Makefile

@@ -132,7 +132,6 @@ stop:
 	@-pkill -f "next start" 2>/dev/null || true
 	@-pkill -f "next-server" 2>/dev/null || true
 	@-pkill -f "next-server" 2>/dev/null || true
-	@-pkill -f "frontend/.next/standalone/server.js" 2>/dev/null || true
 	@-nginx -c $(PWD)/docker/nginx/nginx.local.conf -p $(PWD) -s quit 2>/dev/null || true
 	@sleep 1
 	@-pkill -9 nginx 2>/dev/null || true

frontend/package.json

@@ -75,8 +75,6 @@
     "nanoid": "^5.1.6",
     "next": "^16.1.7",
     "next-themes": "^0.4.6",
-    "nextra": "^4.6.1",
-    "nextra-theme-docs": "^4.6.1",
     "nuxt-og-image": "^5.1.13",
     "ogl": "^1.0.11",
     "react": "^19.0.0",

frontend/src/app/[lang]/docs/[[...mdxPath]]/page.tsx

@@ -1,7 +1,9 @@
-import { importPage } from "nextra/pages";
+import { generateStaticParamsFor, importPage } from "nextra/pages";
 
 import { useMDXComponents as getMDXComponents } from "../../../../mdx-components";
 
+export const generateStaticParams = generateStaticParamsFor("mdxPath");
+
 export async function generateMetadata(props) {
   const params = await props.params;
   const { metadata } = await importPage(params.mdxPath, params.lang);

frontend/src/app/[lang]/docs/layout.tsx

@@ -3,6 +3,7 @@ import { getPageMap } from "nextra/page-map";
 import { Footer, Layout } from "nextra-theme-docs";
 
 import { Header } from "@/components/landing/header";
+import { getLocaleByLang } from "@/core/i18n/locale";
 import "nextra-theme-docs/style.css";
 
 const footer = <Footer>MIT {new Date().getFullYear()} © Nextra.</Footer>;
@@ -26,11 +27,18 @@ function formatPageRoute(base: string, items: PageMapItem[]): PageMapItem[] {
 
 export default async function DocLayout({ children, params }) {
   const { lang } = await params;
+  const locale = getLocaleByLang(lang);
   const pages = await getPageMap(`/${lang}`);
 
   return (
     <Layout
-      navbar={<Header />}
+      navbar={
+        <Header
+          className="relative max-w-full px-10"
+          homeURL="/"
+          locale={locale}
+        />
+      }
       pageMap={formatPageRoute(`/${lang}/docs`, pages)}
       docsRepositoryBase="https://github.com/bytedance/deerflow/tree/main/frontend/src/app/content"
       footer={footer}

frontend/src/app/workspace/agents/[agent_name]/chats/[thread_id]/page.tsx

@@ -2,7 +2,7 @@
 
 import { BotIcon, PlusSquare } from "lucide-react";
 import { useParams, useRouter } from "next/navigation";
-import { useCallback } from "react";
+import { useCallback, useState } from "react";
 
 import type { PromptInputMessage } from "@/components/ai-elements/prompt-input";
 import { Button } from "@/components/ui/button";
@@ -11,7 +11,11 @@ import { ArtifactTrigger } from "@/components/workspace/artifacts";
 import { ChatBox, useThreadChat } from "@/components/workspace/chats";
 import { ExportTrigger } from "@/components/workspace/export-trigger";
 import { InputBox } from "@/components/workspace/input-box";
-import { MessageList } from "@/components/workspace/messages";
+import {
+  MessageList,
+  MESSAGE_LIST_DEFAULT_PADDING_BOTTOM,
+  MESSAGE_LIST_FOLLOWUPS_EXTRA_PADDING_BOTTOM,
+} from "@/components/workspace/messages";
 import { ThreadContext } from "@/components/workspace/messages/context";
 import { ThreadTitle } from "@/components/workspace/thread-title";
 import { TodoList } from "@/components/workspace/todo-list";
@@ -20,17 +24,15 @@ import { Tooltip } from "@/components/workspace/tooltip";
 import { useAgent } from "@/core/agents";
 import { useI18n } from "@/core/i18n/hooks";
 import { useNotification } from "@/core/notification/hooks";
-import { useLocalSettings } from "@/core/settings";
+import { useThreadSettings } from "@/core/settings";
 import { useThreadStream } from "@/core/threads/hooks";
 import { textOfMessage } from "@/core/threads/utils";
 import { env } from "@/env";
 import { cn } from "@/lib/utils";
 
-const MESSAGE_LIST_DEFAULT_PADDING_BOTTOM = 160;
-const MESSAGE_LIST_FOLLOWUPS_EXTRA_PADDING_BOTTOM = 120;
-
 export default function AgentChatPage() {
   const { t } = useI18n();
+  const [showFollowups, setShowFollowups] = useState(false);
   const router = useRouter();
 
   const { agent_name } = useParams<{
@@ -40,7 +42,7 @@ export default function AgentChatPage() {
   const { agent } = useAgent(agent_name);
 
   const { threadId, isNewThread, setIsNewThread } = useThreadChat();
-  const [settings, setSettings] = useLocalSettings();
+  const [settings, setSettings] = useThreadSettings(threadId);
 
   const { showNotification } = useNotification();
   const [thread, sendMessage] = useThreadStream({
@@ -84,12 +86,13 @@ export default function AgentChatPage() {
     await thread.stop();
   }, [thread]);
 
-  const messageListPaddingBottom =
+  const messageListPaddingBottom = showFollowups
-    MESSAGE_LIST_DEFAULT_PADDING_BOTTOM +
+    ? MESSAGE_LIST_DEFAULT_PADDING_BOTTOM +
-    MESSAGE_LIST_FOLLOWUPS_EXTRA_PADDING_BOTTOM;
+      MESSAGE_LIST_FOLLOWUPS_EXTRA_PADDING_BOTTOM
+    : undefined;
 
   return (
-    <ThreadContext.Provider value={{ thread, threadId }}>
+    <ThreadContext.Provider value={{ thread }}>
       <ChatBox threadId={threadId}>
         <div className="relative flex size-full min-h-0 justify-between">
           <header
@@ -163,10 +166,9 @@ export default function AgentChatPage() {
 
                 <InputBox
                   className={cn("bg-background/5 w-full -translate-y-4")}
+                  isNewThread={isNewThread}
                   threadId={threadId}
                   autoFocus={isNewThread}
-                  showWelcomeStyle={isNewThread}
-                  hasSubmitted={!isNewThread}
                   status={
                     thread.error
                       ? "error"
@@ -182,6 +184,7 @@ export default function AgentChatPage() {
                   }
                   disabled={env.NEXT_PUBLIC_STATIC_WEBSITE_ONLY === "true"}
                   onContextChange={(context) => setSettings("context", context)}
+                  onFollowupsVisibilityChange={setShowFollowups}
                   onSubmit={handleSubmit}
                   onStop={handleStop}
                 />

frontend/src/app/workspace/agents/new/page.tsx

@@ -186,7 +186,7 @@ export default function NewAgentPage() {
   // ── Step 2: chat ───────────────────────────────────────────────────────────
 
   return (
-    <ThreadContext.Provider value={{ thread, threadId }}>
+    <ThreadContext.Provider value={{ thread }}>
       <ArtifactsProvider>
         <div className="flex size-full flex-col">
           {header}

frontend/src/app/workspace/chats/[thread_id]/page.tsx

@@ -67,19 +67,10 @@ export default function ChatPage() {
   // 新逻辑：历史渲染和新会话仅由路由 /chats/new 控制，不再读取 isnew/is_chatting 参数。
   const shouldRenderHistory = !showWelcomeStyle;
   const createNewSession = useMemo(() => isNewThread, [isNewThread]);
-  const safeThreadId = useMemo(() => {
-    if (!threadId || threadId === "new") {
-      return undefined;
-    }
-    return threadId;
-  }, [threadId]);
 
   const streamThreadId = useMemo(() => {
-    if (isNewThread && createNewSession) {
+    return isNewThread && createNewSession ? undefined : threadId;
-      return undefined;
+  }, [createNewSession, isNewThread, threadId]);
-    }
-    return safeThreadId;
-  }, [createNewSession, isNewThread, safeThreadId]);
 
   const { showNotification } = useNotification();
 
