On 25 November 2024, Anthropic introduced the Model Context Protocol (MCP), an open standard for connecting AI applications with data sources, business tools and development environments.[1] Until then, each of those connections was usually built as a one-off integration.
Agent products were arriving in the same period. In February 2025 Anthropic released Claude Code alongside Claude 3.7 Sonnet, describing it as a research preview for agentic coding.[3]
On 9 April 2025, Google's Developers Blog announced Agent2Agent (A2A), an open protocol meant to let agents built by different vendors or on different frameworks discover each other, exchange information and collaborate on tasks.[2]
The two protocols work at different layers. MCP covers how one AI application reaches tools and context; A2A covers how separate agents talk to each other. Google positioned A2A as complementary to MCP.[2]
Sources
- 1Introducing the Model Context Protocolanthropic.com
MCP introduced on 2024-11-25 as an open standard
- 2Announcing the Agent2Agent Protocol (A2A)developers.googleblog.com
A2A announced on 2025-04-09; complementary to MCP
- 3Claude 3.7 Sonnet and Claude Codeanthropic.com
Claude Code released with Claude 3.7 Sonnet as a research preview
Cite this page: MCP and A2A: two open protocols for connecting AI agents. TellWorld, https://www.tellworld.net/analysis/mcp-and-a2a. Updated 2026-11-07
