New MCP Roadmap

The Model Context Protocol (MCP) has released an updated roadmap outlining its direction for upcoming specification releases and beyond. The roadmap, developed by Core Maintainers, the community of maintainers, and Working Groups, is organized into five priority areas. These areas build upon work previously listed in the previous roadmap, such as server-initiated events, result type improvements, and agent identity.

Agentic messaging primitives are crucial for modern workloads that can run for extended periods and require mid-flight steering. MCP has introduced Tasks, subscriptions/listen, and progress notifications to meet these requirements. The focus will be on providing appropriate primitives, ensuring they work well together, and advancing the Tasks extension (SEP-2663) for specification inclusion.

HTTP-native transport unification and hardening aim to simplify MCP server and client development by using a single transport for all deployment modes. This unification will enable more streamlined server and client development, as well as support for local servers speaking Streamable HTTP over stdio.

Agent identity and enterprise-ready security aim to provide a standardized way for MCP servers to recognize and trust agent identities. This standardization will be based on existing standards, such as Demonstrating Proof of Possession (DPoP) and Workload Identity Federation, rather than relying on API keys and long-lived tokens. The work will also involve finalizing DPoP and defining an opinionated path for agent identity and delegation.

Improved primitives, specifically tool calling, aim to standardize the contract used for result handling. This change will enable more efficient result handling, as a server developer will know which form a given client will use. The initiative also includes a progressive discovery effort, allowing servers to offer a small entry point and reveal more of their catalog as the conversation narrows.

The SDK developer experience is improved by investing in the ergonomics and conformance of existing SDKs, as well as making them intuitive and well-documented across all supported platforms and languages. This investment is particularly crucial as many developers build MCP clients and servers using existing libraries, where clear APIs and accurate documentation determine the ease of implementation.

Specification Enhancement Proposals (SEPs) falling within these priority areas receive expedited review and have a higher likelihood of acceptance. Proposals outside these priority areas will still be considered but will face reduced priority. Interested contributors can join or comment on Working Groups or Interest Groups, propose or comment on a SEP, start an experimental extension, or contribute directly to the specification, SDKs, and tooling. The MCP community encourages everyone to participate in shaping the future of the protocol.