feat(desktop): render sent images/videos/files

agent/tools/send/send.py

@@ -54,6 +54,11 @@ class Send(BaseTool):
         if not path:
             return ToolResult.fail("Error: path parameter is required")
         
+        # Pass through remote URLs directly (no local file check): the client
+        # renders the link inline, so no download is needed.
+        if path.lower().startswith(("http://", "https://")):
+            return self._build_url_result(path, message)
+        
         # Resolve path
         absolute_path = self._resolve_path(path)
         
@@ -112,6 +117,46 @@ class Send(BaseTool):
 
         return ToolResult.success(result)
     
+    def _build_url_result(self, url: str, message: str) -> ToolResult:
+        """Build a file_to_send result for a remote http(s) URL.
+
+        The URL is passed through as both ``path`` and ``url`` so downstream
+        channels render it inline without downloading it locally.
+        """
+        # Infer file type from the URL path extension (ignore query string).
+        from urllib.parse import urlparse
+        url_path = urlparse(url).path
+        file_ext = Path(url_path).suffix.lower()
+        file_name = Path(url_path).name or "file"
+
+        if file_ext in self.image_extensions:
+            file_type = "image"
+            mime_type = self._get_image_mime_type(file_ext)
+        elif file_ext in self.video_extensions:
+            file_type = "video"
+            mime_type = self._get_video_mime_type(file_ext)
+        elif file_ext in self.audio_extensions:
+            file_type = "audio"
+            mime_type = self._get_audio_mime_type(file_ext)
+        elif file_ext in self.document_extensions:
+            file_type = "document"
+            mime_type = self._get_document_mime_type(file_ext)
+        else:
+            # Default to image: most pass-through URLs are generated images.
+            file_type = "image"
+            mime_type = "image/jpeg"
+
+        result = {
+            "type": "file_to_send",
+            "file_type": file_type,
+            "path": url,
+            "url": url,
+            "file_name": file_name,
+            "mime_type": mime_type,
+            "message": message or f"正在发送 {file_name}",
+        }
+        return ToolResult.success(result)
+
     def _resolve_path(self, path: str) -> str:
         """Resolve path to absolute path"""
         path = expand_path(path)

bridge/agent_bridge.py

@@ -684,11 +684,21 @@ class AgentBridge:
         """
         file_type = file_info.get("file_type", "file")
         file_path = file_info.get("path")
+        # Remote URLs are passed through as-is; local paths get a file:// prefix
+        # so the channel can read them from disk.
+        remote_url = file_info.get("url", "")
+        is_remote = bool(remote_url) and remote_url.lower().startswith(("http://", "https://"))
+
+        def _to_channel_url(p: str) -> str:
+            if is_remote:
+                return remote_url
+            if p and p.lower().startswith(("http://", "https://")):
+                return p
+            return f"file://{p}"
 
         # For images, use IMAGE_URL type (channel will handle upload)
         if file_type == "image":
-            # Convert local path to file:// URL for channel processing
+            file_url = _to_channel_url(file_path)
-            file_url = f"file://{file_path}"
             logger.info(f"[AgentBridge] Sending image: {file_url}")
             reply = Reply(ReplyType.IMAGE_URL, file_url)
             # Attach text message if present (for channels that support text+image)
@@ -698,7 +708,7 @@ class AgentBridge:
         
         # For all file types (document, video, audio), use FILE type
         if file_type in ["document", "video", "audio"]:
-            file_url = f"file://{file_path}"
+            file_url = _to_channel_url(file_path)
             logger.info(f"[AgentBridge] Sending {file_type}: {file_url}")
             reply = Reply(ReplyType.FILE, file_url)
             reply.file_name = file_info.get("file_name", os.path.basename(file_path))
@@ -708,7 +718,7 @@ class AgentBridge:
             return reply
         
         # For all other file types (tar.gz, zip, etc.), also use FILE type
-        file_url = f"file://{file_path}"
+        file_url = _to_channel_url(file_path)
         logger.info(f"[AgentBridge] Sending generic file: {file_url}")
         reply = Reply(ReplyType.FILE, file_url)
         reply.file_name = file_info.get("file_name", os.path.basename(file_path))

channel/web/web_channel.py

@@ -556,14 +556,29 @@ class WebChannel(ChatChannel):
                 file_path = data.get("path", "")
                 file_name = data.get("file_name", os.path.basename(file_path))
                 file_type = data.get("file_type", "file")
-                from urllib.parse import quote
+                # Remote URLs are passed through as-is; local files are served
-                web_url = f"/api/file?path={quote(file_path)}"
+                # via the backend /api/file endpoint.
+                remote_url = data.get("url", "")
+                is_remote = bool(remote_url) and remote_url.lower().startswith(("http://", "https://"))
+                if is_remote:
+                    web_url = remote_url
+                else:
+                    from urllib.parse import quote
+                    web_url = f"/api/file?path={quote(file_path)}"
                 is_image = file_type == "image"
-                q.put({
+                payload = {
                     "type": "image" if is_image else "file",
                     "content": web_url,
                     "file_name": file_name,
-                })
+                    # Preserve the concrete media kind (image/video/audio/...)
+                    # so richer clients can render an inline player.
+                    "file_type": file_type,
+                }
+                # Expose the local absolute path so the desktop client can open
+                # the file directly (Finder / default app) instead of the browser.
+                if not is_remote and file_path:
+                    payload["abs_path"] = file_path
+                q.put(payload)
 
         return on_event
 

desktop/src/main/index.ts

@@ -219,6 +219,15 @@ function setupIPC() {
     return result.canceled ? null : result.filePaths[0]
   })
 
