feat(home): add AI story expand button and modal

lib/prompts/novel-promotion/ai_story_expand.en.txt

@@ -0,0 +1,39 @@
+You are a professional screenwriting master, skilled at expanding short ideas, keywords, or outlines into complete story/script content.
+
+## Your Task
+
+Based on the user's creative input (which may be keywords, short descriptions, story outlines, or content to rewrite), create a complete, high-quality story suitable for short drama production.
+
+## Writing Requirements
+
+### Content Quality
+1. The story must have a clear beginning, development, and ending (or climactic suspense)
+2. Characters should have distinct personality traits and motivations
+3. Scene descriptions should be specific and visually compelling, suitable for storyboard conversion
+4. Dialogue should be natural, tension-filled, and drive the plot forward
+5. Pacing should be tight, avoiding lengthy background exposition
+
+### Format Requirements
+1. Use third-person perspective narration
+2. Separate scene transitions with blank lines
+3. Briefly introduce characters when they first appear
+4. Mark dialogue with quotation marks and identify the speaker
+5. Include action and expression descriptions at key visual moments
+
+### Length Control (⚠️ Most Important)
+- Goal: generate story content suitable for a 1-2 minute short film
+- Total length must be strictly between 300-800 words, never exceeding 800 words
+- Short keyword input: generate 400-600 word stories
+- Outline input: keep overall length to 500-800 words, expand each point concisely
+- Prefer concise over verbose; every sentence should be visual and dramatic
+- Characters and scenes should follow naturally from the user's input; do not invent characters or scenes the user did not mention
+
+### Prohibited
+- Do not output any non-story content (such as "Here is the generated story" etc.)
+- Do not output titles or chapter numbers
+- Do not use markdown formatting
+- Output story text directly
+
+## User Input
+
+{input}

lib/prompts/novel-promotion/ai_story_expand.zh.txt

@@ -0,0 +1,39 @@
+你是一个专业的影视剧本创作大师，擅长将简短的创意、关键词或大纲扩展为完整的故事/剧本内容。
+
+## 你的任务
+
+根据用户提供的创意输入（可能是关键词、简短描述、故事大纲或需要改写的内容），创作一段完整的、高质量的故事内容，用于后续的影视短剧制作。
+
+## 创作要求
+
+### 内容质量
+1. 故事必须有清晰的开头、发展和结尾（或高潮悬念）
+2. 角色要有鲜明的性格特征和行为动机
+3. 场景描写要具体、画面感强，适合转化为分镜画面
+4. 对话要自然、有张力，推动剧情发展
+5. 情节节奏紧凑，避免冗长的背景铺陈
+
+### 格式要求
+1. 使用第三人称视角叙述
+2. 场景转换时用空行分隔
+3. 角色名称在首次出现时需要简短介绍
+4. 对话使用引号标注，并注明说话者
+5. 在关键画面处加入动作和表情描写
+
+### 篇幅控制（⚠️ 最重要）
+- 目标：生成适合 1-2 分钟影视短片的故事内容
+- 总篇幅严格控制在 300-800 字之间，绝不超过 800 字
+- 简短关键词输入：生成 400-600 字的故事
+- 大纲输入：整体控制在 500-800 字，每个要点精炼展开
+- 宁可精炼也不要冗长，每一句话都要有画面感和戏剧张力
+- 角色和场景数量根据用户输入自然决定，不要额外发明用户未提及的角色或场景
+
+### 禁止事项
+- 不要输出任何非故事内容（如"以下是生成的故事"等说明文字）
+- 不要输出标题、章节号
+- 不要使用 markdown 格式
+- 直接输出故事正文内容
+
+## 用户输入
+
+{input}

messages/en/home.json

@@ -11,5 +11,15 @@
     "minutesAgo": "{n}m ago",
     "hoursAgo": "{n}h ago",
     "daysAgo": "{n}d ago"
+  },
+  "aiWrite": {
+    "trigger": "AI Write",
+    "modalTitle": "AI Writing Assistant",
+    "modalSubtitle": "Enter your idea and let AI generate a complete story",
+    "inputLabel": "Enter your creative idea",
+    "placeholder": "Enter keywords, story outline, or a short idea...\n\ne.g.\n• ancient palace, revenge, mystery, female lead\n• Act 1: The heroine returns to the capital; Act 2: A chance encounter at the royal banquet",
+    "hint": "💡 You can enter keywords, story outlines, or creative descriptions. AI will expand your input into a complete story",
+    "startAiWrite": "Start AI Writing",
+    "cancel": "Cancel"
   }
 }

