unifying all image saves to /tmp to bug-proof (#14149)

image-gen feature will have the model saving to /tmp by default + at all
times

Author: won-openai

codex-rs/core/src/codex.rs

@@ -6811,8 +6811,7 @@ async fn handle_assistant_item_done_in_plan_mode(
     {
         maybe_complete_plan_item_from_message(sess, turn_context, state, item).await;
 
-        if let Some(turn_item) =
-            handle_non_tool_response_item(item, true, Some(&turn_context.cwd)).await
+        if let Some(turn_item) = handle_non_tool_response_item(sess, turn_context, item, true).await
         {
             emit_turn_item_in_plan_mode(
                 sess,
@@ -6993,8 +6992,13 @@ async fn try_run_sampling_request(
                 needs_follow_up |= output_result.needs_follow_up;
             }
             ResponseEvent::OutputItemAdded(item) => {
-                if let Some(turn_item) =
-                    handle_non_tool_response_item(&item, plan_mode, Some(&turn_context.cwd)).await
+                if let Some(turn_item) = handle_non_tool_response_item(
+                    sess.as_ref(),
+                    turn_context.as_ref(),
+                    &item,
+                    plan_mode,
+                )
+                .await
                 {
                     let mut turn_item = turn_item;
                     let mut seeded_parsed: Option<ParsedAssistantTextDelta> = None;

codex-rs/core/src/codex_tests.rs

@@ -154,6 +154,26 @@ fn developer_input_texts(items: &[ResponseItem]) -> Vec<&str> {
         .collect()
 }
 
+fn default_image_save_developer_message_text() -> String {
+    let image_output_dir = crate::stream_events_utils::default_image_generation_output_dir();
+    format!(
+        "Generated images are saved to {} as {} by default.",
+        image_output_dir.display(),
+        image_output_dir.join("<image_id>.png").display(),
+    )
+}
+
+fn test_tool_runtime(session: Arc<Session>, turn_context: Arc<TurnContext>) -> ToolCallRuntime {
+    let router = Arc::new(ToolRouter::from_config(
+        &turn_context.tools_config,
+        None,
+        None,
+        turn_context.dynamic_tools.as_slice(),
+    ));
+    let tracker = Arc::new(tokio::sync::Mutex::new(TurnDiffTracker::new()));
+    ToolCallRuntime::new(router, session, turn_context, tracker)
+}
+
 fn make_connector(id: &str, name: &str) -> AppInfo {
     AppInfo {
         id: id.to_string(),
@@ -3123,6 +3143,114 @@ async fn build_initial_context_uses_previous_realtime_state() {
     );
 }
 