+  // Open a local file with the OS default app; falls back to revealing it in
+  // the file manager when no handler exists. Returns '' on success.
+  ipcMain.handle('open-path', async (_event, targetPath: string) => {
+    if (!targetPath) return 'empty path'
+    const err = await shell.openPath(targetPath)
+    if (err) shell.showItemInFolder(targetPath)
+    return err
+  })
+
   // Custom window controls (used by Windows frameless titlebar)
   ipcMain.handle('window-minimize', () => mainWindow?.minimize())
   ipcMain.handle('window-maximize', () => {

desktop/src/main/preload.ts

@@ -6,6 +6,7 @@ contextBridge.exposeInMainWorld('electronAPI', {
   restartBackend: () => ipcRenderer.invoke('restart-backend'),
   selectDirectory: () => ipcRenderer.invoke('select-directory'),
   selectFile: (filters?: Electron.FileFilter[]) => ipcRenderer.invoke('select-file', filters),
+  openPath: (targetPath: string) => ipcRenderer.invoke('open-path', targetPath) as Promise<string>,
 
   // Each listener registrar returns an unsubscribe fn so renderers can clean
   // up on unmount / effect re-run and avoid accumulating duplicate handlers.

desktop/src/renderer/src/components/MessageBubble.tsx

@@ -1,5 +1,5 @@
 import React, { useState } from 'react'
-import { Copy, Check, RefreshCw, Pencil, Trash2, File as FileIcon, Sprout } from 'lucide-react'
+import { Copy, Check, RefreshCw, Trash2, File as FileIcon, Sprout } from 'lucide-react'
 import type { ChatMessage } from '../types'
 import { t } from '../i18n'
 import apiClient from '../api/client'
@@ -11,6 +11,9 @@ interface MessageBubbleProps {
   onRegenerate?: (id: string) => void
   onEdit?: (id: string) => void
   onDelete?: (msg: ChatMessage) => void
+  /** Fired when an inline image/video finishes loading, so the parent can
+   *  re-scroll to the bottom (async media changes bubble height after mount). */
+  onMediaLoad?: () => void
 }
 
 function fmtTime(ts: number): string {
@@ -36,7 +39,7 @@ const HoverAction: React.FC<{ onClick: () => void; title: string; danger?: boole
   </button>
 )
 
-const MessageBubble: React.FC<MessageBubbleProps> = ({ message, onRegenerate, onEdit, onDelete }) => {
+const MessageBubble: React.FC<MessageBubbleProps> = ({ message, onRegenerate, onEdit, onDelete, onMediaLoad }) => {
   const isUser = message.role === 'user'
   const [copied, setCopied] = useState(false)
 
@@ -46,6 +49,16 @@ const MessageBubble: React.FC<MessageBubbleProps> = ({ message, onRegenerate, on
     setTimeout(() => setCopied(false), 1800)
   }
 
+  // Open a sent file: prefer the local path via Electron (Finder / default
+  // app); fall back to the served URL in a browser when unavailable.
+  const openAttachment = (att: { abs_path?: string; preview_url?: string; file_path: string }) => {
+    if (att.abs_path && window.electronAPI?.openPath) {
+      window.electronAPI.openPath(att.abs_path)
+      return
+    }
+    window.open(apiClient.getFileUrl(att.preview_url || att.file_path), '_blank')
+  }
+
   if (isUser) {
     return (
       <div className="group flex flex-col items-end px-4 sm:px-6 py-2">
@@ -73,11 +86,9 @@ const MessageBubble: React.FC<MessageBubbleProps> = ({ message, onRegenerate, on
         </div>
         <div className="flex items-center gap-0.5 mt-1 opacity-0 group-hover:opacity-100 transition-opacity">
           <span className="text-[11px] text-content-tertiary mr-1">{fmtTime(message.timestamp)}</span>
-          {onEdit && message.userSeq != null && (
+          {/* Edit entry hidden: editing a past question cascade-deletes all
-            <HoverAction onClick={() => onEdit(message.id)} title={t('msg_edit')}>
+              subsequent turns, which surprises users. Kept off until we support
-              <Pencil size={13} />
+              non-destructive editing. */}
-            </HoverAction>
-          )}
           {onDelete && message.userSeq != null && (
             <HoverAction onClick={() => onDelete(message)} title={t('msg_delete')} danger>
               <Trash2 size={13} />
@@ -118,6 +129,42 @@ const MessageBubble: React.FC<MessageBubbleProps> = ({ message, onRegenerate, on
           {/* Final answer */}
           {message.content && <Markdown content={message.content} />}
 
