Update agents/README.md

agents/README.md

@@ -0,0 +1 @@
+use livekit-plugins-volcengine==1.2.9

agents/my_basic_agent_1_2_9.py

@@ -1018,6 +1018,16 @@ async def entrypoint(ctx: JobContext, avatar_dispatcher_url: str = None, vision_
         initial_instructions = participant.attributes.get("instructions")
         logger.info(f"User selected instructions: {initial_instructions}")
 
+    # Read talking_mode from frontend state
+    initial_talking_mode = DEFAULT_TALKING_MODE
+    if participant.attributes.get("talking_mode"):
+        frontend_talking_mode = participant.attributes.get("talking_mode")
+        if frontend_talking_mode in ["push_to_talk", "realtime"]:
+            initial_talking_mode = frontend_talking_mode
+            logger.info(f"Initializing talking_mode from frontend: {initial_talking_mode}")
+        else:
+            logger.warning(f"Invalid talking_mode from frontend: {frontend_talking_mode}, using default: {initial_talking_mode}")
+
     # Replace the datetime and weekday placeholders to avoid KeyError from other braces in the prompt
     initial_instructions = initial_instructions.replace("{datetime}", current_time)
     initial_instructions = initial_instructions.replace("{weekday}", current_weekday)
@@ -1128,7 +1138,7 @@ async def entrypoint(ctx: JobContext, avatar_dispatcher_url: str = None, vision_
     )
 
      # disable input audio at the start
-    _talking_mode = DEFAULT_TALKING_MODE
+    _talking_mode = initial_talking_mode
     if _talking_mode == "push_to_talk":
         session.input.set_audio_enabled(False)
     else:
@@ -1192,6 +1202,8 @@ async def entrypoint(ctx: JobContext, avatar_dispatcher_url: str = None, vision_
                 session.input.set_audio_enabled(False)
                 logger.info("Setting audio enabled to False (PTT mode)")
             else:
+                # When switching to realtime mode, clear user turn state to ensure proper initialization
+                session.clear_user_turn()
                 session.input.set_audio_enabled(True)
                 logger.info("Setting audio enabled to True (realtime mode)")
             
@@ -1204,6 +1216,8 @@ async def entrypoint(ctx: JobContext, avatar_dispatcher_url: str = None, vision_
             if _talking_mode == "push_to_talk":
                 session.input.set_audio_enabled(False)
             else:
+                # When switching to realtime mode, clear user turn state
+                session.clear_user_turn()
                 session.input.set_audio_enabled(True)
             return json.dumps({"success": True, "mode": _talking_mode})
 

next.config.js

@@ -4,6 +4,10 @@ const withNextPluginPreval = createNextPluginPreval();
 /** @type {import('next').NextConfig} */
 const nextConfig = {
   reactStrictMode: false,
+  // Explicitly allow znjj.wangxin93.eu.org for Dev Origin, per future Next.js requirement.
+  allowedDevOrigins: [
+    "znjj.wangxin93.eu.org",
+  ],
 };
 
 module.exports = withNextPluginPreval(nextConfig);

src/components/chat/ChatOverlay.tsx

@@ -11,7 +11,7 @@ export interface ChatOverlayProps {
   isVisible: boolean;
   position: { x: number; y: number };
   onPositionChange: (position: { x: number; y: number }) => void;
-  containerRef: React.RefObject<HTMLDivElement>;
+  containerRef: React.RefObject<HTMLDivElement | null>;
   onToggle: () => void;
 }
 

src/components/playground/PhoneSimulator.tsx

@@ -70,6 +70,11 @@ export function PhoneSimulator({
   const pushToTalkButtonRef = useRef<HTMLButtonElement>(null);
   const [showChatOverlay, setShowChatOverlay] = useState(false);
   const [chatOverlayPosition, setChatOverlayPosition] = useState({ x: 0, y: 0 }); // Will be positioned at top-right by ChatOverlay component
+  const [chatTogglePosition, setChatTogglePosition] = useState<{ x?: number; right?: number; y: number }>({ right: 16, y: 56 }); // Initial position on the right
+  const [isDraggingChatToggle, setIsDraggingChatToggle] = useState(false);
+  const chatToggleRef = useRef<HTMLButtonElement>(null);
+  const chatToggleDragOffset = useRef({ x: 0, y: 0 });
+  const chatToggleHasDragged = useRef(false);
 
   useEffect(() => {
     const voiceAttr = config.settings.attributes?.find(a => a.key === "voice");
@@ -78,6 +83,47 @@ export function PhoneSimulator({
     }
   }, [config.settings.attributes]);
 
