[Agentproto] Re: [iesg] IAB review of proposed agentproto charter

["Charles Eckel (eckelcu)" <[e-mail]>]

Hi

On Oct 8, 2026, at 1:18 PM, Yingzhen Qu <[e-mail]> wrote:

Hi Charles,

Thanks for addressing my comments. The new charter looks good to me.

Thanks,
Yingzhen

On Thu, Oct 1, 2026 at 6:28 PM Charles Eckel (eckelcu) <[e-mail]<mailto:[e-mail]>> wrote:
Hi Yingzhen,

Thank you again for the thorough IAB review and suggested changes.
Please see inline for how I propose to address them.

On Sep 25, 2026, at 1:11 PM, Charles Eckel (eckelcu) <[e-mail]<mailto:[e-mail]>> wrote:

Hi Yingzhen,

Thank you for providing this review. I plan to get back to you regarding your comments next week, but in the meantime, I am adding the agentproto list to this initial reply so that others benefit from your review as well.

Thanks again,
Charles

On Sep 24, 2026, at 6:49 AM, Yingzhen Qu <[e-mail]<mailto:[e-mail]>> wrote:

Hi

I have been selected as the IAB reviewer of the proposed Agent Communication Protocols WG (agentproto) working group.

I have reviewed 03 of the proposed charter here: https://datatracker.ietf.org/doc/charter-ietf-agentproto/00-03/, in the context of the areas that the IAB can provide feedback on, here: https://wiki.ietf.org/en/group/iab/IAB_Review_of_Proposed_WG_Charters.

In general, I think the charter is fine, but I do have some comments.

In the "Key Considerations", my understanding of "AI agent chains" is the chain of AI agents that act on users behalf, including those invoked by other AI agents. Please correct me if I'm wrong.

Your understanding matches mine.


Assuming my understanding is correct, here are some suggested editorial changes to the following text.

Current:
"
To protect data exchanged between AI agents and users, other AI agents, and tools over potentially untrusted networks, particularly when handling sensitive information (such as personal data within dialog context), mechanisms are required to establish and verify identity of agents and of the users on whose behalf they act, ensure confidentiality, integrity, authenticity of the exchanged data, and delegated authorization across AI agent chains. This introduces new considerations around protocol-level security and privacy mechanisms.
"

Suggested:
"
To protect data exchanged between AI agents and users, other AI agents, and tools over potentially untrusted networks, particularly when handling sensitive information (such as personal data in dialog context), mechanisms are required to establish and verify the identities of agents and the users on whose behalf they act. These mechanisms need to ensure confidentiality, integrity, and authenticity of exchanged data and delegated authorization across AI agent chains. This raises new requirements for protocol-level security and privacy mechanisms.
"

Thank you for catching this. I propose to incorporate your suggested text as suggested.


In "Use Cases and Requirements", the two "deliverable" don't seem to mean the same thing.
Current:
"
Describe basic use cases and requirements that drive the protocol deliverable. The deliverable will also describe the deployment model(s).
"

Suggested:
"
An informational document will describe the basic use cases and requirements that drive the development of the protocol. It will also describe the deployment model(s).
"

I agree with this suggestion as well; however, I made some changes to match the construction used for the Reference Architecture deliverable and split the deliverable across two bullets.

Please see the corresponding PR:
https://github.com/ietf-artarea/charters/pull/127

Thanks again!
Charles


Thanks,
Yingzhen
_______________________________________________
iesg mailing list -- [e-mail]<mailto:[e-mail]>
To unsubscribe send an email to [e-mail]<mailto:[e-mail]>