+          {/* Media attachments sent via the `send` tool (images / files). */}
+          {message.attachments && message.attachments.length > 0 && (
+            <div className="flex flex-wrap gap-2 mt-2">
+              {message.attachments.map((att, i) =>
+                att.file_type === 'image' ? (
+                  <img
+                    key={i}
+                    src={apiClient.getFileUrl(att.preview_url || att.file_path)}
+                    alt={att.file_name}
+                    onLoad={() => onMediaLoad?.()}
+                    onClick={() => window.open(apiClient.getFileUrl(att.preview_url || att.file_path), '_blank')}
+                    className="max-w-[320px] w-full rounded-xl border border-default cursor-zoom-in"
+                  />
+                ) : att.file_type === 'video' ? (
+                  <video
+                    key={i}
+                    src={apiClient.getFileUrl(att.preview_url || att.file_path)}
+                    controls
+                    onLoadedData={() => onMediaLoad?.()}
+                    className="max-w-[360px] w-full rounded-xl border border-default"
+                  />
+                ) : (
+                  <button
+                    key={i}
+                    type="button"
+                    onClick={() => openAttachment(att)}
+                    className="flex items-center gap-1.5 px-3 py-2 bg-surface-2 rounded-xl text-xs text-content-secondary hover:text-content cursor-pointer"
+                  >
+                    <FileIcon size={13} />
+                    {att.file_name}
+                  </button>
+                )
+              )}
+            </div>
+          )}
+
           {showCursor && (
             <div className="flex items-center gap-1 py-0.5">
               <span className="typing-dot" />

desktop/src/renderer/src/pages/ChatPage.tsx

@@ -182,6 +182,13 @@ const ChatPage: React.FC<ChatPageProps> = ({ baseUrl }) => {
     [deleteMessage, activeId]
   )
 
+  // Inline images/videos load asynchronously and grow the bubble after mount,
+  // so a scroll triggered on message change fires before the final height is
+  // known. Re-scroll once media loads, but only while following the bottom.
+  const handleMediaLoad = useCallback(() => {
+    if (followBottomRef.current) scrollToBottom(false)
+  }, [scrollToBottom])
+
   const handleScroll = useCallback(
     async (e: React.UIEvent<HTMLDivElement>) => {
       const el = e.currentTarget
@@ -250,6 +257,7 @@ const ChatPage: React.FC<ChatPageProps> = ({ baseUrl }) => {
                 onRegenerate={handleRegenerate}
                 onEdit={handleEdit}
                 onDelete={handleDelete}
+                onMediaLoad={handleMediaLoad}
               />
             ))}
             <div ref={bottomRef} />

desktop/src/renderer/src/store/chatStore.ts

@@ -218,6 +218,31 @@ export const useChatStore = create<ChatState>((set, get) => {
           }))
           break
 
+        case 'image':
+        case 'file': {
+          // Media pushed by the `send` tool (file_to_send). `content` is either
+          // a backend /api/file?path=... URL or a passed-through http(s) URL.
+          const url = data.content || ''
+          if (!url) break
+          // Prefer the concrete media kind from the backend (image/video/...);
+          // fall back to the coarse SSE event type.
+          const kind = data.file_type || (data.type === 'image' ? 'image' : 'file')
+          const attType: Attachment['file_type'] =
+            kind === 'image' ? 'image' : kind === 'video' ? 'video' : 'file'
+          const att: Attachment = {
+            file_path: url,
+            file_name: data.file_name || 'file',
+            file_type: attType,
+            preview_url: url,
+            abs_path: data.abs_path,
+          }
+          updateMsg(sid, botId, (m) => ({
+            ...m,
+            attachments: [...(m.attachments || []), att],
+          }))
+          break
+        }
+
         case 'cancelled':
           updateMsg(sid, botId, (m) => ({ ...m, isCancelled: true }))
           break

desktop/src/renderer/src/types.ts

@@ -8,6 +8,8 @@ export interface ElectronAPI {
   restartBackend: () => Promise<boolean>
   selectDirectory: () => Promise<string | null>
   selectFile: (filters?: { name: string; extensions: string[] }[]) => Promise<string | null>
+  /** Open a local file with the OS default app. Resolves to '' on success. */
+  openPath: (targetPath: string) => Promise<string>
   // Listener registrars return an unsubscribe fn for cleanup.
   onBackendStatus: (callback: (data: BackendStatusEvent) => void) => () => void
   onBackendLog: (callback: (line: string) => void) => () => void
@@ -94,6 +96,9 @@ export interface Attachment {
   file_name: string
   file_type: 'image' | 'video' | 'file' | 'directory'
   preview_url?: string
+  /** Local absolute path (set for files sent via the `send` tool) so the
+   *  desktop client can open them directly with the OS default app. */
+  abs_path?: string
 }
 
 /** Live tool event during SSE streaming. */
@@ -137,6 +142,7 @@ export interface StreamEvent {
   execution_time?: number
   has_tool_calls?: boolean
   path?: string
+  abs_path?: string
   file_name?: string
   file_type?: string
   web_url?: string