[Agentproto] Re: Topic: Existing practice, use cases, and deployment models

[Iman Schrock <[e-mail]>]

Dear Orie, Leslie,

One concrete example for the existing recovery use case: a dialog resumes
and the agent recovers its outstanding task, but a downstream tool
operation is still unresolved.

Rosenberg and Jennings already cover recovering outstanding tasks in
Section 3.3.3:
https://www.ietf.org/archive/id/draft-rosenberg-agentproto-usecases-00.html#section-3.3.3

Now consider a refund task. The payment provider commits the refund, but
its response is lost. Recovering the enclosing task does not by itself tell
the agent whether the refund happened. The tool needs to retain that
uncertainty until its provider can resolve it.

Our executor-side treatment is in EP Architecture -03, Sections 3.4–3.6,
and Action Evidence Boundary -07, Sections 5.12–5.14. AEB Section 9
describes the deployment assumptions:
https://www.ietf.org/archive/id/draft-schrock-ep-architecture-03.html#section-3.4
https://www.ietf.org/archive/id/draft-schrock-action-evidence-boundary-07.html#section-5.12

A runnable payment-release example exercises the same response-loss case:
https://github.com/emiliaprotocol/emilia-protocol/tree/e507acdf8efbe8951cb4294801d4c440f0b86a5a/examples/indeterminate-effect-reconciliation

I reran its three tests: provider commitment followed by response loss,
refusal of a duplicate attempt and later reconciliation; rejection of
tampered evidence; and rejection of a signed result for the wrong
destination. It uses a mock provider and memory-only state, not a restart
or cross-vendor recovery test.

For the use-case document, I suggest this as a variant of Section 3.3.3.
For the architecture, it makes the deployment assumptions explicit: which
component retains operation state, and which can resolve the outcome.
Existing A2A/MCP task-status mechanisms should be reused where they
suffice. The dialog-to-task association belongs in the continuity
discussion; payment, retry and outcome rules stay with the tool and
provider.

Best,
Team EMILIA

On Fri, Oct 2, 2026 at 10:10 AM Orie <[e-mail]> wrote:

> Hi,
>
> This is the first in a series of topics, as noted here:
>
>
> https://mailarchive.ietf.org/arch/msg/agentproto/3_WBraJwt-jwDpZYb3uUoCRMrA0/
>
> Please identify Internet-Drafts describing existing systems, protocols,
> use cases, or deployment models relevant to agentproto.
>
> Point to the sections that explain how users, agents, and tools interact,
> what works today, and where interoperability gaps arise. Implementation and
> deployment experience will help us assess these contributions.
>
> The expected output is a set of candidate use cases, deployment models,
> and supporting references for **Use Cases and Requirements**, together with
> deployment assumptions for the **Reference Architecture**.
>
> Regards,
>
> Orie and Leslie
> agentproto co-chairs
>
> _______________________________________________
> Agentproto mailing list -- [e-mail]
> To unsubscribe send an email to [e-mail]
>