+  // Set talking_mode attribute when connected or when mode changes
+  const lastTalkingModeRef = useRef<string | null>(null);
+  const configAttributesRef = useRef(config.settings.attributes);
+  
+  // Update config attributes ref when it changes
+  useEffect(() => {
+    configAttributesRef.current = config.settings.attributes;
+  }, [config.settings.attributes]);
+  
+  useEffect(() => {
+    if (roomState === ConnectionState.Connected && localParticipant) {
+      const talkingMode = isPushToTalkMode ? "push_to_talk" : "realtime";
+      
+      // Only update if the mode actually changed
+      if (lastTalkingModeRef.current === talkingMode) {
+        return;
+      }
+      lastTalkingModeRef.current = talkingMode;
+      
+      try {
+        // Get current attributes from config to preserve them
+        const attributesToSet: Record<string, string> = {};
+        const configAttributes = configAttributesRef.current || [];
+        configAttributes.forEach(attr => {
+          if (attr.key && attr.value) {
+            attributesToSet[attr.key] = attr.value;
+          }
+        });
+        // Add talking_mode
+        attributesToSet.talking_mode = talkingMode;
+        
+        localParticipant.setAttributes(attributesToSet);
+      } catch (error) {
+        console.error("Failed to set talking_mode attribute:", error);
+      }
+    } else if (roomState === ConnectionState.Disconnected) {
+      // Reset ref when disconnected
+      lastTalkingModeRef.current = null;
+    }
+  }, [roomState, localParticipant, isPushToTalkMode]);
+
   const [currentTime, setCurrentTime] = useState("");
 
   const [visualizerPosition, setVisualizerPosition] = useState({
@@ -148,6 +194,96 @@ export function PhoneSimulator({
     };
   }, [isDragging]);
 
+  // Chat toggle button drag handlers
+  const handleChatToggleDragStart = (e: React.MouseEvent | React.TouchEvent) => {
+    e.preventDefault();
+    e.stopPropagation(); // Prevent triggering the button click
+    setIsDraggingChatToggle(true);
+    chatToggleHasDragged.current = false;
+    if (!phoneContainerRef.current || !chatToggleRef.current) return;
+    
+    const containerRect = phoneContainerRef.current.getBoundingClientRect();
+    const buttonRect = chatToggleRef.current.getBoundingClientRect();
+    const clientX = 'touches' in e ? e.touches[0].clientX : e.clientX;
+    const clientY = 'touches' in e ? e.touches[0].clientY : e.clientY;
+    
+    // If using right positioning, convert to x for dragging
+    if (chatTogglePosition.right !== undefined && chatTogglePosition.x === undefined) {
+      const currentX = containerRect.width - chatTogglePosition.right - buttonRect.width;
+      setChatTogglePosition({ x: currentX, y: chatTogglePosition.y });
+      chatToggleDragOffset.current = {
+        x: clientX - containerRect.left - currentX,
+        y: clientY - containerRect.top - chatTogglePosition.y,
+      };
+    } else {
+      // Already using x positioning
+      const currentX = chatTogglePosition.x ?? 0;
+      chatToggleDragOffset.current = {
+        x: clientX - containerRect.left - currentX,
+        y: clientY - containerRect.top - chatTogglePosition.y,
+      };
+    }
+  };
+
+  const handleChatToggleDragMove = (e: MouseEvent | TouchEvent) => {
+    if (!isDraggingChatToggle || !phoneContainerRef.current || !chatToggleRef.current) return;
+
+    e.preventDefault();
+    chatToggleHasDragged.current = true; // Mark that we've actually dragged
+
+    const containerRect = phoneContainerRef.current.getBoundingClientRect();
+    const buttonRect = chatToggleRef.current.getBoundingClientRect();
+
+    const clientX = 'touches' in e ? e.touches[0].clientX : e.clientX;
+    const clientY = 'touches' in e ? e.touches[0].clientY : e.clientY;
+
+    // Calculate new position relative to container
+    let newX = clientX - containerRect.left - chatToggleDragOffset.current.x;
+    let newY = clientY - containerRect.top - chatToggleDragOffset.current.y;
+
+    // Constrain within container
+    const maxX = containerRect.width - buttonRect.width;
+    const maxY = containerRect.height - buttonRect.height;
+    // On mobile (width < 768px), status bar is hidden, so allow dragging to top (y=0)
+    // On desktop, keep status bar height constraint (48px)
+    const isMobile = typeof window !== 'undefined' && window.innerWidth < 768;
+    const minY = isMobile ? 0 : 48; // statusBarHeight = 48px
+
+    newX = Math.max(0, Math.min(newX, maxX));
+    newY = Math.max(minY, Math.min(newY, maxY));
+
+    setChatTogglePosition({
+      x: newX,
+      y: newY,
+    });
+  };
+
+  const handleChatToggleDragEnd = () => {
+    setIsDraggingChatToggle(false);
+    // Reset the flag after a short delay to allow onClick to check it
+    setTimeout(() => {
+      chatToggleHasDragged.current = false;
+    }, 100);
+  };
+
+  useEffect(() => {
+    if (isDraggingChatToggle) {
+      window.addEventListener("mouseup", handleChatToggleDragEnd);
+      window.addEventListener("mousemove", handleChatToggleDragMove);
+      window.addEventListener("touchend", handleChatToggleDragEnd);
+      window.addEventListener("touchmove", handleChatToggleDragMove, { passive: false });
+    }
+    return () => {
+      window.removeEventListener("mouseup", handleChatToggleDragEnd);
+      window.removeEventListener("mousemove", handleChatToggleDragMove);
+      window.removeEventListener("touchend", handleChatToggleDragEnd);
+      window.removeEventListener("touchmove", handleChatToggleDragMove);
+    };
+  }, [isDraggingChatToggle]);
+
+  // Initialize chat toggle button position - keep it on the right using 'right' CSS property
+  // Only convert to 'x' (left positioning) when user drags it
+
   useEffect(() => {
     if (showCameraMenu) {
       Room.getLocalDevices("videoinput").then(setCameras);
@@ -851,22 +987,31 @@ export function PhoneSimulator({
         </div>
       </div>
 
