Status: IETF Independent Submission concluded; charter 2.1 de-confliction lifts
Members,
A status update on our upstream dependency, the referenced protocol specification.
The IETF Independent Submissions Editor has concluded consideration of draft-saihm-memory-protocol and released it from the ISE queue; the datatracker stream is now set to None. The direction from the ISE is to bring this work to the IETF proper. Agent-to-agent and agent-to-tool protocol work is now being taken up there — an “agentproto” Birds-of-a-Feather session was held at IETF 126 (Vienna, July 2026) to identify what needs standardizing and whether to charter a working group — and the Independent stream defers to IETF working-group work, active or forming, in the same space (RFC 4846, RFC 8730).
What this means for the Group:
Charter section 2.1 (de-confliction). Our charter placed a time-bound restriction on developing a competing normative specification, in force “until the IETF Independent Submission concludes (… is published as an RFC, or is withdrawn or declined).” That condition is now met, so the section 2.1 restriction lifts by its own terms.
Our posture is unchanged. Lifting the restriction does not change what this Group does. We continue to operate one layer above the protocol — interoperability profiles, a use-case catalogue, and conformance criteria and test vectors — and we do not re-specify the protocol here (charter sections 3 and 4).
Forward normative home. The natural home for the protocol’s normative evolution is the IETF working-group work now forming in this space. The Group will track that work and align our profiles and test vectors to it. When it produces a citable document, I will bring an amendment (section 11) to re-point our normative reference — currently draft-saihm-memory-protocol-01 — to that document, for your review and consensus.
No charter change today. I am not proposing an amendment in this note. The -01 draft remains our current normative reference and stays citable on the datatracker; I will keep it available as the reference text in the interim.
If you follow agent-memory or interop work at the IETF and would like to engage the agentproto effort directly, its BoF materials and mailing list are on the IETF datatracker; I will circulate the specific list address once I have confirmed it. I will keep this list posted as the IETF work takes shape.
Best regards,
Russell Jackson
Chair, AI Agent Memory Interoperability Community Group