀⠀⠀⠘⠛⠛⠀⠘⠛⠛⠀⠀⠀⠛⠛⠀⠀⠀⠛⠛⠀⠀⠙⠻⠿⠟⠋⠛⠛⠀⠘⠛⠛⠛⠛⠛⠛⠃⠀⠈⠛⠿⠿⠿⠛⠁⠀⠀⠘⠛⠃⠀⠀⠘⠛⠛⠀⠀⠀⠀⠀⠀⠀⠀⠙⠛⠿⢿⣿⣿⣋⠀⠀
⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠀⠈⠛⠿⢿⡧
╭─────────────────────────────── Did you know? ────────────────────────────────╮
│ │
│ Use /model to select the model to use for this conversation │
│ │
╰──────────────────────────────────────────────────────────────────────────────╯
/help all commands • ctrl + j new lines • ctrl + s fuzzy search
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
🤖 You are chatting with claude-4-sonnet
MCPサーバーの稼働確認
Strands Agents MCPサーバーが有効になったので、呼び出せるか試してみます。
> Strands Agentsを利用して簡単なアプリを作成して
> Strands Agentsを使って簡単なアプリを作成するお手伝いをします。まず、Strands Agentsの基本的な情報を確認させてください。
🛠️ Using tool: quickstart from mcp server strands_agents
⋮
● Running quickstart with the param:
⋮ {
⋮ "arguments": {},
⋮ "name": "quickstart"
Allow this action? Use 't' to trust (always allow) this tool for the session. [y/n/t]:
> t
⋮ }
⋮
● Completed in 0.18s
> それでは、Strands Agentsを使って簡単なアプリを作成しましょう。まず、プロジェクトの構造を作成します。
・
・
・
strands_agents MCPサーバーの quickstart をちゃんと呼び出せたようです。
次に、Strands Agentsの機能についても聞いてみました。
> Strands Agentsについて説明して
> Strands Agentsについて詳しく説明します。先ほど取得した情報を基に、主要な概念と使い方をご説明しますね。
## Strands Agentsとは
Strands AgentsはAIエージェントを構築するためのPython SDKです。エージェントは主に以下の3つの要素で定義されます:
1. モデル - 推論とツール使用をサポートするAIモデル
2. ツール - エージェントの機能を拡張する外部システムとの連携機能
3. プロンプト - システムプロンプトとユーザーメッセージ
## 主要な概念
### エージェント (Agent)
• モデル、ツール、プロンプトを組み合わせて定義
• 入力クエリとコンテキストに基づいて自動的にツールを使用
### モデル (Model)
• デフォルト:Amazon BedrockのAnthropic Claude Sonnet 3.7 (us-west-2リージョン)
• Amazon Bedrock、Anthropic、LiteLLM経由で多数のモデルをサポート
### ツール (Tools)
• エージェントの機能拡張の主要メカニズム
• 外部システムとの連携、データアクセス、環境操作が可能
• @toolデコレータでPython関数をツール化
• Model Context Protocol (MCP) サーバーもサポート
## パッケージ構成
strands-agents>=0.1.0 # メインSDK
strands-agents-tools>=0.1.0 # サンプルツール集
## 基本的な使用例
python
from strands import Agent, tool
from strands_tools import calculator, current_time, python_repl
・
・
・
ちゃんと理解してくれているようです。
これで書き方を確認してくれるので、Strands Agentsを使ったアプリ作成が簡単にできそうです!
参考URL