The Model Context Protocol (MCP), an open source standard for AI system interactions, has undergone its largest update since its introduction. The core of the protocol is now stateless, meaning requests are no longer tied to a session with a specific server instance. This change aims to overcome long-standing scalability challenges. According to the blog post by lead maintainers David Soria Parra and Den Delimarsky, the update represents a major shift from a bidirectional stateful protocol to a request/response stateless one. It was one of the most requested features from developers seeking better reliability and scalability for their MCP servers. The update is described as 'MCP’s most important since remote MCP first launched over a year ago,' by Soria Parra. The changes include features such as multi round-trip requests, header-based routing, cacheable list results, authorization hardening, a formal extensions framework, and updated Tier 1 SDKs. The goal is to make enterprise deployment easier through improved scalability and reliability. The protocol's documentation details these changes, which are intended to support broader adoption in enterprise environments. The new spec marks a significant shift from MCP's original design, which ran on a local machine and connected models to local apps. The update also introduces a deprecation policy ensuring at least 12 months between formal deprecation and removal of a feature, with a narrow exception for critical security updates. This aligns with the overall focus on improving enterprise scalability. MCP is managed by the Agentic AI Foundation (AAIF), which is part of the Linux Foundation. The protocol was originally introduced by Anthropic and is still influenced by the company, though it has grown beyond its initial scope. OpenAI, Google, Microsoft, and Amazon also contribute to the project, and it is supported by many developer tools and software used for knowledge and creative work. The responsibility for maintaining MCP lies with individual maintainers, not the contributing companies, though some principal maintainers currently work at Anthropic. The update reflects a broader effort to enhance MCP's capabilities for enterprise-scale applications.
Source: arstechnica
