MCP, an open source standard for how AI systems interact with external tools and data sources, has received its largest update since introduction. The key change transforms MCP's protocol core from a bidirectional stateful protocol into a request/response stateless protocol, addressing what lead maintainers call one of the most highly-requested features from developers seeking better reliability and scalability.
The update, announced in a blog post by Anthropic engineers David Soria Parra and Den Delimarsky, also adds Multi Round-Trip Requests, header-based routing, cacheable list results, authorization hardening, a formal extensions framework, and updated Tier 1 SDKs. A new deprecation policy guarantees at least 12 months between a feature's formal deprecation and its removal, with narrow exceptions for critical security updates.
MCP, originally introduced by Anthropic nearly two years ago and managed by the Agentic AI Foundation under the Linux Foundation, has grown beyond its originator to include contributions from OpenAI, Google, Microsoft, and Amazon. Lead maintainers note this is "MCP's most important update since remote MCP first launched over a year ago," with the changes designed to enable widespread enterprise deployment.