@@ -88,7 +79,7 @@ export default function ChatPage() {
     skillError: selectedSkillError,
     clearSkillError: clearSelectedSkillError,
     isBootstrapping: isSelectedSkillBootstrapping,
-  } = useSelectedSkillListener({ threadId: safeThreadId ?? null });
+  } = useSelectedSkillListener({ threadId });
   // 对话行为控制器
   const [thread, sendMessage, isUploading] = useThreadStream({
     threadId: streamThreadId,
@@ -100,7 +91,7 @@ export default function ChatPage() {
       setIsNewThread(false);
       // if (!shouldStayOnNewRoute) {
         // Keep /new in history so router.back() can return to it.
-        router.replace(`/workspace/chats/${currentThreadId}?is_chatting=true`);
+        router.replace(`/workspace/chats/${currentThreadId}`);
       // }
       // history.pushState(null, "", pathOfThread(currentThreadId));
     },
@@ -504,7 +495,7 @@ export default function ChatPage() {
                   if (threadId && threadId !== "new") {
                     nextQuery.set("thread_id", threadId);
                   }
-                  router.replace(`/workspace/chats/${threadId}?is_chatting=false`);
+                  router.replace(`/workspace/chats/new?${nextQuery.toString()}`);
                 }}
               >
                 确定

frontend/src/components/workspace/chats/chat-box.tsx

@@ -133,7 +133,7 @@ const ChatBox: React.FC<{
             artifactPanelOpen ? "translate-x-0" : "translate-x-full",
           )}
         >
-          {selectedArtifact && threadId ? (
+          {selectedArtifact ? (
             <ArtifactFileDetail
               // className="size-full"
               filepath={selectedArtifact}
@@ -167,7 +167,7 @@ const ChatBox: React.FC<{
                     <ArtifactFileList
                       className="max-w-(--container-width-sm) p-4 pt-12"
                       files={thread.values.artifacts ?? []}
-                      threadId={threadId ?? ""}
+                      threadId={threadId}
                     />
                   </main>
                 </div>

frontend/src/components/workspace/chats/use-thread-chat.ts

@@ -1,7 +1,7 @@
 "use client";
 
 import { useParams, usePathname, useSearchParams } from "next/navigation";
-import { useEffect, useState } from "react";
+import { useCallback, useEffect, useState } from "react";
 
 
 export function useThreadChat() {
@@ -26,13 +26,53 @@ export function useThreadChat() {
   const rawPathThreadId = params?.thread_id ?? threadIdFromPathname;
 
   const isNewRoute = rawPathThreadId === "new";
-  const threadIdFromPathOrParams = isNewRoute
+  const threadIdFromPathOrParams:string = isNewRoute
-    ? normalizeThreadId(threadIdFromSearchParams)
+    ? threadIdFromSearchParams?? params.thread_id
-    : normalizeThreadId(rawPathThreadId);
+    : params.thread_id;
   // console.log("[useThreadChat] pathname", pathname);
   // console.log("[useThreadChat] params.thread_id", params?.thread_id);
   // console.log("[useThreadChat] threadIdFromPathname", threadIdFromPathname);
   // console.log("[useThreadChat] threadIdFromPath", threadIdFromPath);
+  // 持久化兜底：用于处理首屏水合或 params 时序问题。
+  const readStoredThreadId = () => {
+    if (typeof window === "undefined") {
+      return undefined;
+    }
+    const stored = window.sessionStorage.getItem("workspace.thread_id");
+    return isValidThreadId(stored) ? stored : undefined;
+  };
+
+  // 读取 query 的 thread_id（先用 hook，必要时用 window 兜底）。
+  const readQueryThreadId = () => {
+    const fromHook = threadIdFromSearchParams;
+    if (isValidThreadId(fromHook)) {
+      return fromHook;
+    }
+    if (typeof window === "undefined") {
+      return undefined;
+    }
+    const fromLocation = new URLSearchParams(window.location.search).get(
+      "thread_id",
+    );
+    if (isValidThreadId(fromLocation)) {
+      return fromLocation.trim();
+    }
+    return undefined;
+  };
+
+  const queryThreadIdFromParams = readQueryThreadId();
+  // console.log("[useThreadChat] query.thread_id", queryThreadIdFromParams);
+  // 归一化：当值为 "new" 时，替换为 query 中的 thread_id（如果存在）。
+  const normalizeThreadId = useCallback(
+    (value?: string | null) => {
+      if (!value) {
+        return undefined;
+      }
+      return value === "new" ? queryThreadIdFromParams : value;
+    },
+    [queryThreadIdFromParams],
+  );
+
   // New session is only controlled by `/workspace/chats/new`.
   const [isNewThread, setIsNewThread] = useState(() => isNewRoute);
 
@@ -42,7 +82,7 @@ export function useThreadChat() {
   // console.log("[useThreadChat] effectiveThreadIdFromPath", effectiveThreadIdFromPath);
 
   const [threadId, setThreadId] = useState<string>(() => {
-    return threadIdFromPathOrParams ?? "";
+    return threadIdFromPathOrParams;
   });
 
 
@@ -53,10 +93,11 @@ export function useThreadChat() {
     }
     setIsNewThread(isNewRoute);
     // Prefer path thread id, fall back to query thread_id when path is /new.
-    setThreadId(threadIdFromPathOrParams ?? "");
+    setThreadId(threadIdFromPathOrParams);
     setShowWelcomeStyle(isNewRoute || !isChattingFromQuery);
   }, [
     isNewRoute,
+    normalizeThreadId,
     pathname,
     searchParams,
     isChattingFromQuery,
@@ -73,11 +114,6 @@ export function useThreadChat() {
   };
 }
 
-function normalizeThreadId(value?: string | null): string | undefined {
-  if (!value) return undefined;
-  return isValidThreadId(value) ? value.trim() : undefined;
-}
-
 function isValidThreadId(value?: string | null): value is string {
   if (!value) return false;
   const normalized = value.trim().toLowerCase();

frontend/src/components/workspace/input-box.tsx

@@ -59,10 +59,10 @@ import {
 import { useI18n } from "@/core/i18n/hooks";
 import type {
   SelectedSkillPayloadItem,
+  SuggestionSkillChildren,
 } from "@/core/i18n/locales/types";
 import { POST_MESSAGE_TYPES, sendToParent } from "@/core/iframe-messages";
 import { useModels } from "@/core/models/hooks";
-import { bootstrapRemoteSkill } from "@/core/skills/api";
 import type { AgentThreadContext } from "@/core/threads";
 import { useIframeSkill } from "@/hooks/use-iframe-skill";
 import { cn } from "@/lib/utils";
@@ -429,7 +429,6 @@ export function InputBox({
 
       {showWelcomeStyle && !hasSubmitted && searchParams.get("mode") !== "skill" && (
         <SuggestionListContainer
-          threadId={threadId}
           sendSelectSkill={iframeSkill.sendSelectSkill}
         />
       )}
@@ -494,31 +493,26 @@ export function InputBox({
 
 // SuggestionList 容器
 function SuggestionListContainer({
-  threadId,
   sendSelectSkill,
 }: {
-  threadId: string;
   sendSelectSkill: (selectedSkills: SelectedSkillPayloadItem[]) => void;
 }) {
   return (
     <div className="absolute right-0 bottom-0 left-0 z-0 flex translate-y-full items-center justify-center pt-4">
-      <SuggestionList threadId={threadId} sendSelectSkill={sendSelectSkill} />
+      <SuggestionList sendSelectSkill={sendSelectSkill} />
     </div>
   );
 }
 
