claude-agent-sdk

Expert guide for building AI agents with the Claude Agent SDK (Python & TypeScript). Use this skill PROACTIVELY whenever: - Building or configuring AI agents with `claude_agent_sdk` or `@anthropic-ai/claude-agent-sdk` - Working with `query()`, `ClaudeAgentOptions`, `AgentDefinition`, hooks, sessions, subagents, MCP, or custom tools - User asks how to automate tasks, build pipelines, or orchestrate multi-agent workflows with Claude - Implementing permission controls, `canUseTool` callbacks, or tool restrictions - Connecting external services (databases, APIs, browsers) to a Claude agent via MCP - Debugging or optimizing agent behavior, streaming, sessions, or hooks - Any mention of "Claude agent", "agent SDK", "agentic loop", or autonomous Claude workflows Covers the full SDK: query API, built-in tools, hooks lifecycle, session management, subagents, MCP servers, custom in-process tools, permission modes, user input flows, and authentication.