+#[tokio::test]
+async fn build_initial_context_omits_default_image_save_location_with_image_history() {
+    let (session, turn_context) = make_session_and_context().await;
+    session
+        .replace_history(
+            vec![ResponseItem::ImageGenerationCall {
+                id: "ig-test".to_string(),
+                status: "completed".to_string(),
+                revised_prompt: Some("a tiny blue square".to_string()),
+                result: "Zm9v".to_string(),
+            }],
+            None,
+        )
+        .await;
+
+    let initial_context = session.build_initial_context(&turn_context).await;
+    let developer_texts = developer_input_texts(&initial_context);
+    assert!(
+        !developer_texts
+            .iter()
+            .any(|text| text.contains("Generated images are saved to")),
+        "expected initial context to omit image save instructions even with image history, got {developer_texts:?}"
+    );
+}
+
+#[tokio::test]
+async fn build_initial_context_omits_default_image_save_location_without_image_history() {
+    let (session, turn_context) = make_session_and_context().await;
+
+    let initial_context = session.build_initial_context(&turn_context).await;
+    let developer_texts = developer_input_texts(&initial_context);
+
+    assert!(
+        !developer_texts
+            .iter()
+            .any(|text| text.contains("Generated images are saved to")),
+        "expected initial context to omit image save instructions without image history, got {developer_texts:?}"
+    );
+}
+
+#[tokio::test]
+async fn handle_output_item_done_records_image_save_message_after_successful_save() {
+    let (session, turn_context) = make_session_and_context().await;
+    let session = Arc::new(session);
+    let turn_context = Arc::new(turn_context);
+    let call_id = "ig_history_records_message";
+    let expected_saved_path = crate::stream_events_utils::default_image_generation_output_dir()
+        .join(format!("{call_id}.png"));
+    let _ = std::fs::remove_file(&expected_saved_path);
+    let item = ResponseItem::ImageGenerationCall {
+        id: call_id.to_string(),
+        status: "completed".to_string(),
+        revised_prompt: Some("a tiny blue square".to_string()),
+        result: "Zm9v".to_string(),
+    };
+
+    let mut ctx = HandleOutputCtx {
+        sess: Arc::clone(&session),
+        turn_context: Arc::clone(&turn_context),
+        tool_runtime: test_tool_runtime(Arc::clone(&session), Arc::clone(&turn_context)),
+        cancellation_token: CancellationToken::new(),
+    };
+    handle_output_item_done(&mut ctx, item.clone(), None)
+        .await
+        .expect("image generation item should succeed");
+
+    let history = session.clone_history().await;
+    let expected_message: ResponseItem =
+        DeveloperInstructions::new(default_image_save_developer_message_text()).into();
+    assert_eq!(history.raw_items(), &[expected_message, item]);
+    assert_eq!(
+        std::fs::read(&expected_saved_path).expect("saved file"),
+        b"foo"
+    );
+    let _ = std::fs::remove_file(&expected_saved_path);
+}
+
+#[tokio::test]
+async fn handle_output_item_done_skips_image_save_message_when_save_fails() {
+    let (session, turn_context) = make_session_and_context().await;
+    let session = Arc::new(session);
+    let turn_context = Arc::new(turn_context);
+    let call_id = "ig_history_no_message";
+    let expected_saved_path = crate::stream_events_utils::default_image_generation_output_dir()
+        .join(format!("{call_id}.png"));
+    let _ = std::fs::remove_file(&expected_saved_path);
+    let item = ResponseItem::ImageGenerationCall {
+        id: call_id.to_string(),
+        status: "completed".to_string(),
+        revised_prompt: Some("broken payload".to_string()),
+        result: "_-8".to_string(),
+    };
+
+    let mut ctx = HandleOutputCtx {
+        sess: Arc::clone(&session),
+        turn_context: Arc::clone(&turn_context),
+        tool_runtime: test_tool_runtime(Arc::clone(&session), Arc::clone(&turn_context)),
+        cancellation_token: CancellationToken::new(),
+    };
+    handle_output_item_done(&mut ctx, item.clone(), None)
+        .await
+        .expect("image generation item should still complete");
+
+    let history = session.clone_history().await;
+    assert_eq!(history.raw_items(), &[item]);
+    assert!(!expected_saved_path.exists());
+}
+
 #[tokio::test]
 async fn build_initial_context_uses_previous_turn_settings_for_realtime_end() {
     let (session, turn_context) = make_session_and_context().await;

codex-rs/core/src/context_manager/history_tests.rs

@@ -434,9 +434,6 @@ fn for_prompt_rewrites_image_generation_calls_when_images_are_supported() {
                     ContentItem::InputImage {
                         image_url: "data:image/png;base64,Zm9v".to_string(),
                     },
-                    ContentItem::InputText {
-                        text: "Saved to: CWD".to_string(),
-                    },
                 ],
                 end_turn: None,
                 phase: None,
@@ -503,9 +500,6 @@ fn for_prompt_rewrites_image_generation_calls_when_images_are_unsupported() {
                         text: "image content omitted because you do not support image input"
                             .to_string(),
                     },
-                    ContentItem::InputText {
-                        text: "Saved to: CWD".to_string(),
-                    },
                 ],
                 end_turn: None,
                 phase: None,

codex-rs/core/src/context_manager/normalize.rs

@@ -242,9 +242,6 @@ pub(crate) fn rewrite_image_generation_calls_for_stateless_input(items: &mut Vec
                             text: format!("Prompt: {revised_prompt}"),
                         },
                         ContentItem::InputImage { image_url },
-                        ContentItem::InputText {
-                            text: "Saved to: CWD".to_string(),
-                        },
                     ],
                     end_turn: None,
                     phase: None,