Model Context Protocol
Model Context Protocol (MCP) is an open standard for connecting AI models to external services and tools, jointly developed by Anthropic, OpenAI, Google, and Microsoft. In 2026, MCP underwent a major architectural shift to version 2.0, transitioning from a stateful, session-based protocol to a stateless HTTP-based system that simplifies implementation, improves security, reduces complexity for smaller models, and enables better enterprise scalability and distributed deployment.
Updated 3 August 2026
Specifications
No specifications recorded yet.
Latest developments
Stateless MCP has recaptured my interest (and inspired mcp-explorer and datasette-mcp)
Simon Willison · 2 days ago ·
3
New MCP specification addresses the main barrier to enterprise adoption
Ars Technica · 3 days ago ·
14
How AgentCore Gateway supports the MCP 2026-07-28 spec
AWS Machine Learning · 5 days ago ·
13
MCP’s biggest update removes the machinery many servers were built around
The New Stack · 1 week ago ·
46
AI’s most important protocol is getting a little bit easier to use
TechCrunch AI · 1 week ago ·
47
The MCP debate has a context problem
The New Stack · 2 weeks ago ·
41
Building News Agents for Daily News Recaps with MCP, Q, and tmux
Eugene Yan · 1 year ago ·
41
Tiny Agents: an MCP-powered agent in 50 lines of code
Hugging Face Blog · 1 year ago ·
31
July 2026
Model Context Protocol releases stateless 2.0 specification removing session requirements Feature update
- Stateless MCP has recaptured my interest (and inspired mcp-explorer and datasette-mcp)
- New MCP specification addresses the main barrier to enterprise adoption
- How AgentCore Gateway supports the MCP 2026-07-28 spec
- MCP’s biggest update removes the machinery many servers were built around
- AI’s most important protocol is getting a little bit easier to use
- The MCP debate has a context problem