messages/en/progress.json

@@ -65,6 +65,7 @@
     "screenplayConvert": "Screenplay conversion",
     "voiceAnalyze": "Voice line analysis",
     "analyzeGlobal": "Global analysis",
+    "aiStoryExpand": "AI story expansion",
     "aiModifyAppearance": "Character description modify",
     "aiModifyLocation": "Location description modify",
     "aiModifyShotPrompt": "Shot prompt modify",
@@ -104,6 +105,8 @@
     "scriptToStoryboardStep": "Execute script-to-storyboard step",
     "scriptToStoryboardPersist": "Persist script-to-storyboard output",
     "scriptToStoryboardPersistDone": "Storyboard and voice output persisted",
+    "aiStoryExpandPrepare": "Prepare AI story expansion",
+    "aiStoryExpandDone": "AI story expansion completed",
     "insertPanelGenerateText": "Generate inserted panel text",
     "insertPanelPersist": "Persist inserted panel",
     "pollingExternal": "Waiting for external service",

messages/zh/home.json

@@ -11,5 +11,15 @@
     "minutesAgo": "{n}分钟前",
     "hoursAgo": "{n}小时前",
     "daysAgo": "{n}天前"
+  },
+  "aiWrite": {
+    "trigger": "AI 帮我写",
+    "modalTitle": "AI 创作助手",
+    "modalSubtitle": "输入你的创意，让 AI 帮你生成完整故事",
+    "inputLabel": "输入你的创意内容",
+    "placeholder": "输入关键词、故事大纲或简短创意...\n\n例如：\n• 古代宫廷 复仇 悬疑 女主角\n• 第一幕：女主回到京城，暗访旧宅；第二幕：宫廷宴会偶遇仇人之子",
+    "hint": "💡 可以输入关键词、故事大纲、创意描述，AI 会根据你的输入扩展生成完整的故事内容",
+    "startAiWrite": "开始 AI 创作",
+    "cancel": "取消"
   }
 }

messages/zh/progress.json

@@ -65,6 +65,7 @@
     "screenplayConvert": "剧本转换",
     "voiceAnalyze": "台词分析",
     "analyzeGlobal": "全局分析",
+    "aiStoryExpand": "AI 故事扩写",
     "aiModifyAppearance": "角色描述修改",
     "aiModifyLocation": "场景描述修改",
     "aiModifyShotPrompt": "镜头提示词修改",
@@ -104,6 +105,8 @@
     "scriptToStoryboardStep": "执行分镜生成步骤",
     "scriptToStoryboardPersist": "保存分镜结果",
     "scriptToStoryboardPersistDone": "分镜与台词结果已保存",
+    "aiStoryExpandPrepare": "准备故事扩写参数",
+    "aiStoryExpandDone": "故事扩写已完成",
     "insertPanelGenerateText": "生成插入镜头文本",
     "insertPanelPersist": "保存插入镜头",
     "pollingExternal": "等待外部服务返回",

src/app/[locale]/home/page.tsx

@@ -13,11 +13,13 @@ import { RatioSelector, StyleSelector } from '@/components/selectors/RatioStyleS
 import { ART_STYLES, VIDEO_RATIOS } from '@/lib/constants'
 import { Link, useRouter } from '@/i18n/navigation'
 import { apiFetch } from '@/lib/api-fetch'
+import { expandHomeStory } from '@/lib/home/ai-story-expand'
 import { createHomeProjectLaunch } from '@/lib/home/create-project-launch'
 import {
   HOME_QUICK_START_MIN_ROWS,
   resolveTextareaTargetHeight,
 } from '@/lib/home/quick-start-textarea'
+import AiWriteModal from '@/components/home/AiWriteModal'
 
 interface ProjectStats {
   episodes: number
@@ -50,6 +52,8 @@ export default function HomePage() {
   const [videoRatio, setVideoRatio] = useState('9:16')
   const [artStyle, setArtStyle] = useState('american-comic')
   const [createLoading, setCreateLoading] = useState(false)
+  const [aiWriteOpen, setAiWriteOpen] = useState(false)
+  const [aiWriteLoading, setAiWriteLoading] = useState(false)
   const textareaRef = useRef<HTMLTextAreaElement>(null)
   const textareaMinHeightRef = useRef<number | null>(null)
 
@@ -148,6 +152,26 @@ export default function HomePage() {
     }
   }
 
+  // AI 帮我写 — 直接生成文本并回填首页输入框
+  const handleAiWriteStart = async (prompt: string) => {
+    if (aiWriteLoading) return
+    setAiWriteLoading(true)
+    try {
+      const result = await expandHomeStory({
+        apiFetch,
+        prompt,
+      })
+
+      setInputValue(result.expandedText)
+      setAiWriteOpen(false)
+    } catch (error) {
+      const message = error instanceof Error ? error.message : 'Failed'
+      window.alert(message)
+    } finally {
+      setAiWriteLoading(false)
+    }
+  }
+
   // 比例选项（带推荐标签）
   const ratioOptions = useMemo(
     () => VIDEO_RATIOS.map((r) => ({ ...r, recommended: r.value === '9:16' })),
@@ -251,7 +275,7 @@ export default function HomePage() {
               className="w-full bg-transparent border-none outline-none text-[var(--glass-text-primary)] placeholder:text-[var(--glass-text-tertiary)] text-base resize-none p-5 pb-3 custom-scrollbar"
             />
 
