The Model Context Protocol (MCP), a foundational interoperability protocol in the artificial intelligence field, is set to receive a major update next week. As a critical link connecting AI models with external data sources and services, MCP has long served as an underlying 'pipeline,' enabling applications like chatbots to securely access calendars, databases, or internal tools without requiring engineers to develop custom interfaces for each connection. This update spans three key directions: enhanced security, expanded functionality, and simplified specifications. Specific changes include: removing support for JSON-RPC batching to streamline protocol complexity; introducing structured data validation and elicitation mechanisms to support complex data structures in tool call results and allow servers to dynamically request additional user information during interactions; implementing stricter version negotiation policies by mandating clients to declare protocol versions via the MCP-Protocol-Version request header; classifying MCP servers as OAuth resource servers and adding protected resource metadata, while requiring clients to implement RFC 8707 resource indicator mechanisms to prevent malicious servers from obtaining access tokens. Additionally, the update supports resource link types in tool call results, enabling the return of external resource references instead of directly embedding content, and adds a security best practices page along with 'Security Considerations' documentation.