[codex] Add user input client ids (#24653)

## Summary

Adds an optional `clientId` field to app-server v2 `UserInput` and
carries it through the core `UserInput` model so clients can correlate
echoed user input items without relying on payload equality.

## Details

- Adds `client_id: Option<String>` to core `UserInput` variants.
- Exposes the v2 app-server field as `clientId` on the wire and in
generated TypeScript.
- Preserves the id when converting between app-server v2 and core
protocol types.
- Regenerates app-server schema fixtures.

## Validation

- `just fmt`
- `just write-app-server-schema`
- `cargo test -p codex-app-server-protocol`
- `cargo test -p codex-protocol`
- `just fix -p codex-app-server-protocol`
- `just fix -p codex-protocol`
- `git diff --check`

codex-rs/protocol/src/items.rs

@@ -56,6 +56,9 @@ pub enum TurnItem {
 #[derive(Debug, Clone, Deserialize, Serialize, TS, JsonSchema)]
 pub struct UserMessageItem {
     pub id: String,
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    #[ts(optional)]
+    pub client_id: Option<String>,
     pub content: Vec<UserInput>,
 }
 
@@ -237,6 +240,7 @@ impl UserMessageItem {
     pub fn new(content: &[UserInput]) -> Self {
         Self {
             id: uuid::Uuid::new_v4().to_string(),
+            client_id: None,
             content: content.to_vec(),
         }
     }
@@ -245,6 +249,7 @@ impl UserMessageItem {
         // Legacy user-message events flatten only text inputs into `message` and
         // rebase text element ranges onto that concatenated text.
         EventMsg::UserMessage(UserMessageEvent {
+            client_id: self.client_id.clone(),
             message: self.message(),
             images: Some(self.image_urls()),
             image_details: self.image_details(),
@@ -272,6 +277,7 @@ impl UserMessageItem {
             if let UserInput::Text {
                 text,
                 text_elements,
+                ..
             } = input
             {
                 // Text element ranges are relative to each text chunk; offset them so they align

codex-rs/protocol/src/models.rs

@@ -1237,7 +1237,9 @@ impl From<Vec<UserInput>> for ResponseInputItem {
                 .into_iter()
                 .flat_map(|c| match c {
                     UserInput::Text { text, .. } => vec![ContentItem::InputText { text }],
-                    UserInput::Image { image_url, detail } => {
+                    UserInput::Image {
+                        image_url, detail, ..
+                    } => {
                         image_index += 1;
                         let detail = detail.unwrap_or(DEFAULT_IMAGE_DETAIL);
                         vec![ContentItem::InputImage {
@@ -1245,7 +1247,7 @@ impl From<Vec<UserInput>> for ResponseInputItem {
                             detail: Some(detail),
                         }]
                     }
-                    UserInput::LocalImage { path, detail } => {
+                    UserInput::LocalImage { path, detail, .. } => {
                         image_index += 1;
                         let detail = detail.unwrap_or(DEFAULT_IMAGE_DETAIL);
                         match std::fs::read(&path) {

codex-rs/protocol/src/protocol.rs

@@ -129,6 +129,9 @@ pub struct Submission {
     pub id: String,
     /// Payload
     pub op: Op,
+    /// Client-provided id for the user message represented by `Op::UserInput`.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub client_user_message_id: Option<String>,
     /// Optional W3C trace carrier propagated across async submission handoffs.
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub trace: Option<W3cTraceContext>,
@@ -2164,6 +2167,8 @@ pub struct AgentMessageEvent {
 
 #[derive(Debug, Clone, Default, Deserialize, Serialize, JsonSchema, TS)]
 pub struct UserMessageEvent {
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub client_id: Option<String>,
     pub message: String,
     /// Image URLs sourced from `UserInput::Image`. These are safe
     /// to replay in legacy UI history events and correspond to images sent to
@@ -5088,6 +5093,7 @@ mod tests {
     #[test]
     fn user_message_event_serializes_empty_metadata_vectors() -> Result<()> {
         let event = UserMessageEvent {
+            client_id: None,
             message: "hello".to_string(),
             images: None,
             local_images: Vec::new(),
@@ -5133,7 +5139,7 @@ mod tests {
     #[test]
     fn user_message_item_legacy_event_preserves_image_details() {
         let local_path = PathBuf::from("/tmp/local.png");
-        let item = UserMessageItem::new(&[
+        let mut item = UserMessageItem::new(&[
             crate::user_input::UserInput::Image {
                 image_url: "https://example.com/first.png".to_string(),
                 detail: Some(ImageDetail::Original),
@@ -5147,6 +5153,7 @@ mod tests {
                 detail: Some(ImageDetail::Original),
             },
         ]);
+        item.client_id = Some("client-message-1".to_string());
 
         let EventMsg::UserMessage(event) = item.as_legacy_event() else {
             panic!("expected user message event");
@@ -5159,6 +5166,7 @@ mod tests {
                 "https://example.com/second.png".to_string(),
             ])
         );
+        assert_eq!(event.client_id, Some("client-message-1".to_string()));
         assert_eq!(event.image_details, vec![Some(ImageDetail::Original)]);
         assert_eq!(event.local_images, vec![local_path]);
         assert_eq!(event.local_image_details, vec![Some(ImageDetail::Original)]);