Can I use this SDK with Amazon Bedrock or Google Vertex AI?

Yes, the SDK supports multiple providers. You can enable them by setting environment variables like CLAUDE_CODE_USE_BEDROCK=1 or CLAUDE_CODE_USE_VERTEX=1 and configuring the respective cloud credentials.

How do I enable file-based Skills in my SDK project?

You must explicitly set setting_sources=['user', 'project'] in your ClaudeAgentOptions. Without this configuration, the SDK will not discover Skills located in .claude/skills/ directories.

What are lifecycle hooks used for in AI agents?

Hooks allow you to execute custom logic at specific points in the agent's execution, such as intercepting and blocking dangerous commands in PreToolUse or logging activities in PostToolUse.

How do I define custom tools in the Claude Agent SDK?

You can define custom tools using the @tool decorator and then incorporate them into your agent via an in-process SDK MCP server using create_sdk_mcp_server().

What is the difference between query() and ClaudeSDKClient?

The query() function is designed for stateless, one-shot tasks like automation scripts, while ClaudeSDKClient supports stateful, multi-turn conversations where context is preserved across multiple interactions.

Claude Agent SDK (Python)

Name: Claude Agent SDK (Python)
Author: WalterSumbon

byWalterSumbon

•

Herramientas para Desarrolladores

Provides expert guidance and implementation patterns for building production-ready AI agents using the Claude Agent SDK in Python.

This skill serves as a comprehensive technical companion for developers building autonomous agents with the official Claude Agent SDK. It offers deep-dive documentation on interaction modes like stateless one-shot queries and stateful multi-turn conversations, alongside advanced architectural patterns such as subagent delegation, lifecycle hooks, and custom MCP tool integration. By centralizing best practices for session management, permission modes, and multi-cloud authentication (including AWS Bedrock and GCP Vertex), it enables the creation of robust, secure, and scalable AI-driven workflows within the Python ecosystem.

Características Principales

015 GitHub stars

02Orchestration guides for subagent delegation and specialized agent definitions.

03Cross-provider authentication support for Anthropic API, Bedrock, Vertex AI, and Azure Foundry.

04Lifecycle hook configurations for PreToolUse, PostToolUse, and Session events.

05Full coverage of interaction modes including stateless query() and stateful ClaudeSDKClient.

06Detailed implementation patterns for custom tools via in-process SDK MCP servers.

Casos de Uso

01Integrating proprietary business logic into Claude via custom-defined SDK tools and MCP servers.

02Building interactive terminal assistants with persistent conversation state and session memory.

03Developing automated code review pipelines that utilize specialized subagents for security and testing.

Características Principales

015 GitHub stars

02Orchestration guides for subagent delegation and specialized agent definitions.

03Cross-provider authentication support for Anthropic API, Bedrock, Vertex AI, and Azure Foundry.

04Lifecycle hook configurations for PreToolUse, PostToolUse, and Session events.

05Full coverage of interaction modes including stateless query() and stateful ClaudeSDKClient.

06Detailed implementation patterns for custom tools via in-process SDK MCP servers.

Casos de Uso

01Integrating proprietary business logic into Claude via custom-defined SDK tools and MCP servers.

02Building interactive terminal assistants with persistent conversation state and session memory.

03Developing automated code review pipelines that utilize specialized subagents for security and testing.