-      {/* Chat Toggle Button - Top Right, aligned with audio visualizer */}
+      {/* Chat Toggle Button - Top Right, aligned with audio visualizer (Draggable) */}
       {roomState === ConnectionState.Connected && 
        voiceAssistant.agent && 
        phoneMode !== "important_message" && 
        phoneMode !== "capture" && (
         <button
-          className={`absolute right-2 z-50 p-3 rounded-full backdrop-blur-md transition-colors shadow-lg ${
+          ref={chatToggleRef}
+          className={`absolute z-50 p-3 rounded-full backdrop-blur-md transition-colors shadow-lg cursor-move select-none touch-none ${
             showChatOverlay
               ? "bg-blue-500/80 text-white"
               : "bg-gray-800/70 text-white hover:bg-gray-800/90"
           }`}
-          onClick={() => setShowChatOverlay(!showChatOverlay)}
-          title={showChatOverlay ? "Hide chat" : "Show chat"}
+          onClick={(e) => {
+            // Only toggle if we didn't just drag
+            if (!chatToggleHasDragged.current) {
+              setShowChatOverlay(!showChatOverlay);
+            }
+          }}
+          onMouseDown={handleChatToggleDragStart}
+          onTouchStart={handleChatToggleDragStart}
+          title={showChatOverlay ? "Hide chat (drag to move)" : "Show chat (drag to move)"}
           style={{
-            top: '56px', // Align with audio visualizer initial position
-            right: '8px',
+            ...(chatTogglePosition.x !== undefined ? { left: chatTogglePosition.x } : {}),
+            ...(chatTogglePosition.right !== undefined ? { right: chatTogglePosition.right } : {}),
+            top: chatTogglePosition.y,
           }}
         >
           <ChatIcon className="w-5 h-5 md:w-6 md:h-6" />
@@ -1031,7 +1176,7 @@ export function PhoneSimulator({
         <ChatOverlay
           agentAudioTrack={voiceAssistant.audioTrack}
           accentColor={config.settings.theme_color}
-          inputDisabled={phoneMode === "important_message" || phoneMode === "hand_off"}
+          inputDisabled={phoneMode === "hand_off"}
           isVisible={showChatOverlay}
           position={chatOverlayPosition}
           onPositionChange={setChatOverlayPosition}
@@ -1255,7 +1400,7 @@ export function PhoneSimulator({
               )}
 
               {/* Realtime Mode Layout */}
-              {!isPushToTalkMode && (
+              {!isPushToTalkMode && phoneMode !== "hand_off" && (
                 <>
                   {/* Important Message Mode - Centered End Call Button */}
                   {phoneMode === "important_message" ? (
@@ -1270,7 +1415,6 @@ export function PhoneSimulator({
                   ) : (
                     <div className="w-full flex items-center justify-center gap-4">
                       {/* Mic Toggle */}
-                      {phoneMode !== "hand_off" && (
                         <button
                           className={`p-4 rounded-full backdrop-blur-md transition-colors ${
                             !isMicEnabled
@@ -1285,7 +1429,6 @@ export function PhoneSimulator({
                             <MicOffIcon className="w-6 h-6" />
                           )}
                         </button>
-                      )}
 
                       {/* End Call Button */}
                       <button