China has released an open-source Deep Research multi-agent framework that supports the Model-Centric Protocol (MCP), enabling AI agents to connect with specialized tools for tasks such as web search, crawling, and Python code execution. MCP is rapidly emerging as a new AI-native standard for linking tools and data to large language models (LLMs), facilitating more advanced automation and agentic AI protocols. This protocol allows AI agents, including ChatGPT and Claude, to move beyond passive interactions and perform real actions within digital systems, such as managing emails by drafting and sending messages. The integration of MCP with the MyShell ecosystem, which had been previously aligned with Web2, positions it advantageously within the evolving Web3 AI landscape. Industry experts highlight MCP's potential to become the new API standard for developers building intelligent, autonomous AI systems.