On July 23, the Internet Engineering Task Force held a Birds-of-a-Feather session called "agentproto" (Agent Communication Protocols) at IETF 126 in Vienna, one of three AI-agent-related BoFs at the meeting. This is the first time agent-to-agent communication has been put in front of the IETF as a possible chartered Working Group rather than treated purely as a vendor feature. A BoF isn't a vote and doesn't produce a standard on the spot. As one write-up of the session put it, the IETF works by rough consensus, and any resulting standard is years away. What the session does mark is a shift in framing: agent-to-agent messaging, task handoff, and delegation are now being discussed as internet infrastructure — the same category as DNS, TLS, or HTTP — rather than as a feature bundled into whichever vendor's SDK you happen to use. The timing lines up with a run of protocol maturation elsewhere this year. The Agent2Agent (A2A) project shipped its first stable v1.0.0 specification back in March, removing legacy OAuth implicit and password flows, adding Device Code and PKCE support, and formalizing signed Agent Card verification using JWS and JSON Canonicalization — a concrete step toward agents being able to prove who they are to each other before a task handoff happens. The Model Context Protocol's maintainers, meanwhile, moved to the Agentic AI Foundation under the Linux Foundation and are working through a large 2026-07-28 revision addressing authorization hardening. None of this guarantees agentproto becomes an actual IETF Working Group — BoFs sometimes stall or get folded into existing groups. But the fact that agent-to-agent orchestration is being pitched at the same standards body that governs email, routing, and web transport is a signal worth logging. If a Working Group does get chartered, it would be the first attempt to standardize agent coordination at the protocol layer rather than leaving it to whichever framework (A2A, MCP, or a proprietary vendor mesh) wins market share first. Nothing here is finalized. The session was a discussion, not a decision, and the realistic path from BoF to published RFC runs on the order of years, not months. What changed this week is that the conversation formally started.