 // 快速选择skillbutton
 function SuggestionList({
-  threadId,
   sendSelectSkill,
 }: {
-  threadId: string;
   sendSelectSkill: (selectedSkills: SelectedSkillPayloadItem[]) => void;
 }) {
   const { t } = useI18n();
-  const searchParams = useSearchParams();
   const { textInput } = usePromptInputController();
-  const suggestions = t.inputBox.suggestions;
+  const suggestions =t.inputBox.suggestions
   const promptSuggestions = suggestions.filter(
     (
       suggestion,
@@ -531,35 +525,14 @@ function SuggestionList({
       suggestion: {
         prompt: string;
         skill_id?: string[];
-        children?: SelectedSkillPayloadItem[];
+        children?: SuggestionSkillChildren[];
         suggestion: string;
       },
     ) => {
-      const languageTypeRaw =
-        searchParams.get("languageType")?.trim() ??
-        searchParams.get("language_type")?.trim();
-      const languageType = languageTypeRaw ? Number(languageTypeRaw) : 0;
-      const bootstrapByIds = (ids: string[]) => {
-        const content_ids = Array.from(
-          new Set(
-            ids
-              .map((id) => Number(id))
-              .filter((id) => Number.isFinite(id) && id > 0),
-          ),
-        );
-        if (!threadId || content_ids.length === 0) return;
-        void bootstrapRemoteSkill({
-          thread_id: threadId,
-          content_ids,
-          language_type: languageType,
-          target_dir: "/mnt/user-data/uploads/skill",
-          clear_target: true,
-        });
-      };
-
       // 优先从 children 中提取 skill_id 数组，转换为 selectedSkills 发送给宿主页
       const childSkillIds = (suggestion.children ?? [])
-        .map((item) => String(item.id).trim())
+        .flatMap((item) => item.skill_id)
+        .map((item) => item.trim())
         .filter((id): id is string => Boolean(id));
       if (childSkillIds.length > 0) {
         sendSelectSkill(
@@ -568,7 +541,6 @@ function SuggestionList({
             name: suggestion.suggestion,
           })),
         );
-        bootstrapByIds(childSkillIds);
         return;
       }
       if (suggestion.skill_id && suggestion.skill_id.length > 0) {
@@ -578,7 +550,6 @@ function SuggestionList({
             name: suggestion.suggestion,
           })),
         );
-        bootstrapByIds(suggestion.skill_id);
         return;
       }
       // 原有逻辑
@@ -598,7 +569,7 @@ function SuggestionList({
         }
       }, 500);
     },
-    [textInput, sendSelectSkill, threadId, searchParams],
+    [textInput, sendSelectSkill],
   );
   return (
     <Suggestions className="min-h-16 w-fit items-start" data-testid="welcome-suggestions">

frontend/src/core/i18n/locales/zh-CN.ts

@@ -58,7 +58,7 @@ export const zhCN: Translations = {
   // Welcome
   welcome: {
     // TODO: 测试环境标识
-    greeting: "轻办公 · XClaw Tag：v3.2.0 --- Skill功能施工中 --- refactor(frontend): 将 SELECT_SKILL 重命名为 SELECT_SKILLS.",
+    greeting: "轻办公 · XClaw Tag：v3.1.0 fix：适配md图片的更多情况",
     description:
       "欢迎使用 🦌 DeerFlow，一个完全开源的超级智能体。通过内置和自定义的 Skills，\nDeerFlow 可以帮你搜索网络、分析数据，还能为你生成幻灯片、\n图片、视频、播客及网页等，几乎可以做任何事情。",
 
@@ -117,31 +117,31 @@ export const zhCN: Translations = {
         prompt:
           "为[主题/产品]撰写吸引人的自媒体文案，包括标题、正文和话题标签。",
         icon: PenLineIcon,
-        children: [{ id: "1245", name: "微信文章撰写" }],
+        children: [{ id: "1245", name: "自媒体文案" }],
       },
       {
         suggestion: "需求文档",
         prompt: "编写[项目/功能]的需求文档，包含功能描述、用户故事和验收标准。",
         icon: CompassIcon,
-        children: [{ id: "520", name: "分解功能产品需求文档" }],
+        children: [{ id: "520", name: "需求文档" }],
       },
       {
         suggestion: "使用指南",
         prompt: "编写[产品/功能]的使用指南，包含操作步骤、注意事项和常见问题。",
         icon: GraduationCapIcon,
-        children: [{ id: "409", name: "用户指南编写" }],
+        children: [{ id: "409", name: "使用指南" }],
       },
       {
         suggestion: "Excel数据分析",
         prompt: "对[Excel文件/数据]进行分析，生成数据洞察和可视化建议。",
         icon: MicroscopeIcon,
-        children: [{ id: "5", name: "数据分析" }],
+        children: [{ id: "5", name: "Excel数据分析" }],
       },
       {
         suggestion: "市场调研",
         prompt: "针对[行业/产品]进行市场调研，分析市场规模、竞品和趋势。",
         icon: ShapesIcon,
-        children: [{ id: "1216", name: "市场研究报告" }],
+        children: [{ id: "1216", name: "市场调研" }],
       },
     ],
     suggestionsCreate: [

frontend/src/core/iframe-messages.ts

@@ -12,7 +12,7 @@ export const POST_MESSAGE_TYPES = {
   // 会话是否处于聊天态
   IS_CHATTING: "isChatting",
   // 选择预定义 skill
-  SELECT_SKILLS: "selectedSkills",
+  SELECT_SKILL: "selectedSkills",
   // 打开 skill 选择对话框
   OPEN_SKILL_DIALOG: "openSkillDialog",
 } as const;
@@ -41,7 +41,7 @@ export interface IsChattingMessage {
 }
 
 export interface SelectSkillMessage {
-  type: typeof POST_MESSAGE_TYPES.SELECT_SKILLS;
+  type: typeof POST_MESSAGE_TYPES.SELECT_SKILL;
   selectedSkills: SelectedSkillPayloadItem[];
 }
 

frontend/src/core/threads/hooks.ts

@@ -193,17 +193,14 @@ export function useThreadStream({
       handleStreamStart(meta.thread_id);
       setOnStreamThreadId(meta.thread_id);
     },
-    onLangChainEvent:
+    onLangChainEvent(event) {
-      onToolEnd == null
+      if (event.event === "on_tool_end") {
-        ? undefined
+        listeners.current.onToolEnd?.({
-        : (event) => {
+          name: event.name,
-            if (event.event === "on_tool_end") {
+          data: event.data,
-              listeners.current.onToolEnd?.({
+        });
-                name: event.name,
+      }
-                data: event.data,
+    },
-              });
-            }
-          },
     onUpdateEvent(data) {
       const updates: Array<Partial<AgentThreadState> | null> = Object.values(
         data || {},
@@ -328,20 +325,15 @@ export function useThreadStream({
       }
       setOptimisticMessages(newOptimistic);
 
