Claude Agent SDK Demos — 8 Official Example Projects

What it is

This is the official demo collection by Anthropic showcasing the Claude Agent SDK. It includes eight example projects: an email agent, a research agent, a resume generator, a chat application, an Excel processing agent, and more. Each demo demonstrates a different SDK pattern for building production AI agents.

The collection targets developers learning the Claude Agent SDK who want working reference implementations rather than documentation alone.

How it saves time or tokens

Starting from a working demo eliminates the blank-page problem. Each example includes the agent loop, tool definitions, error handling, and state management patterns that you would otherwise build from scratch. Copy the closest example, modify it for your use case, and deploy.

The demos also serve as integration tests for the SDK. They demonstrate patterns that are verified to work, reducing debugging time.

How to use

1. Clone the repository: git clone https://github.com/anthropics/anthropic-cookbook
2. Navigate to the agent SDK demos directory
3. Install dependencies: pip install anthropic
4. Run any demo: python email_agent.py

Example

# Simplified research agent pattern from the demos
import anthropic

client = anthropic.Anthropic()

def research_agent(query: str) -> str:
    tools = [
        {
            'name': 'web_search',
            'description': 'Search the web for information',
            'input_schema': {
                'type': 'object',
                'properties': {
                    'query': {'type': 'string', 'description': 'Search query'}
                },
                'required': ['query']
            }
        }
    ]

    messages = [{'role': 'user', 'content': query}]

    while True:
        response = client.messages.create(
            model='claude-sonnet-4-20250514',
            max_tokens=4096,
            tools=tools,
            messages=messages
        )
        if response.stop_reason == 'end_turn':
            return response.content[0].text
        # Handle tool use and continue the loop
        messages.append({'role': 'assistant', 'content': response.content})
        # Process tool calls...

Related on TokRepo

• AI agent tools -- Agent frameworks and SDKs
• Coding tools -- Developer tools and examples

Common pitfalls

• Demos use hardcoded API keys in some examples; replace with environment variables before deploying
• The agent loop pattern requires proper stop condition handling; infinite loops occur if tool results are not formatted correctly
• Some demos depend on external services (email, web search); configure those services before running