Facilitates the development and integration of Claude-powered agents using the official TypeScript SDK.
This skill provides comprehensive guidance for developers building with the Claude Agent SDK, offering deep expertise in core functions like query management, custom tool creation, and MCP server configuration. It simplifies complex tasks such as implementing permission hooks, managing file checkpointing for safety, and handling session persistence, ensuring your AI agents are robust, secure, and performant. Whether you are defining subagent patterns or configuring sandboxed environments, this skill delivers accurate API references and production-ready TypeScript examples.
主要功能
01Detailed guidance on Model Context Protocol (MCP) server implementation
021 GitHub stars
03Production-grade TypeScript code examples with full type support
04Security-focused advice on permission modes and sandbox configurations
05Comprehensive API reference for query(), tool(), and MCP functions
06Best practices for session management and file checkpointing/rewind
使用场景
01Implementing complex authorization hooks and permission logic for secure tool use
02Configuring stateful AI workflows with session resumption and file-based checkpoints
03Building custom AI agents with specialized tools and MCP server integrations