-      // For "new chat with prefilled thread_id" flows, calling onStart before
+      if (resolvedThreadId) {
-      // submit can trigger route switch too early, which causes the new page to
-      // fetch history before the thread/run is actually created.
-      // Let useStream.onCreated -> handleStreamStart drive onStart instead.
-      if (resolvedThreadId && !createNewSession) {
         _handleOnStart(resolvedThreadId);
       }
 
       let uploadedFileInfo: UploadedFileInfo[] = [];
 
       try {
-        // 新会话模式下，仅在本地已有历史消息时才重置旧线程。
+        // 新会话模式下，删除旧线程并创建同名新线程
-        // 对于全新 thread_id，避免多发一次 DELETE /threads/{id}（通常会 404）。
+        if (createNewSession && resolvedThreadId) {
-        if (createNewSession && resolvedThreadId && thread.messages.length > 0) {
           await apiClient.threads.delete(resolvedThreadId).catch(() => undefined);
         }
 

frontend/src/hooks/use-iframe-skill.ts

@@ -70,7 +70,7 @@ export function useIframeSkill(): UseIframeSkillReturn {
 
   // 发送选择预定义 skill
   const sendSelectSkill = useCallback((selectedSkills: SelectedSkillPayloadItem[]) => {
-    const message = { type: POST_MESSAGE_TYPES.SELECT_SKILLS, selectedSkills };
+    const message = { type: POST_MESSAGE_TYPES.SELECT_SKILL, selectedSkills };
     console.log("[useIframeSkill] sendSelectSkill:", message);
     sendToParent(message);
   }, []);
@@ -89,7 +89,7 @@ export function useIframeSkill(): UseIframeSkillReturn {
   const clearSkill = useCallback(() => {
     setSelectedSkill(null);
     // 发送空数组给主页，通知取消选择
-    const message = { type: POST_MESSAGE_TYPES.SELECT_SKILLS, selectedSkills: [] };
+    const message = { type: POST_MESSAGE_TYPES.SELECT_SKILL, selectedSkills: [] };
     console.log("[useIframeSkill] clearSkill, sending selectedSkills=[]:", message);
     sendToParent(message);
   }, []);

scripts/deploy-frontend-standalone.sh

@@ -1,72 +0,0 @@
-#!/usr/bin/env bash
-
-set -euo pipefail
-
-REPO_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
-cd "$REPO_ROOT"
-
-# Default deploy target (can be overridden by positional args)
-DEFAULT_REMOTE="root@111.228.39.147"
-DEFAULT_REMOTE_APP_DIR="/root/deerflow2"
-
-usage() {
-  cat <<'EOF'
-Usage:
-  ./scripts/deploy-frontend-standalone.sh [remote] [remote_app_dir]
-
-Arguments:
-  remote          SSH target, optional. Default: root@111.228.39.147
-  remote_app_dir  Remote deerflow2 root dir, optional. Default: /root/deerflow2
-
-Example:
-  ./scripts/deploy-frontend-standalone.sh
-  ./scripts/deploy-frontend-standalone.sh ubuntu@1.2.3.4 /opt/deerflow2
-
-Notes:
-  - Run this script on your local machine (build machine), not on the server.
-  - Requires: pnpm, rsync, ssh
-  - Script will run:
-      sudo systemctl daemon-reload
-      sudo systemctl restart deerflow.service
-EOF
-}
-
-if [ "${1:-}" = "-h" ] || [ "${1:-}" = "--help" ]; then
-  usage
-  exit 0
-fi
-
-if [ "$#" -gt 2 ]; then
-  usage
-  exit 1
-fi
-
-REMOTE="${1:-$DEFAULT_REMOTE}"
-REMOTE_APP_DIR="${2:-$DEFAULT_REMOTE_APP_DIR}"
-
-echo "==> Deploy target:"
-echo "    remote: $REMOTE"
-echo "    app dir: $REMOTE_APP_DIR"
-
-echo "==> Building frontend (standalone)..."
-pnpm -C frontend build
-
-echo "==> Uploading standalone server..."
-rsync -azP --delete --info=progress2 \
-  frontend/.next/standalone/ \
-  "$REMOTE:$REMOTE_APP_DIR/frontend/.next/standalone/"
-
-echo "==> Uploading static assets..."
-rsync -azP --info=progress2 \
-  frontend/.next/static/ \
-  "$REMOTE:$REMOTE_APP_DIR/frontend/.next/standalone/.next/static/"
-
-echo "==> Uploading public assets..."
-rsync -azP --info=progress2 \
-  frontend/public/ \
-  "$REMOTE:$REMOTE_APP_DIR/frontend/.next/standalone/public/"
-
-echo "==> Reloading systemd and restarting deerflow.service on remote host..."
-ssh "$REMOTE" "sudo systemctl daemon-reload && sudo systemctl restart deerflow.service"
-
-echo "==> Done."

scripts/serve.sh

@@ -38,43 +38,7 @@ else
         echo "Python is required to generate BETTER_AUTH_SECRET, but neither python3 nor python was found."
         exit 1
     fi
-    FRONTEND_STANDALONE_DIR="$REPO_ROOT/frontend/.next/standalone"
+    FRONTEND_CMD="env BETTER_AUTH_SECRET=$($PYTHON_BIN -c 'import secrets; print(secrets.token_hex(16))') pnpm run preview"
-    FRONTEND_STANDALONE_SERVER="$FRONTEND_STANDALONE_DIR/server.js"
-    FRONTEND_STANDALONE_STATIC_DIR="$FRONTEND_STANDALONE_DIR/.next/static"
-    FRONTEND_SOURCE_STATIC_DIR="$REPO_ROOT/frontend/.next/static"
-    FRONTEND_STANDALONE_PUBLIC_DIR="$FRONTEND_STANDALONE_DIR/public"
-    FRONTEND_SOURCE_PUBLIC_DIR="$REPO_ROOT/frontend/public"
-
-    if [ ! -f "$FRONTEND_STANDALONE_SERVER" ]; then
-        echo "✗ Frontend standalone server not found: $FRONTEND_STANDALONE_SERVER"
-        echo "  make start (prod) now uses prebuilt standalone assets and will not run local build."
-        echo "  Please prepare frontend artifacts first (example):"
-        echo "    pnpm -C frontend build"
-        exit 1
-    fi
-
-    # Align local runtime layout with deploy-frontend-standalone.sh:
-    # standalone + .next/static + public under frontend/.next/standalone/
-    if [ ! -d "$FRONTEND_STANDALONE_STATIC_DIR" ] && [ -d "$FRONTEND_SOURCE_STATIC_DIR" ]; then
-        mkdir -p "$FRONTEND_STANDALONE_DIR/.next"
-        cp -R "$FRONTEND_SOURCE_STATIC_DIR" "$FRONTEND_STANDALONE_STATIC_DIR"
-    fi
-    if [ ! -d "$FRONTEND_STANDALONE_PUBLIC_DIR" ] && [ -d "$FRONTEND_SOURCE_PUBLIC_DIR" ]; then
-        cp -R "$FRONTEND_SOURCE_PUBLIC_DIR" "$FRONTEND_STANDALONE_PUBLIC_DIR"
-    fi
-
-    if [ ! -d "$FRONTEND_STANDALONE_STATIC_DIR" ]; then
-        echo "✗ Missing standalone static assets: $FRONTEND_STANDALONE_STATIC_DIR"
-        echo "  Please ensure .next/static is available for standalone runtime."
-        exit 1
-    fi
-    if [ ! -d "$FRONTEND_STANDALONE_PUBLIC_DIR" ]; then
-        echo "✗ Missing standalone public assets: $FRONTEND_STANDALONE_PUBLIC_DIR"
-        echo "  Please ensure public/ is available for standalone runtime."
-        exit 1
-    fi
-
-    FRONTEND_CMD="env BETTER_AUTH_SECRET=$($PYTHON_BIN -c 'import secrets; print(secrets.token_hex(16))') HOSTNAME=0.0.0.0 PORT=3000 node server.js"
 fi
 
