The Agentic AI Foundation published a roadmap for MCP, the protocol connecting AI tools.
It focuses on AI agent messaging, HTTP unification, and enterprise security.
These changes aim to support long-running agents and easier scaling.
What happened
The Agentic AI Foundation (AAIF), which oversees the Model Context Protocol (MCP) specification under the Linux Foundation, published a roadmap outlining five priority areas for the protocol's future evolution.
Why it matters
MCP has become the de facto industry standard for connecting AI tools and services. The roadmap signals a shift from simple chat-based interactions to supporting long-running AI agents, unifying all communication over HTTP, and adding enterprise-ready security features such as agent identity and token exchange protocols.
What to watch
The roadmap includes implementing server-initiated events so servers can push results to clients without polling, and improving the developer experience for SDKs used by both humans and AI. These changes are expected to make MCP more scalable and easier to deploy in enterprise environments.
Ask the AI about this article →
MCP started as a protocol for connecting Anthropic's Claude to external data sources in November 2024. After moving to the Linux Foundation in December 2025, it became an open standard. This roadmap builds on that foundation, addressing the shift from chat-based AI to agentic AI that can perform long-running tasks in the background. The focus on server-initiated events and HTTP unification reflects a need for more robust, scalable communication. Enterprise security is also a key concern, with plans to implement token exchange protocols and permission separation so agents can act on behalf of users without relying on human approval. These changes are likely to make MCP more viable for production use in large organizations.
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytime
Ask AI anything about this article. Q&As are published on this page for other readers too.
Microsoft co-founder Bill Gates published an almost-6,000-word essay Tuesday warning that the transition to th…
SpaceX disclosed a partnership with Nvidia to deploy Vera CPUs as part of the architecture behind SpaceXAI's S…
A Semafor analysis found that over the past month, 10 out of 310 guest submissions to The Wall Street Journal…
OpenAI revealed benchmark results for its custom inference chip Jalapeño, claiming 1.5–1.9× more work per watt…
Z.ai Co. released the code for GLM-5.3-Flash, an LLM with 320 billion parameters that activates 18 billion per…
Deep Cogito Inc. raised $43 million in a Series A round led by TQ Ventures, with participation from Benchmark…
