[Audit] Re: Individual draft on resource accounting for agent runs: draft-arsentev-agent-run-metrics

[Tom Sato <[e-mail]>]

Audit mailing list friends

As Mirja pointed out, this list is solely for the coming Audit BoF at IEFT
SFO.

However, many developers are in need of discussion forums, right now as
they are drafting and implementing.

I have a slack channel for this audience, originally set up for my
accountability architecture but now open to all discussions on audit
related drafts.
In this slack channel, you are not limited to IETF regulations so you can
connect, discuss and promote your work.
It is a private workspace of my own but people are welcome. Once Audit WG
is established, WG chairs can decide what kind of discussion forums they
want to set up.
https://join.slack.com/t/agentaccountability/shared_invite/zt-4ao9cb5zm-choYVp4NoocFygc90RHkRA

Mirja, ADs, I hope this is OK.

Tom Sato

2026年10月12日(月) 19:15 Mirja Kuehlewind (IETF) <ietf=
[e-mail]>:

> Hi all,
>
> Thanks for you interest in this work and implementation work in this
> space. There is currently no working group yet. This mailing is focused on
> scoping discussions in order to form a future working group.
>
> If you want to discuss implementation details about your system, I
> recommend to do that directly among yourself off-list. Also for a future
> working group, standardization inputs, meaning directly concrete in-scoped
> proposals that address parts of the agreed future charter directly, not
> necessarily whole systems that go beyond that, as well as comments on
> adopted working group items are of course welcome on a working group
> mailing list. But again, coordination on concrete implementation details,
> especially when not directly relevant to the standards part of the work,
> are best done using other channels. E.g. some working groups also use slack
> for that purpose.
>
> Thanks!
> Mirja
>
>
>
> On 12. Oct 2026, at 07:33, Nicholas Templeman <[e-mail]> wrote:
>
> Hi Evgenii,
>
> Yes, an optional pointer to a detached record fits what I had in mind. The
> accounting format need not adopt a CSOAI-specific schema or define the
> record itself.
>
> For the first JSON example, I suggest making the referenced artifact's
> immutable identity or digest, format and scope explicit. A link alone
> should not imply that the record was retrieved or its signature checked.
> The reporter's signed assertion, provider reconciliation and independent
> measurement remain separate checks. An inaccessible reference stays
> uncheckable; it must not become a zero or a pass.
>
> Your additional case is useful too: where the namespace, units, window and
> counting contract match, known non-overlapping totals that already exceed
> the claimed subtree total contradict that claim, even with a child report
> missing.
>
> Please post the four plain-JSON cases you outlined. We can use those to
> agree the expected collector findings and a concrete detached-record
> reference before trying an implementation.
>
> Best,
>
> Nicholas Templeman
>
> CSOAI Ltd / Council of AI
>
> [e-mail]
> _______________________________________________
> Audit mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>
>
> _______________________________________________
> Audit mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>


-- 

Tom Sato
佐藤俊之
代表取締役CEO
マイオーベルジュ株式会社
090-6315-1325
[e-mail]