 # ── Stop existing services ────────────────────────────────────────────────────
@@ -84,7 +48,6 @@ pkill -f "langgraph dev" 2>/dev/null || true
 pkill -f "uvicorn app.gateway.app:app" 2>/dev/null || true
 pkill -f "next dev" 2>/dev/null || true
 pkill -f "next-server" 2>/dev/null || true
-pkill -f "frontend/.next/standalone/server.js" 2>/dev/null || true
 nginx -c "$REPO_ROOT/docker/nginx/nginx.local.conf" -p "$REPO_ROOT" -s quit 2>/dev/null || true
 sleep 1
 pkill -9 nginx 2>/dev/null || true
@@ -147,7 +110,6 @@ cleanup() {
     pkill -f "next dev" 2>/dev/null || true
     pkill -f "next start" 2>/dev/null || true
     pkill -f "next-server" 2>/dev/null || true
-    pkill -f "frontend/.next/standalone/server.js" 2>/dev/null || true
     # Kill nginx using the captured PID first (most reliable),
     # then fall back to pkill/killall for any stray nginx workers.
     if [ -n "${NGINX_PID:-}" ] && kill -0 "$NGINX_PID" 2>/dev/null; then
@@ -219,11 +181,7 @@ echo "Starting Gateway API..."
 echo "✓ Gateway API started on localhost:8001"
 
 echo "Starting Frontend..."
-if $DEV_MODE; then
+(cd frontend && $FRONTEND_CMD > ../logs/frontend.log 2>&1) &
-    (cd frontend && $FRONTEND_CMD > ../logs/frontend.log 2>&1) &
-else
-    (cd frontend/.next/standalone && $FRONTEND_CMD > ../../../logs/frontend.log 2>&1) &
-fi
 ./scripts/wait-for-port.sh 3000 120 "Frontend" || {
     echo "  See logs/frontend.log for details"
     tail -20 logs/frontend.log

skills/public/image-generation/.env.example

@@ -1,6 +0,0 @@
-# RunningHub API Configuration
-# Copy this file to .env and fill in your actual API key
-
-# RunningHub API Key for image generation
-# Get your API key from: https://www.runninghub.cn
-RUNNINGHUB_API_KEY=your_api_key_here

skills/public/image-generation/.gitignore

@@ -1,31 +0,0 @@
-# Environment variables
-.env
-.env.local
-.env.*.local
-
-# Python
-__pycache__/
-*.py[cod]
-*$py.class
-*.so
-.Python
-venv/
-env/
-ENV/
-
-# Output files
-*.jpg
-*.jpeg
-*.png
-*.webp
-outputs/
-
-# IDE
-.vscode/
-.idea/
-*.swp
-*.swo
-
-# OS
-.DS_Store
-Thumbs.db

skills/public/image-generation/SKILL.md

@@ -7,47 +7,14 @@ description: Use this skill when the user requests to generate, create, imagine,
 
 ## Overview
 
-This skill generates high-quality images using RunningHub API with structured prompts and a Python script. The workflow includes creating JSON-formatted prompts and executing image generation through asynchronous task submission.
+This skill generates high-quality images using structured prompts and a Python script. The workflow includes creating JSON-formatted prompts and executing image generation with optional reference images.
 
 ## Core Capabilities
 
 - Create structured JSON prompts for AIGC image generation
-- Generate images through RunningHub's Z-Image Turbo LoRA API
+- Support multiple reference images for style/composition guidance
-- Support asynchronous task submission and status polling
+- Generate images through automated Python script execution
 - Handle various image generation scenarios (character design, scenes, products, etc.)
-- Support multiple aspect ratios and output formats (PNG, JPEG, WEBP)
-
-## Configuration
-
-### API Key Setup
-
-This skill uses RunningHub API for image generation. You need to configure your API key before using the skill.
-
-**Option 1: Environment Variable (Recommended)**
-```bash
-# Set the RUNNINGHUB_API_KEY environment variable
-export RUNNINGHUB_API_KEY=your_api_key_here
-
-# Or on Windows:
-set RUNNINGHUB_API_KEY=your_api_key_here
-```
-
-**Option 2: .env File**
-1. Copy `.env.example` to `.env`:
-   ```bash
-   cp .env.example .env
-   ```
-2. Edit `.env` and add your API key:
-   ```
-   RUNNINGHUB_API_KEY=your_api_key_here
-   ```
-3. The `.env` file is automatically excluded from version control via `.gitignore`
-
-**Security Notes:**
-- Never commit `.env` files to version control
-- Never hardcode API keys in source code
-- Rotate your API keys if they are accidentally exposed
-- Get your API key from: https://www.runninghub.cn
 
 ## Workflow
 
@@ -71,20 +38,20 @@ Call the Python script:
 ```bash
 python /mnt/skills/public/image-generation/scripts/generate.py \
   --prompt-file /mnt/user-data/workspace/prompt-file.json \
-  --output-file /mnt/user-data/outputs/generated-image.jpg \
+  --reference-images /path/to/ref1.jpg /path/to/ref2.png \
+  --output-file /mnt/user-data/outputs/generated-image.jpg
   --aspect-ratio 16:9
 ```
 
 Parameters:
 
 - `--prompt-file`: Absolute path to JSON prompt file (required)
+- `--reference-images`: Absolute paths to reference images (optional, space-separated)
 - `--output-file`: Absolute path to output image file (required)
 - `--aspect-ratio`: Aspect ratio of the generated image (optional, default: 16:9)
 
 [!NOTE]
-- The script uses RunningHub API which requires `RUNNINGHUB_API_KEY` environment variable to be set
+Do NOT read the python file, just call it with the parameters.
-- Do NOT read the python file, just call it with the parameters
-- The script automatically handles task submission, status polling, and image download
 
