A VS Code extension that turns plain .chat.md files into editable AI chat sessions with MCP tool support across LLM providers.
anthropics/claude-agent-sdk-python
Python SDK for building agents on Claude Code, with query(), an interactive client, in-process custom tools and hooks.
- Stars
- 8,304
- Forks
- 1,363
- Open issues
- 571
- Last commit
- 2 days ago
What is it?
The Claude Agent SDK for Python lets Python applications drive Claude Code programmatically. Install it with pip and call query() to send a prompt and iterate over the streamed response messages, or use ClaudeSDKClient for bidirectional, multi-turn conversations.
The package bundles the Claude Code CLI, so no separate installation is needed by default, though a system-wide install or a specific CLI path can be configured. By default Claude has access to the full Claude Code toolset such as Read, Write, Edit and Bash, and allowed_tools, disallowed_tools and permission_mode control approvals.
With ClaudeSDKClient you can define custom tools as Python functions that run as in-process MCP servers, and register hooks that run at points in the agent loop, such as blocking a Bash command before it executes. Typed options, message classes and error classes are provided for handling results.
Key features
- 01Stream Claude Code responses from an async query() function.
- 02Run multi-turn, interactive sessions with ClaudeSDKClient.
- 03Define custom tools as in-process MCP servers using a Python decorator.
- 04Register Python hooks that run at points in the agent loop.
- 05Control tool permissions with allowed and disallowed tool lists and permission modes.
Use cases
- Embedding a Claude coding agent inside a Python application or internal tool.
- Adding domain-specific Python functions as tools that Claude can call.
- Enforcing policies such as blocking certain shell commands with hooks.
- Building automated workflows that run Claude Code against a project directory.
Quick start
- 1
Install the SDK from PyPI
pip install claude-agent-sdk - 2
Save the README quick start code as a Python file and run it with anyio
Things to know
- Use is governed by Anthropic's Commercial Terms of Service, not the MIT license alone.
- Requires Python 3.10 or later.
- Querying Claude requires access to Anthropic's service and credentials.
Topics
FAQ
- What is claude-agent-sdk-python?
- Python SDK for building agents on Claude Code, with query(), an interactive client, in-process custom tools and hooks. The Claude Agent SDK for Python lets Python applications drive Claude Code programmatically. Install it with pip and call query() to send a prompt and iterate over the streamed response messages, or use ClaudeSDKClient for bidirectional, multi-turn conversations.
- Is claude-agent-sdk-python free to use? What license does it have?
- claude-agent-sdk-python is open source under the MIT license. Check the license terms before using it in commercial products.
- Is claude-agent-sdk-python still maintained?
- Actively maintained. The last commit to claude-agent-sdk-python was on Oct 19, 2026.
- What can I use claude-agent-sdk-python for?
- Embedding a Claude coding agent inside a Python application or internal tool. Adding domain-specific Python functions as tools that Claude can call. Enforcing policies such as blocking certain shell commands with hooks. Building automated workflows that run Claude Code against a project directory.
- How do I install claude-agent-sdk-python?
- Install the SDK from PyPI: pip install claude-agent-sdk Save the README quick start code as a Python file and run it with anyio
- Where can I get claude-agent-sdk-python?
- The source code and installation instructions are on GitHub: https://github.com/anthropics/claude-agent-sdk-python
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
FastGPT is an open-source AI Agent building platform with data processing, RAG knowledge bases, and visual workflow orchestration for LLM apps.
Agent-native memory infrastructure that turns LLM agent conversations and execution into persistent, structured memory for Python and TypeScript apps.
An agent framework that connects digital humans (2.5D, 3D, mobile, PC, web) and LLMs (OpenAI-compatible, DeepSeek) to business systems.
Memory operating system for LLMs and AI agents that stores, retrieves, and manages long-term memory as an editable graph.
Enterprise-grade agentic workflow platform for building and running AI agents, with model management, MCP tool integration and RPA automation.