Add Frameless Bidi support for realtime conversations

codex-rs/app-server-protocol/src/protocol/common.rs

@@ -3304,7 +3304,7 @@ mod tests {
                 prompt: Some(Some("You are on a call".to_string())),
                 realtime_session_id: Some("sess_456".to_string()),
                 transport: None,
-                version: Some(RealtimeConversationVersion::V1),
+                version: Some(RealtimeConversationVersion::V3),
                 voice: Some(RealtimeVoice::Marin),
             },
         };
@@ -3325,7 +3325,7 @@ mod tests {
                     "prompt": "You are on a call",
                     "realtimeSessionId": "sess_456",
                     "transport": null,
-                    "version": "v1",
+                    "version": "v3",
                     "voice": "marin"
                 }
             }),

codex-rs/app-server-protocol/src/protocol/v2/realtime.rs

@@ -80,9 +80,9 @@ pub struct ThreadRealtimeStartParams {
     /// Optional prefix added to automatic Codex response items when `codexResponsesAsItems` is true.
     #[ts(optional = nullable)]
     pub codex_response_item_prefix: Option<String>,
-    /// Optional prefix added to automatic V1 Codex commentary sent with
-    /// `conversation.handoff.append` when `codexResponsesAsItems` is not true. Final answers are
-    /// sent without the prefix.
+    /// Optional prefix added to automatic V1 or V3 Codex commentary sent through the selected
+    /// Bidi handoff wire event when `codexResponsesAsItems` is not true. Final answers are sent
+    /// without the prefix.
     #[ts(optional = nullable)]
     pub codex_response_handoff_prefix: Option<String>,
     /// Overrides the configured realtime model for this session only.

codex-rs/app-server/README.md

@@ -172,7 +172,7 @@ Example with notification opt-out:
 - `thread/inject_items` — append raw Responses API items to a loaded thread’s model-visible history without starting a user turn; returns `{}` on success.
 - `turn/steer` — add user input to an already in-flight regular turn without starting a new turn; returns the active `turnId` that accepted the input. `clientUserMessageId` is optional; when supplied, the corresponding `userMessage` item echoes it as `clientId`. Review and manual compaction turns reject `turn/steer`.
 - `turn/interrupt` — request cancellation of an in-flight turn by `(thread_id, turn_id)`; success is an empty `{}` response and the turn finishes with `status: "interrupted"`.
-- `thread/realtime/start` — start a thread-scoped realtime session (experimental); pass `outputModality: "text"` or `outputModality: "audio"` to choose model output, optionally pass `model` and, for websocket transport only, `version` to override configured realtime selection for this session only, and pass `includeStartupContext: false` to omit Codex's generated startup context. By default, automatic Codex text follows the protocol's speakable output path. Pass `clientManagedHandoffs: true` to disable automatic Codex response delivery so only the client's explicit append calls produce handoffs. Pass `codexResponsesAsItems: true` to send automatic Codex responses as realtime conversation items instead, and optionally pass `codexResponseItemPrefix` to prepend experiment instructions to those items. For V1 sessions, pass `codexResponseHandoffPrefix` while item mode is disabled to route automatic Codex commentary through `conversation.handoff.append` with that prefix; final answers remain unprefixed. Returns `{}` and streams `thread/realtime/*` notifications. Omit `transport` for the websocket transport, or pass `{ "type": "webrtc", "sdp": "..." }` to create an AVAS/v1 WebRTC session from a browser-generated SDP offer; the remote answer SDP is emitted as `thread/realtime/sdp`. Explicit `version: "v2"` requests are rejected for WebRTC.
+- `thread/realtime/start` — start a thread-scoped realtime session (experimental); pass `outputModality: "text"` or `outputModality: "audio"` to choose model output, optionally pass `model` and `version` to override configured realtime selection for this session only, and pass `includeStartupContext: false` to omit Codex's generated startup context. Version `"v1"` uses legacy Bidi `conversation.handoff.*`, `"v2"` uses the Realtime Voice API, and `"v3"` preserves V1 Codex Voice behavior while using Frameless Bidi `delegation.*`. By default, automatic Codex text follows the protocol's speakable output path. Pass `clientManagedHandoffs: true` to disable automatic Codex response delivery so only the client's explicit append calls produce handoffs. Pass `codexResponsesAsItems: true` to send automatic Codex responses as realtime conversation items instead, and optionally pass `codexResponseItemPrefix` to prepend experiment instructions to those items. For V1 and V3 sessions, pass `codexResponseHandoffPrefix` while item mode is disabled to route automatic Codex commentary through the selected Bidi handoff wire event with that prefix; final answers remain unprefixed. Returns `{}` and streams `thread/realtime/*` notifications. Omit `transport` for the websocket transport, or pass `{ "type": "webrtc", "sdp": "..." }` to create a Bidi WebRTC session from a browser-generated SDP offer; the remote answer SDP is emitted as `thread/realtime/sdp`. Conversation `version: "v2"` requests remain unsupported for WebRTC.
 - `thread/realtime/appendAudio` — append an input audio chunk to the active realtime session (experimental); returns `{}`.
 - `thread/realtime/appendText` — append text input to the active realtime session with a required `role` of `user`, `developer`, or `assistant` (experimental); returns `{}`. Older clients that omit `role` default to `user`.
 - `thread/realtime/appendSpeech` — append text that the realtime model should speak to the user (experimental); returns `{}`.
@@ -897,10 +897,9 @@ Omit `prompt` to use Codex's default realtime backend prompt. Send `prompt: null
 `prompt: ""` when the session should start without that default backend prompt.
 Clients may also pass `model` on `thread/realtime/start` to select a
 different realtime session configuration without changing thread or user config.
-For websocket transport, clients may pass `version` to select the realtime
-protocol version for this session only. WebRTC sessions always use AVAS with
-realtime v1; omitting `version` uses v1, and explicitly passing `version: "v2"`
-is rejected.
+Clients may pass `version` to select the realtime protocol for this session
+only. WebRTC uses AVAS and supports legacy Bidi `"v1"` or Frameless Bidi
+`"v3"`; Realtime Voice `"v2"` is rejected for WebRTC.
 Pass `includeStartupContext: false` to skip Codex's startup context for this
 session while still using the selected backend prompt.
 Pass `clientManagedHandoffs: true` to suppress automatic Codex response handoffs
@@ -911,10 +910,10 @@ Pass `codexResponsesAsItems: true` to inject automatic Codex responses with
 path. When using that mode, `codexResponseItemPrefix` can prepend short
 experiment instructions to each automatic Codex response item. Omit
 `codexResponsesAsItems`, or pass `false`, to preserve the default speakable
-behavior. For V1 sessions, `codexResponseHandoffPrefix` instead routes automatic
-Codex commentary through `conversation.handoff.append` and prepends the provided
-text. Final answers remain unprefixed. Item mode takes precedence when
-`codexResponsesAsItems` is true.
+behavior. For V1 and V3 sessions, `codexResponseHandoffPrefix` instead routes
+automatic Codex commentary through the selected Bidi handoff wire event and
+prepends the provided text. Final answers remain unprefixed. Item mode takes
+precedence when `codexResponsesAsItems` is true.
 Call
 `thread/realtime/appendText` to append app-provided realtime text items, or
 `thread/realtime/appendSpeech` when the app decides a realtime update should be