 ## Character Generation Example
 
@@ -119,6 +86,40 @@ python /mnt/skills/public/image-generation/scripts/generate.py \
   --aspect-ratio 2:3
 ```
 
+With reference images:
+```json
+{
+  "characters": [{
+    "gender": "based on [Image 1]",
+    "age": "based on [Image 1]",
+    "ethnicity": "human from [Image 1] adapted to Star Wars universe",
+    "body_type": "based on [Image 1]",
+    "facial_features": "matching [Image 1] with slight weathered look from space travel",
+    "clothing": "Star Wars style outfit - worn leather jacket with utility vest, cargo pants with tactical pouches, scuffed boots, belt with holster",
+    "accessories": "blaster pistol on hip, comlink device on wrist, goggles pushed up on forehead, satchel with supplies, personal vehicle based on [Image 2]",
+    "era": "Star Wars universe, post-Empire era"
+  }],
+  "prompt": "Character inspired by [Image 1] standing next to a vehicle inspired by [Image 2] on a bustling alien planet street in Star Wars universe aesthetic. Character wearing worn leather jacket with utility vest, cargo pants with tactical pouches, scuffed boots, belt with blaster holster. The vehicle adapted to Star Wars aesthetic with weathered metal panels, repulsor engines, desert dust covering, parked on the street. Exotic alien marketplace street with multi-level architecture, weathered metal structures, hanging market stalls with colorful awnings, alien species walking by as background characters. Twin suns casting warm golden light, atmospheric dust particles in air, moisture vaporators visible in distance. Gritty lived-in Star Wars aesthetic, practical effects look, film grain texture, cinematic composition.",
+  "negative_prompt": "clean futuristic look, sterile environment, overly CGI appearance, fantasy medieval elements, Earth architecture, modern city",
+  "style": "Star Wars original trilogy aesthetic, lived-in universe, practical effects inspired, cinematic film look, slightly desaturated with warm tones",
+  "composition": "medium wide shot, character in foreground with alien street extending into background, environmental storytelling, rule of thirds",
+  "lighting": "warm golden hour lighting from twin suns, rim lighting on character, atmospheric haze, practical light sources from market stalls",
+  "color_palette": "warm sandy tones, ochre and sienna, dusty blues, weathered metals, muted earth colors with pops of alien market colors",
+  "technical": {
+    "aspect_ratio": "9:16",
+    "quality": "high",
+    "detail_level": "highly detailed with film-like texture"
+  }
+}
+```
+```bash
+python /mnt/skills/public/image-generation/scripts/generate.py \
+  --prompt-file /mnt/user-data/workspace/star-wars-scene.json \
+  --reference-images /mnt/user-data/uploads/character-ref.jpg /mnt/user-data/uploads/vehicle-ref.jpg \
+  --output-file /mnt/user-data/outputs/star-wars-scene-01.jpg \
+  --aspect-ratio 16:9
+```
+
 ## Common Scenarios
 
 Use different JSON schemas for different scenarios.
@@ -157,10 +158,30 @@ After generation:
 - Provide brief description of the generation result
 - Offer to iterate if adjustments needed
 
+## Tips: Enhancing Generation with Reference Images
+
+For scenarios where visual accuracy is critical, **use the `image_search` tool first** to find reference images before generation.
+
+**Recommended scenarios for using image_search tool:**
+- **Character/Portrait Generation**: Search for similar poses, expressions, or styles to guide facial features and body proportions
+- **Specific Objects or Products**: Find reference images of real objects to ensure accurate representation
+- **Architectural or Environmental Scenes**: Search for location references to capture authentic details
+- **Fashion and Clothing**: Find style references to ensure accurate garment details and styling
+
+**Example workflow:**
+1. Call the `image_search` tool to find suitable reference images:
+   ```
+   image_search(query="Japanese woman street photography 1990s", size="Large")
+   ```
+2. Download the returned image URLs to local files
+3. Use the downloaded images as `--reference-images` parameter in the generation script
+
+This approach significantly improves generation quality by providing the model with concrete visual guidance rather than relying solely on text descriptions.
+
 ## Notes
 
 - Always use English for prompts regardless of user's language
 - JSON format ensures structured, parsable prompts
+- Reference images enhance generation quality significantly
 - Iterative refinement is normal for optimal results
 - For character generation, include the detailed character object plus a consolidated prompt field
-- The script automatically polls task status and downloads the generated image

skills/public/image-generation/scripts/generate.py

@@ -1,14 +1,8 @@
 import base64
-import json
 import os
-import time
-from typing import List
 
 import requests
 from PIL import Image
-from dotenv import load_dotenv
-
-load_dotenv()
 
 
 def validate_image(image_path: str) -> bool:
@@ -23,171 +17,77 @@ def validate_image(image_path: str) -> bool:
     """
     try:
         with Image.open(image_path) as img:
-            img.verify()
+            img.verify()  # Verify that it's a valid image
+        # Re-open to check if it can be fully loaded (verify() may not catch all issues)
         with Image.open(image_path) as img:
-            img.load()
+            img.load()  # Force load the image data
         return True
     except Exception as e:
         print(f"Warning: Image '{image_path}' is invalid or corrupted: {e}")
         return False
 
 
-def submit_generation_task(prompt: str, aspect_ratio: str = "16:9", output_format: str = "png") -> str:
-    """
-    Submit image generation task to RunningHub API.
-    
-    Args:
-        prompt: Text prompt for image generation
-        aspect_ratio: Aspect ratio of the generated image
-        output_format: Output image format (png, jpeg, webp)
-        
-    Returns:
-        Task ID for tracking the generation
-    """
-    api_key = os.getenv("RUNNINGHUB_API_KEY")
-    if not api_key:
-        raise Exception("RUNNINGHUB_API_KEY environment variable is not set")
-    
-    url = "https://www.runninghub.cn/openapi/v2/rhart-image/z-image/turbo-lora"
-    
-    headers = {
-        "Authorization": f"Bearer {api_key}",
-        "Content-Type": "application/json",
-    }
-    
-    payload = {
-        "prompt": prompt,
-        "aspectRatio": aspect_ratio,
-        "lora_name": "Z-Image _ 清纯高颜值_脸模版V1.0.safetensors",
-        "lora_strength": 1,
-        "outputFormat": output_format
-    }
-    
-    response = requests.post(url, headers=headers, json=payload)
-    response.raise_for_status()
-    result = response.json()
-    
-    if result.get("status") not in ["QUEUED", "RUNNING", "SUCCESS"]:
-        raise Exception(f"Task submission failed: {result.get('errorMessage', 'Unknown error')}")
-    
-    return result.get("taskId")
-
-
-def query_task_status(task_id: str) -> dict:
-    """
-    Query the status of a generation task.
-    
-    Args:
-        task_id: Task ID to query
-        
-    Returns:
-        Task status information
-    """
-    api_key = os.getenv("RUNNINGHUB_API_KEY")
-    if not api_key:
-        raise Exception("RUNNINGHUB_API_KEY environment variable is not set")
-    
-    url = "https://www.runninghub.cn/openapi/v2/query"
-    
-    headers = {
-        "Authorization": f"Bearer {api_key}",
-        "Content-Type": "application/json",
-    }
-    
-    payload = {
-        "taskId": task_id
-    }
-    
-    response = requests.post(url, headers=headers, json=payload)
-    response.raise_for_status()
-    return response.json()
-
-
-def download_image(url: str, output_path: str) -> None:
-    """
-    Download image from URL and save to file.
-    
-    Args:
-        url: Image URL to download
-        output_path: Local path to save the image
-    """
-    response = requests.get(url, stream=True)
-    response.raise_for_status()
-    
-    with open(output_path, "wb") as f:
-        for chunk in response.iter_content(chunk_size=8192):
-            f.write(chunk)
-
-
 def generate_image(
     prompt_file: str,
-    reference_images: List[str],
+    reference_images: list[str],
     output_file: str,
     aspect_ratio: str = "16:9",
 ) -> str:
-    """
-    Generate image using RunningHub API.
-    
-    Args:
-        prompt_file: Path to JSON prompt file
-        reference_images: List of reference image paths (currently not supported by RunningHub API)
-        output_file: Output path for generated image
-        aspect_ratio: Aspect ratio of the generated image
-        
-    Returns:
-        Success message with output file path
-    """
     with open(prompt_file, "r", encoding="utf-8") as f:
-        prompt_data = json.load(f)
+        prompt = f.read()
+    parts = []
+    i = 0
     
-    if reference_images:
+    # Filter out invalid reference images
-        print("Note: RunningHub API does not support reference images in this version. Reference images will be ignored.")
+    valid_reference_images = []
-    
+    for ref_img in reference_images:
-    prompt_text = prompt_data.get("prompt", "")
+        if validate_image(ref_img):
-    if not prompt_text:
+            valid_reference_images.append(ref_img)
-        prompt_text = json.dumps(prompt_data, ensure_ascii=False)
-    
-    output_format = "png"
-    if output_file.lower().endswith(".jpg") or output_file.lower().endswith(".jpeg"):
-        output_format = "jpeg"
-    elif output_file.lower().endswith(".webp"):
-        output_format = "webp"
-    
-    print(f"Submitting generation task...")
-    task_id = submit_generation_task(prompt_text, aspect_ratio, output_format)
-    print(f"Task submitted successfully. Task ID: {task_id}")
