Claude Agents

Official collection of agent patterns and recipes from Anthropic. Includes tool use, multi-agent orchestration, memory management, and real-world agent implementations.

A multi-tool agent that analyzes financial data, queries databases, generates charts, and produces structured investment research reports using Claude.

Anthropic's quickstart for building a production-ready customer support agent with Claude. Includes conversation management, tool use for order lookups, and escalation logic.

Official MCP SDK for building Claude tool integrations. Enables agents to connect Claude to any external system — databases, APIs, filesystems, and custom tools.

Official Anthropic GitHub Action that runs Claude Code as a CI agent. Automatically reviews PRs, fixes bugs, and implements features directly in your GitHub workflow.

Anthropic's reference patterns for orchestrating multiple Claude agents. Covers parallel agents, sequential pipelines, hierarchical agents, and error recovery strategies.

Anthropic's reference implementation of a computer-use agent. Claude controls a virtual desktop — browsing the web, running commands, and interacting with GUI applications.

The official Python SDK for building Claude-powered agents. Provides a clean API for multi-turn conversations, tool definitions, streaming, and structured outputs.

The official TypeScript/Node.js SDK for Claude. Build agents with type-safe tool definitions, streaming responses, and comprehensive error handling.

Deploy Claude agents on AWS using Amazon Bedrock. Includes examples for knowledge base integration, action groups, and enterprise-scale agent deployments.