MCP Got a Sibling: A2A Joins the Same Governance Body, Completing the Open Agent Protocol Stack

TL;DR: Google transferred A2A — the protocol for agent-to-agent handoffs — to the Agentic AI Foundation (AAIF) on Aug 20, placing it alongside Anthropic's MCP under neutral Linux Foundation governance.

For anyone building multi-agent systems: MCP connects agents to tools and data; A2A is the protocol for agents to delegate work to other agents. Both now live under the Agentic AI Foundation (AAIF), a Linux Foundation project with 250+ member organizations — AWS, Anthropic, Cloudflare, Google, Microsoft, and OpenAI are platinum members.

A2A v1.0 (March 2026) is already in production in supply chain, financial services, and IT operations. The move follows the same path MCP took when it moved to AAIF governance earlier this year.

The limit: this is a governance transfer, not a spec update. MCP and A2A still require separate implementations, and neither protocol mandates a specific orchestration approach — "A2A + MCP" is a stack you compose, not a product you install.

Why it matters: if you're building agents that delegate work to other agents — not just call tools — A2A + MCP is now the pair of stable, community-governed open specs to build against, with reduced risk that a vendor pivot strands your architecture.

Sources: Axios · Linux Foundation · DevOpsDigest