-    
-    max_retries = 60
-    retry_interval = 2
-    
-    for attempt in range(max_retries):
-        print(f"Checking task status... (Attempt {attempt + 1}/{max_retries})")
-        status_result = query_task_status(task_id)
-        status = status_result.get("status")
-        
-        if status == "SUCCESS":
-            print("Task completed successfully!")
-            results = status_result.get("results", [])
-            if results and len(results) > 0:
-                image_url = results[0].get("url")
-                if image_url:
-                    print(f"Downloading image from: {image_url}")
-                    download_image(image_url, output_file)
-                    return f"Successfully generated image to {output_file}"
-                else:
-                    raise Exception("No image URL found in task results")
-            else:
-                raise Exception("No results found in task response")
-        elif status == "FAILED":
-            error_msg = status_result.get("errorMessage", "Unknown error")
-            raise Exception(f"Task failed: {error_msg}")
-        elif status in ["QUEUED", "RUNNING"]:
-            print(f"Task status: {status}. Waiting...")
-            time.sleep(retry_interval)
         else:
-            raise Exception(f"Unknown task status: {status}")
+            print(f"Skipping invalid reference image: {ref_img}")
     
-    raise Exception(f"Task did not complete within {max_retries * retry_interval} seconds")
+    if len(valid_reference_images) < len(reference_images):
+        print(f"Note: {len(reference_images) - len(valid_reference_images)} reference image(s) were skipped due to validation failure.")
+    
+    for reference_image in valid_reference_images:
+        i += 1
+        with open(reference_image, "rb") as f:
+            image_b64 = base64.b64encode(f.read()).decode("utf-8")
+        parts.append(
+            {
+                "inlineData": {
+                    "mimeType": "image/jpeg",
+                    "data": image_b64,
+                }
+            }
+        )
+
+    api_key = os.getenv("GEMINI_API_KEY")
+    if not api_key:
+        return "GEMINI_API_KEY is not set"
+    response = requests.post(
+        "https://generativelanguage.googleapis.com/v1beta/models/gemini-3-pro-image-preview:generateContent",
+        headers={
+            "x-goog-api-key": api_key,
+            "Content-Type": "application/json",
+        },
+        json={
+            "generationConfig": {"imageConfig": {"aspectRatio": aspect_ratio}},
+            "contents": [{"parts": [*parts, {"text": prompt}]}],
+        },
+    )
+    response.raise_for_status()
+    json = response.json()
+    parts: list[dict] = json["candidates"][0]["content"]["parts"]
+    image_parts = [part for part in parts if part.get("inlineData", False)]
+    if len(image_parts) == 1:
+        base64_image = image_parts[0]["inlineData"]["data"]
+        # Save the image to a file
+        with open(output_file, "wb") as f:
+            f.write(base64.b64decode(base64_image))
+        return f"Successfully generated image to {output_file}"
+    else:
+        raise Exception("Failed to generate image")
 
 
 if __name__ == "__main__":

skills/public/video-generation/.gitignore

@@ -1,16 +0,0 @@
-.env
-*.mp4
-*.avi
-*.mov
-*.mkv
-outputs/
-workspace/
-__pycache__/
-*.pyc
-*.pyo
-*.pyd
-.Python
-*.so
-*.egg-info/
-dist/
-build/

skills/public/video-generation/SETUP.md

@@ -1,93 +0,0 @@
-# Video Generation Skill - Setup Guide
-
-## Quick Start
-
-1. **Install Dependencies**
-   ```bash
-   pip install requests python-dotenv
-   ```
-
-2. **Configure API Key**
-   
-   Create a `.env` file in the project root directory:
-   ```env
-   RUNNINGHUB_API_KEY=your_api_key_here
-   ```
-   
-   Or set it as an environment variable:
-   ```bash
-   # Windows PowerShell
-   $env:RUNNINGHUB_API_KEY="your_api_key_here"
-   
-   # Linux/Mac
-   export RUNNINGHUB_API_KEY="your_api_key_here"
-   ```
-
-3. **Generate a Video**
-   ```bash
-   python scripts/generate.py --prompt-file workspace/your-prompt.json --output-file outputs/video.mp4 --duration 5
-   ```
-
-## Parameters
-
-- `--prompt-file`: Path to JSON prompt file (required)
-- `--output-file`: Output video file path (required)
-- `--aspect-ratio`: Video aspect ratio (optional, default: 16:9)
-- `--duration`: Video duration in seconds (optional, default: 5, range: 1-16)
-
-## Getting API Key
-
-To use this skill, you need a RunningHub API key:
-
-1. Visit [RunningHub](https://www.runninghub.cn/)
-2. Sign up for an account
-3. Get your API key from the dashboard
-4. Add it to your `.env` file
-
-## Example Prompt
-
-Create a JSON file with your video description:
-
-```json
-{
-  "title": "Your Video Title",
-  "description": "Description of what you want to generate",
-  "visual": {
-    "scene": "Scene description",
-    "elements": ["element1", "element2"],
-    "colors": "Color palette",
-    "lighting": "Lighting description"
-  },
-  "camera": {
-    "movement": "Camera movement",
-    "focus": "Focus description"
-  },
-  "audio": {
-    "background": "Background music description",
-    "effects": "Sound effects description"
-  }
-}
-```
-
-## Notes
-
-- The `.env` file is already in `.gitignore` and won't be committed to version control
-- Never share your API key or commit it to public repositories
-- The script automatically loads environment variables from `.env` file
-- Video generation may take several minutes depending on the complexity
-
-## Troubleshooting
-
-**Error: RUNNINGHUB_API_KEY is not set**
-- Make sure you've created the `.env` file with your API key
-- Or set the environment variable before running the script
-
-**Error: Failed to submit task**
-- Check that your API key is valid
-- Ensure you have sufficient credits in your RunningHub account
-- Verify your internet connection
-
-**Video generation takes too long**
-- This is normal for AI video generation
-- The script will automatically poll for status until completion
-- You can check the RunningHub dashboard for task progress

skills/public/video-generation/SKILL.md

@@ -7,14 +7,13 @@ description: Use this skill when the user requests to generate, create, or imagi
 
 ## Overview
 
-This skill generates high-quality videos using structured prompts and a Python script. The workflow includes creating JSON-formatted prompts and executing video generation through RunningHub API.
+This skill generates high-quality videos using structured prompts and a Python script. The workflow includes creating JSON-formatted prompts and executing video generation with optional reference image.
 
 ## Core Capabilities
 
 - Create structured JSON prompts for AIGC video generation
-- Generate videos through RunningHub Vidu model (text-to-video-q3-turbo)
+- Support reference image as guidance or the first/last frame of the video
-- Support up to 16 seconds video generation with audio
+- Generate videos through automated Python script execution
-- Automatic camera switching and dialogue generation
 
 ## Workflow
 
@@ -22,16 +21,21 @@ This skill generates high-quality videos using structured prompts and a Python s
 
 When a user requests video generation, identify:
 
-- Subject/content: What should be in the video
+- Subject/content: What should be in the image
 - Style preferences: Art style, mood, color palette
-- Technical specs: Aspect ratio, resolution, duration
+- Technical specs: Aspect ratio, composition, lighting
-- Audio requirements: Background music, dialogue, sound effects
+- Reference image: Any image to guide generation
+- You don't need to check the folder under `/mnt/user-data`
 
 ### Step 2: Create Structured Prompt
 
 Generate a structured JSON file in `/mnt/user-data/workspace/` with naming pattern: `{descriptive-name}.json`
 
-The prompt should include visual descriptions, camera movements, and audio specifications in a natural language format.
+### Step 3: Create Reference Image (Optional when image-generation skill is available)
+
+Generate reference image for the video generation.
+
+- If only 1 image is provided, use it as the guided frame of the video
 
 ### Step 3: Execute Generation
 
@@ -39,6 +43,7 @@ Call the Python script:
 ```bash
 python /mnt/skills/public/video-generation/scripts/generate.py \
   --prompt-file /mnt/user-data/workspace/prompt-file.json \
+  --reference-images /path/to/ref1.jpg \
   --output-file /mnt/user-data/outputs/generated-video.mp4 \
   --aspect-ratio 16:9
 ```
@@ -46,28 +51,20 @@ python /mnt/skills/public/video-generation/scripts/generate.py \
 Parameters:
 
 - `--prompt-file`: Absolute path to JSON prompt file (required)
-- `--output-file`: Absolute path to output video file (required)
+- `--reference-images`: Absolute paths to reference image (optional)
-- `--aspect-ratio`: Aspect ratio of the generated video (optional, default: 16:9)
+- `--output-file`: Absolute path to output image file (required)
+- `--aspect-ratio`: Aspect ratio of the generated image (optional, default: 16:9)
 
 [!NOTE]
 Do NOT read the python file, instead just call it with the parameters.
 
-## Environment Variables
-
-Set the following environment variable before running the script:
-
-- `RUNNINGHUB_API_KEY`: Your RunningHub API key
-
-Example:
-```bash
-export RUNNINGHUB_API_KEY=a73d0e93afb4432c978e5bff30b7517e
-```
-
 ## Video Generation Example
 
 User request: "Generate a short video clip depicting the opening scene from "The Chronicles of Narnia: The Lion, the Witch and the Wardrobe"
 
-Step 1: Create a JSON prompt file with the following content:
+Step 1: Search for the opening scene of "The Chronicles of Narnia: The Lion, the Witch and the Wardrobe" online
+
+Step 2: Create a JSON prompt file with the following content:
 
