claude-agent-sdk - Build autonomous AI agents with Claude Agent SDK
Build autonomous AI agents with structured outputs, plugins, and event-driven workflows
Tags
Updated: 2026-02-09Capabilities
Typical Inputs
Typical Outputs
What this skill does
- validate JSON schema
- define output structure
- load local plugins
- register hook events
- configure MCP servers
- orchestrate subagents
- manage sessions
- control permissions
- configure sandbox
- enable file checkpointing
Inputs
- prompt
- JSON schema
- Zod schema
- plugin paths
- MCP server configuration
- subagent definitions
- permission rules
- sandbox settings
Outputs
- structured JSON output
- validated messages
- tool execution logs
- session state
- file checkpoints
- permission decisions
- MCP tool results
Requirements
- @anthropic-ai/claude-agent-sdk package
- Node.js environment
- structured-outputs-2025-11-13 beta header
- TypeScript for Zod schemas
- MCP server availability