MCP specification upgrades to stateless design, boosting enterprise scalability

The Model Context Protocol, an open-source framework for connecting AI systems to external tools and data, has undergone its most significant revision since its launch. The core of the protocol is now stateless, meaning requests no longer rely on a persistent session tied to a specific server instance, a shift that promises greater reliability and enterprise-level scalability. Lead maintainers David Soria Parra and Den Delimarsky, who work at Anthropic, described the change as the most important update since the protocol’s remote version debuted over a year ago.

The release also adds multi-round-trip request handling, header-based routing, cacheable list results, hardened authorization, a formal extensions framework, and updated Tier 1 SDKs. These enhancements aim to meet developer demand for more robust and scalable MCP deployments.