[app-server] type client response payloads

[rhan-oai]

Why

pr17088 adds typed server-originated request/response plumbing, but successful client responses are still erased into bare JSON-RPC result values before app-server can make any typed decision about them.

This precursor PR keeps successful client responses typed until the outgoing response seam. It is intentionally limited to protocol/app-server plumbing so the analytics behavior change can review separately on top.

What changed

• Add ClientResponsePayload as the pre-serialization client response body type.
• Route app-server successful response paths through the typed payload seam while preserving existing handler-local analytics behavior.
• Keep InterruptConversation JSON-RPC-only because it has no ClientResponse variant.
• Move the new payload conversion tests into a dedicated protocol test module.

Verification

• cargo check -p codex-app-server
• cargo test -p codex-app-server-protocol

[owenlin0]

Is there a way to avoid updating every callsite that sends an RPC response with the payload type? This seems boilerplatey, especially since the result is typed already.

Codex suggests:

Yes. The cleanest reduction is to make the protocol macro generate From for ClientResponsePayload for response types that are unambiguous, then change send_result/send_response to accept Into.

[owenlin0]

There's one annoying gotcha, which is that we have at least one reused response type:

ConfigValueWrite => response: v2::ConfigWriteResponse
ConfigBatchWrite => response: v2::ConfigWriteResponse

but maybe there's a fix/workaround for this one?

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: a1caa34dbe

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Restrict request analytics to turn requests

Avoid tracking every initialized request here. track_request now clones and enqueues all ClientRequests, but the analytics reducer only processes TurnStart/TurnSteer and discards the rest. High-volume or large-payload methods (for example command/exec/write or fs/writeFile with data_base64) can saturate the 256-item analytics queue, causing dropped telemetry and skewed experiment data.

Useful? React with 👍 / 👎.