【入門】Amazon Bedrock AgentCoreのCode Interpreterを動かしてみた

mkdir bedrock-agentcore-tutorial
cd bedrock-agentcore-tutorial

bedrock-agentcore-tutorial/
├── myenv/                    # Python仮想環境（自動生成）
├── hello.py                  # Hello World サンプル
└── strands_ci_agent.py      # Code Interpreter サンプル

# 1. 仮想環境を作成（任意のディレクトリ名を指定）
python3 -m venv myenv

# 2. 仮想環境を有効化
source myenv/bin/activate

# 3. パッケージをインストール
pip install boto3 bedrock-agentcore strands-agents strands-agents-tools

{
    "UserId": "AIDAXXXXXXXXXXXXX",
    "Account": "123456789012",
    "Arn": "arn:aws:iam::123456789012:user/your-user-name"
}

import boto3
import json

code_to_execute = """
print("Hello World!!!")
"""

client = boto3.client(
    "bedrock-agentcore",
    region_name="us-west-2",  # 使用するリージョンを指定
    endpoint_url="https://bedrock-agentcore.us-west-2.amazonaws.com"
)

session_id = client.start_code_interpreter_session(
    codeInterpreterIdentifier="aws.codeinterpreter.v1",
    name="my-code-session",
    sessionTimeoutSeconds=900
)["sessionId"]

execute_response = client.invoke_code_interpreter(
    codeInterpreterIdentifier="aws.codeinterpreter.v1",
    sessionId=session_id,
    name="executeCode",
    arguments={
        "language": "python",
        "code": code_to_execute
    }
)

# Extract and print the text output from the stream
for event in execute_response['stream']:
    if 'result' in event:
        result = event['result']
        if 'content' in result:
            for content_item in result['content']:
                if content_item['type'] == 'text':
                    print(content_item['text'])

# Don't forget to stop the session when you're done
client.stop_code_interpreter_session(
    codeInterpreterIdentifier="aws.codeinterpreter.v1",
    sessionId=session_id
)

python hello.py

Hello World!!!

import json
from strands import Agent, tool
from bedrock_agentcore.tools.code_interpreter_client import code_session
import asyncio

#Define the detailed system prompt for the assistant
SYSTEM_PROMPT = """You are a helpful AI assistant that validates all answers through code execution.

VALIDATION PRINCIPLES:
1. When making claims about code, algorithms, or calculations - write code to verify them
2. Use execute_python to test mathematical calculations, algorithms, and logic
3. Create test scripts to validate your understanding before giving answers
4. Always show your work with actual code execution
5. If uncertain, explicitly state limitations and validate what you can

APPROACH:
- If asked about a programming concept, implement it in code to demonstrate
- If asked for calculations, compute them programmatically AND show the code
- If implementing algorithms, include test cases to prove correctness
- Document your validation process for transparency
- The state is maintained between executions, so you can refer to previous results

TOOL AVAILABLE:
- execute_python: Run Python code and see output

RESPONSE FORMAT: The execute_python tool returns a JSON response with:
- sessionId: The code interpreter session ID
- id: Request ID
- isError: Boolean indicating if there was an error
- content: Array of content objects with type and text/data
- structuredContent: For code execution, includes stdout, stderr, exitCode, executionTime

For successful code execution, the output will be in content[0].text and also in structuredContent.stdout.
Check isError field to see if there was an error.

Be thorough, accurate, and always validate your answers when possible."""

#Define and configure the code interpreter tool 
@tool
def execute_python(code: str, description: str = "") -> str:
    """Execute Python code"""

    if description:
        code = f"# {description}\n{code}"

    #Print code to be executed
    print(f"\n Code: {code}")

    # Call the Invoke method and execute the generated code, within the initialized code interpreter session
    with code_session("<Region>") as code_client:
        response = code_client.invoke("executeCode", {
        "code": code,
        "language": "python",
        "clearContext": False
    })

    for event in response["stream"]:
        return json.dumps(event["result"])

#configure the strands agent including the tool(s)
agent=Agent(
        tools=[execute_python],
        system_prompt=SYSTEM_PROMPT,
        callback_handler=None)

query="Can all the planets in the solar system fit between the earth and moon?"

# Invoke the agent asynchcronously and stream the response
async def main():
    response_text = ""
    async for event in agent.stream_async(query):
        if "data" in event:
            # Stream text response
            chunk = event["data"]
            response_text += chunk
            print(chunk, end="")

asyncio.run(main())

query="Can all the planets in the solar system fit between the earth and moon?"

日本語訳：
太陽系のすべての惑星は、地球と月の間に収まりますか？

python strands_ci_agent.py

I'll calculate whether all the planets in our solar system can fit between Earth and the Moon by comparing their combined diameters to the Earth-Moon distance.
 Code: # Planet diameters in kilometers
planets = {
    'Mercury': 4879,
    'Venus': 12104,
    'Mars': 6792,
    'Jupiter': 142984,
    'Saturn': 120536,
    'Uranus': 51118,
    'Neptune': 49528
}

~省略~

Based on the known planetary diameters and Earth-Moon distance:

**Planet Diameters (km):**
- Mercury: 4,879 km
- Venus: 12,104 km  
- Mars: 6,792 km
- Jupiter: 142,984 km
- Saturn: 120,536 km
- Uranus: 51,118 km
- Neptune: 49,528 km

**Total diameter of all planets:** 387,941 km

**Earth-Moon distance:** 384,400 km (average)

**Answer: NO** - All the planets cannot fit between Earth and the Moon.

地球と月の間に太陽系のすべての惑星が収まるかどうかを、惑星の直径の合計と地球と月の距離を比較して計算してみます。

コード:
```python
# 惑星の直径（キロメートル単位）
planets = {
    'Mercury': 4879,
    'Venus': 12104,
    'Mars': 6792,
    'Jupiter': 142984,
    'Saturn': 120536,
    'Uranus': 51118,
    'Neptune': 49528
}

~省略~

既知の惑星の直径と地球と月の距離に基づくと：

**惑星の直径（km）：**
- 水星：4,879 km  
- 金星：12,104 km  
- 火星：6,792 km  
- 木星：142,984 km  
- 土星：120,536 km  
- 天王星：51,118 km  
- 海王星：49,528 km  

**すべての惑星の直径の合計：** 387,941 km

**地球と月の距離（平均）：** 384,400 km

**答え：いいえ** — すべての惑星は地球と月の間には収まりません。