anthropics/claude-agent-sdk-demos

What it solves

This repository provides a collection of practical examples to help developers learn how to use the Claude Agent SDK. It bridges the gap between documentation and implementation by showing how to build various AI-powered applications, from simple chat interfaces to complex multi-agent systems.

How it works

The project consists of several independent demo applications that utilize the Claude Agent SDK to interact with Claude. These demos demonstrate different SDK features, such as session persistence, streaming responses over WebSockets, and the use of specialized tools. Some demos implement advanced patterns like multi-agent coordination where a main agent spawns parallel sub-agents to perform research tasks.

Who it’s for

Developers who are using or planning to use the Claude Agent SDK to build AI agents and applications.

Highlights

• Multi-Agent Research: A system that breaks research requests into subtopics and coordinates parallel researcher agents to generate reports.
• Agentic Email Assistant: An IMAP-integrated assistant capable of searching and managing an inbox.
• Visual Previews: A branding assistant that renders user options as HTML preview cards instead of plain text.
• Resume Generation: An automated tool that searches the web for a person's professional profile and generates a .docx resume.
• Session Management: Examples of multi-turn conversations and session persistence using the V2 Session API.