diff --git a/skills b/skills
index 0685a6af4..e856b8e75 160000
--- a/skills
+++ b/skills
@@ -1 +1 @@
-Subproject commit 0685a6af4ef3ddea4fb26364dd72ea574f307314
+Subproject commit e856b8e756ad7ea86f36dbd1697542e76a8d3571
diff --git a/src-tauri/Cargo.lock b/src-tauri/Cargo.lock
index bb512679f..9b02b47de 100644
--- a/src-tauri/Cargo.lock
+++ b/src-tauri/Cargo.lock
@@ -4,7 +4,7 @@ version = 4
 
 [[package]]
 name = "OpenHuman"
-version = "0.49.12"
+version = "0.49.15"
 dependencies = [
  "aes-gcm",
  "android_logger",
diff --git a/src-tauri/src/commands/chat.rs b/src-tauri/src/commands/chat.rs
index 2d0919d0a..54fb36668 100644
--- a/src-tauri/src/commands/chat.rs
+++ b/src-tauri/src/commands/chat.rs
@@ -436,15 +436,9 @@ pub async fn chat_send(
         chat_state_arc.remove(&thread_id_clone);
 
         if let Err(e) = result {
-            let _ = app_clone.emit(
-                "chat:error",
-                ChatErrorEvent {
-                    thread_id: thread_id_clone,
-                    message: e,
-                    error_type: "inference".to_string(),
-                    round: None,
-                },
-            );
+            // chat_send_inner already emits chat:error for known error paths.
+            // Only log here to avoid duplicate error events.
+            log::error!("[chat] chat_send_inner failed: {e}");
         }
     });
 
@@ -501,15 +495,9 @@ pub async fn chat_send(
         chat_state_arc.remove(&thread_id_clone);
 
         if let Err(e) = result {
-            let _ = app_clone.emit(
-                "chat:error",
-                ChatErrorEvent {
-                    thread_id: thread_id_clone,
-                    message: e,
-                    error_type: "inference".to_string(),
-                    round: None,
-                },
-            );
+            // chat_send_mobile already emits chat:error for known error paths.
+            // Only log here to avoid duplicate error events.
+            log::error!("[chat] chat_send_mobile failed: {e}");
         }
     });
 
@@ -546,7 +534,16 @@ async fn chat_send_inner(
     let client = reqwest::Client::builder()
         .use_rustls_tls()
         .build()
-        .map_err(|e| format!("Failed to build HTTP client: {}", e))?;
+        .map_err(|e| {
+            let msg = format!("Failed to build HTTP client: {}", e);
+            let _ = app.emit("chat:error", ChatErrorEvent {
+                thread_id: thread_id.to_string(),
+                message: msg.clone(),
+                error_type: "network".to_string(),
+                round: None,
+            });
+            msg
+        })?;
 
     // ── Step 1: Load AI context ─────────────────────────────────────────
     let openclaw_context = load_openclaw_context(app);
@@ -775,7 +772,16 @@ async fn chat_send_inner(
         let completion: ChatCompletionResponse = response
             .json()
             .await
-            .map_err(|e| format!("Failed to parse inference response: {}", e))?;
+            .map_err(|e| {
+                let msg = format!("Failed to parse inference response: {}", e);
+                let _ = app.emit("chat:error", ChatErrorEvent {
+                    thread_id: thread_id.to_string(),
+                    message: msg.clone(),
+                    error_type: "inference".to_string(),
+                    round: Some(round),
+                });
+                msg
+            })?;
 
         // Accumulate token usage
         if let Some(ref usage) = completion.usage {
@@ -786,7 +792,16 @@ async fn chat_send_inner(
         let choice = completion
             .choices
             .first()
-            .ok_or_else(|| "No choices in inference response".to_string())?;
+            .ok_or_else(|| {
+                let msg = "No choices in inference response".to_string();
+                let _ = app.emit("chat:error", ChatErrorEvent {
+                    thread_id: thread_id.to_string(),
+                    message: msg.clone(),
+                    error_type: "inference".to_string(),
+                    round: Some(round),
+                });
+                msg
+            })?;
 
         log::info!(
             "[chat] Round {} — finish_reason={:?}, tool_calls={}",
@@ -1002,7 +1017,16 @@ async fn chat_send_mobile(
     let client = reqwest::Client::builder()
         .use_rustls_tls()
         .build()
-        .map_err(|e| format!("Failed to build HTTP client: {}", e))?;
+        .map_err(|e| {
+            let msg = format!("Failed to build HTTP client: {}", e);
+            let _ = app.emit("chat:error", ChatErrorEvent {
+                thread_id: thread_id.to_string(),
+                message: msg.clone(),
+                error_type: "network".to_string(),
+                round: None,
+            });
+            msg
+        })?;
 
     // ── Step 1: Load AI context ─────────────────────────────────────────
     let openclaw_context = load_openclaw_context(app);
@@ -1146,7 +1170,16 @@ async fn chat_send_mobile(
     let completion: ChatCompletionResponse = response
         .json()
         .await
-        .map_err(|e| format!("Failed to parse inference response: {}", e))?;
+        .map_err(|e| {
+            let msg = format!("Failed to parse inference response: {}", e);
+            let _ = app.emit("chat:error", ChatErrorEvent {
+                thread_id: thread_id.to_string(),
+                message: msg.clone(),
+                error_type: "inference".to_string(),
+                round: None,
+            });
+            msg
+        })?;
 
     let (total_input_tokens, total_output_tokens) = completion
         .usage
@@ -1156,7 +1189,16 @@ async fn chat_send_mobile(
     let choice = completion
         .choices
         .first()
-        .ok_or_else(|| "No choices in inference response".to_string())?;
+        .ok_or_else(|| {
+            let msg = "No choices in inference response".to_string();
+            let _ = app.emit("chat:error", ChatErrorEvent {
+                thread_id: thread_id.to_string(),
+                message: msg.clone(),
+                error_type: "inference".to_string(),
+                round: None,
+            });
+            msg
+        })?;
 
     let full_response = choice.message.content.clone().unwrap_or_default();
 
diff --git a/src/pages/Conversations.tsx b/src/pages/Conversations.tsx
index 877322287..5c90e0cd5 100644
--- a/src/pages/Conversations.tsx
+++ b/src/pages/Conversations.tsx
@@ -293,6 +293,20 @@ const Conversations = () => {
         });
 
         if (event.error_type !== 'cancelled') {
+          // Deduplicate: skip if the last message is already an error
+          const currentState = store.getState() as {
+            thread: { messagesByThreadId: Record<string, ThreadMessage[]> };
+          };
+          const threadMessages =
+            currentState.thread.messagesByThreadId[event.thread_id] || [];
+          const lastMsg = threadMessages[threadMessages.length - 1];
+          if (
+            lastMsg?.sender === 'agent' &&
+            lastMsg?.content === 'Something went wrong — please try again.'
+          ) {
+            return;
+          }
+
           dispatch(
             addInferenceResponse({
               content: 'Something went wrong — please try again.',