 ```json
 {
@@ -111,11 +108,16 @@ Step 1: Create a JSON prompt file with the following content:
 }
 ```
 
-Step 2: Use the generate.py script to generate the video
+Step 3: Use the image-generation skill to generate the reference image
+
+Load the image-generation skill and generate a single reference image `narnia-farewell-scene-01.jpg` according to the skill.
+
+Step 4: Use the generate.py script to generate the video
 ```bash
 python /mnt/skills/public/video-generation/scripts/generate.py \
   --prompt-file /mnt/user-data/workspace/narnia-farewell-scene.json \
-  --output-file /mnt/user-data/outputs/narnia-farewell-scene.mp4 \
+  --reference-images /mnt/user-data/outputs/narnia-farewell-scene-01.jpg \
+  --output-file /mnt/user-data/outputs/narnia-farewell-scene-01.mp4 \
   --aspect-ratio 16:9
 ```
 > Do NOT read the python file, just call it with the parameters.
@@ -125,7 +127,7 @@ python /mnt/skills/public/video-generation/scripts/generate.py \
 After generation:
 
 - Videos are typically saved in `/mnt/user-data/outputs/`
-- Share generated videos with user using `present_files` tool
+- Share generated videos (come first) with user as well as generated image if applicable, using `present_files` tool
 - Provide brief description of the generation result
 - Offer to iterate if adjustments needed
 
@@ -133,7 +135,5 @@ After generation:
 
 - Always use English for prompts regardless of user's language
 - JSON format ensures structured, parsable prompts
-- RunningHub Vidu model supports up to 16 seconds video generation
+- Reference image enhance generation quality significantly
-- Audio is automatically generated including dialogue and sound effects
-- The model has "director thinking" capability for automatic camera switching
 - Iterative refinement is normal for optimal results

skills/public/video-generation/scripts/generate.py

@@ -1,83 +1,75 @@
+import base64
 import os
 import time
-from typing import List
 
 import requests
-from dotenv import load_dotenv
-
-load_dotenv()
 
 
 def generate_video(
     prompt_file: str,
-    reference_images: List[str],
+    reference_images: list[str],
     output_file: str,
     aspect_ratio: str = "16:9",
-    duration: str = "5",
 ) -> str:
     with open(prompt_file, "r", encoding="utf-8") as f:
         prompt = f.read()
-    
+    referenceImages = []
-    api_key = os.getenv("RUNNINGHUB_API_KEY")
+    i = 0
-    if not api_key:
+    json = {
-        return "RUNNINGHUB_API_KEY is not set"
+        "instances": [{"prompt": prompt}],
-    
-    json_data = {
-        "prompt": prompt,
-        "style": "general",
-        "aspectRatio": aspect_ratio,
-        "resolution": "720p",
-        "duration": duration,
-        "audio": True
     }
-    
+    for reference_image in reference_images:
+        i += 1
+        with open(reference_image, "rb") as f:
+            image_b64 = base64.b64encode(f.read()).decode("utf-8")
+        referenceImages.append(
+            {
+                "image": {"mimeType": "image/jpeg", "bytesBase64Encoded": image_b64},
+                "referenceType": "asset",
+            }
+        )
+    if i > 0:
+        json["instances"][0]["referenceImages"] = referenceImages
+    api_key = os.getenv("GEMINI_API_KEY")
+    if not api_key:
+        return "GEMINI_API_KEY is not set"
     response = requests.post(
-        "https://www.runninghub.cn/openapi/v2/vidu/text-to-video-q3-turbo",
+        "https://generativelanguage.googleapis.com/v1beta/models/veo-3.1-generate-preview:predictLongRunning",
         headers={
-            "Authorization": f"Bearer {api_key}",
+            "x-goog-api-key": api_key,
             "Content-Type": "application/json",
         },
-        json=json_data,
+        json=json,
     )
-    
+    json = response.json()
-    response_json = response.json()
+    operation_name = json["name"]
-    
-    if "taskId" not in response_json:
-        return f"Failed to submit task: {response_json}"
-    
-    task_id = response_json["taskId"]
-    
     while True:
-        response = requests.post(
+        response = requests.get(
-            "https://www.runninghub.cn/openapi/v2/query",
+            f"https://generativelanguage.googleapis.com/v1beta/{operation_name}",
             headers={
-                "Authorization": f"Bearer {api_key}",
+                "x-goog-api-key": api_key,
-                "Content-Type": "application/json",
             },
-            json={"taskId": task_id},
         )
-        
+        json = response.json()
-        response_json = response.json()
+        if json.get("done", False):
-        status = response_json.get("status")
+            sample = json["response"]["generateVideoResponse"]["generatedSamples"][0]
-        
+            url = sample["video"]["uri"]
-        if status == "SUCCESS":
+            download(url, output_file)
-            results = response_json.get("results", [])
+            break
-            if results and len(results) > 0:
-                url = results[0].get("url")
-                if url:
-                    download(url, output_file)
-                    break
-        elif status == "FAILED":
-            error_message = response_json.get("errorMessage", "Unknown error")
-            return f"Video generation failed: {error_message}"
-        
         time.sleep(3)
-    
     return f"The video has been generated successfully to {output_file}"
 
 
 def download(url: str, output_file: str):
-    response = requests.get(url)
+    api_key = os.getenv("GEMINI_API_KEY")
+    if not api_key:
+        return "GEMINI_API_KEY is not set"
+    response = requests.get(
+        url,
+        headers={
+            "x-goog-api-key": api_key,
+        },
+    )
     with open(output_file, "wb") as f:
         f.write(response.content)
 
@@ -85,28 +77,28 @@ def download(url: str, output_file: str):
 if __name__ == "__main__":
     import argparse
 
-    parser = argparse.ArgumentParser(description="Generate videos using RunningHub API")
+    parser = argparse.ArgumentParser(description="Generate videos using Gemini API")
     parser.add_argument(
         "--prompt-file",
         required=True,
         help="Absolute path to JSON prompt file",
     )
+    parser.add_argument(
+        "--reference-images",
+        nargs="*",
+        default=[],
+        help="Absolute paths to reference images (space-separated)",
+    )
     parser.add_argument(
         "--output-file",
         required=True,
-        help="Output path for generated video",
+        help="Output path for generated image",
     )
     parser.add_argument(
         "--aspect-ratio",
         required=False,
         default="16:9",
-        help="Aspect ratio of the generated video",
+        help="Aspect ratio of the generated image",
-    )
-    parser.add_argument(
-        "--duration",
-        required=False,
-        default="5",
-        help="Duration of the generated video in seconds (1-16)",
     )
 
     args = parser.parse_args()
@@ -115,10 +107,9 @@ if __name__ == "__main__":
         print(
             generate_video(
                 args.prompt_file,
-                [],
+                args.reference_images,
                 args.output_file,
                 args.aspect_ratio,
-                args.duration,
             )
         )
     except Exception as e: