Claude Code Launches Agent SDK for Building AI Agent Teams

Anthropic has released the Claude Agent SDK, enabling developers to build production-grade AI agents that work together as coordinated teams using Python and TypeScript.

The SDK, renamed from the Claude Code SDK, ships with built-in tools for file operations, terminal commands, code search, and web browsing — eliminating the need to build infrastructure from scratch. Developers can have agents reading files, fixing bugs, and editing code within minutes.

The standout feature is **Subagents**: a main agent can delegate specialized tasks to focused sub-agents, each with their own toolset and instructions. A code-reviewer agent can analyze quality while a bug-fixer agent patches issues simultaneously, with results flowing back to the orchestrating agent.

Key capabilities include **Model Context Protocol (MCP)** integration connecting to hundreds of external systems, **Hooks** for injecting custom logic at lifecycle events, **Sessions** for maintaining context across conversations, and granular **Permissions** controlling exactly which tools each agent can access.

The SDK supports deployment through Anthropic's API, Amazon Bedrock, Google Vertex AI, and Microsoft Azure, making it suitable for CI/CD pipelines and production automation at scale.

With example agents already available — from email assistants to research agents — the barrier to building multi-agent AI systems has dropped significantly.