modelcontextprotocol/python-sdk
The official Python SDK for the Model Context Protocol (MCP) enabling developers to build standardized MCP servers and clients for LLM context and interaction.
googleapis/mcp-toolbox-sdk-python
mcp-toolbox-sdk-python is the official Google repository of Python client SDKs for MCP Toolbox, a server that centralises tool definitions, in particular database access tools, for generative AI applications. The SDKs connect an application to a running Toolbox server and load the tools it defines, presenting them either as ordinary Python functions and objects or in the shape a given agent framework expects, so developers do not hand-write per-tool wrappers or duplicate credentials and connection logic in every agent. Four packages live in this monorepo. toolbox-core, under packages/toolbox-core, is framework-agnostic and intended for custom applications that call tools directly. toolbox-adk integrates with the Google Agent Development Kit. toolbox-langchain targets LangChain and LangGraph, exposing Toolbox tools in the form those frameworks bind to a model. toolbox-llamaindex does the same for LlamaIndex. Each package is published to PyPI separately and installed on its own with pip, and each has both a package README and a guide on the mcp-toolbox.dev documentation site. Getting started has two parts: standing up an MCP Toolbox server following the server's own guide, then installing whichever SDK matches the orchestration framework in use; a local quickstart tutorial on the documentation site walks through the full path. The repository holds only the client SDKs, while the Toolbox server itself lives in the googleapis/mcp-toolbox repository, and the README directs general questions to that project's issue tracker and SDK-specific ones here. It is licensed under Apache 2.0 and supports the Python versions listed on the PyPI badge. It is aimed at Python developers building agents that need governed, reusable tool and data access.
https://github.com/googleapis/mcp-toolbox-sdk-python
Also in Python
PrefectHQ/fastmcp
FastMCP offers a Pythonic framework for developing and interacting with Model Context Protocol (MCP) servers and clients, facilitating efficient communication with LLMs.
tadata-org/fastapi_mcp
FastAPI-MCP is a tool that exposes FastAPI endpoints as Model Context Protocol (MCP) tools with built-in authentication, preserving schemas and documentation, and supporting flexible deployment using FastAPI's native ASGI interface.
mcp-use/mcp-use
mcp-use is an open-source client library that enables easy connection of any tool-capable LLM to MCP servers for building custom agents with diverse tool access.
golf-mcp/golf
Golf is a production-ready framework for building, deploying, and scaling secure MCP servers that power AI agents, featuring automatic component discovery, built-in authentication, observability, debugging, telemetry, and runtime management.
cognitivecomputations/dolphin-mcp
Dolphin MCP is a flexible Python library and CLI tool that enables conversational interaction with multiple Model Context Protocol (MCP) servers using various large language models.
rectalogic/langchain-mcp
langchain-mcp provides Model Context Protocol (MCP) tool support for the LangChain framework, enabling integration of MCP-compliant tools to enhance language model applications.
mkinf-io/mkinf
Python SDK that pulls hosted MCP servers from the mkinf hub and exposes them as LangChain-compatible tools through a single unified interface.