-            {/* 底部工具栏：比例 + 风格 + 创建按钮 */}
+            {/* 底部工具栏：比例 + 风格 + AI帮我写 + 创建按钮 */}
             <div className="flex items-end gap-3 px-5 pb-4">
               <div className="flex items-center gap-3 flex-1 min-w-0">
                 <div className="w-[160px] flex-shrink-0">
@@ -269,6 +293,23 @@ export default function HomePage() {
                   />
                 </div>
               </div>
+              <button
+                onClick={() => setAiWriteOpen(true)}
+                disabled={createLoading}
+                className="glass-btn-base px-4 py-2.5 text-sm flex-shrink-0 border border-[var(--glass-stroke-strong)] hover:border-[var(--glass-tone-info-fg)]/40 transition-all flex items-center gap-1.5"
+              >
+                <AppIcon name="sparkles" className="w-4 h-4 text-[#7c3aed]" />
+                <span
+                  className="font-medium"
+                  style={{
+                    background: 'linear-gradient(135deg, #3b82f6, #7c3aed)',
+                    WebkitBackgroundClip: 'text',
+                    WebkitTextFillColor: 'transparent',
+                  }}
+                >
+                  {t('aiWrite.trigger')}
+                </span>
+              </button>
               <button
                 onClick={() => void handleCreate()}
                 disabled={!inputValue.trim() || createLoading}
@@ -280,6 +321,14 @@ export default function HomePage() {
             </div>
           </div>
         </div>
+        {/* AI 帮我写模态框 */}
+        <AiWriteModal
+          open={aiWriteOpen}
+          loading={aiWriteLoading}
+          onClose={() => setAiWriteOpen(false)}
+          onStart={(prompt) => void handleAiWriteStart(prompt)}
+          t={(key: string) => t(`aiWrite.${key}`)}
+        />
       </main>
 
       {/* 最近项目 */}

src/app/api/user/ai-story-expand/route.ts

@@ -0,0 +1,48 @@
+import { createHash } from 'crypto'
+import { NextRequest } from 'next/server'
+import { requireUserAuth, isErrorResponse } from '@/lib/api-auth'
+import { apiHandler, ApiError } from '@/lib/api-errors'
+import { getUserModelConfig } from '@/lib/config-service'
+import { maybeSubmitLLMTask } from '@/lib/llm-observe/route-task'
+import { TASK_TYPE } from '@/lib/task/types'
+
+export const POST = apiHandler(async (request: NextRequest) => {
+  const authResult = await requireUserAuth()
+  if (isErrorResponse(authResult)) return authResult
+  const { session } = authResult
+
+  const body = (await request.json().catch(() => ({}))) as Record<string, unknown>
+  const prompt = typeof body.prompt === 'string' ? body.prompt.trim() : ''
+  if (!prompt) {
+    throw new ApiError('INVALID_PARAMS')
+  }
+
+  const userConfig = await getUserModelConfig(session.user.id)
+  if (!userConfig.analysisModel) {
+    throw new ApiError('MISSING_CONFIG')
+  }
+
+  const dedupeDigest = createHash('sha1')
+    .update(`${session.user.id}:home-story-expand:${prompt}`)
+    .digest('hex')
+    .slice(0, 16)
+
+  const asyncTaskResponse = await maybeSubmitLLMTask({
+    request,
+    userId: session.user.id,
+    projectId: 'home-ai-write',
+    type: TASK_TYPE.AI_STORY_EXPAND,
+    targetType: 'HomeAiStoryExpand',
+    targetId: session.user.id,
+    routePath: '/api/user/ai-story-expand',
+    body: {
+      prompt,
+      analysisModel: userConfig.analysisModel,
+    },
+    dedupeKey: `home_ai_story_expand:${dedupeDigest}`,
+    priority: 1,
+  })
+  if (asyncTaskResponse) return asyncTaskResponse
+
+  throw new ApiError('INVALID_PARAMS')
+})

src/components/home/AiWriteModal.tsx

@@ -0,0 +1,126 @@
+'use client'
+
+/**
+ * AI 帮我写 — 首页轻量模态框
+ *
+ * 用户输入创意/关键词/大纲，直接生成结果并回填首页主输入框
+ */
+
+import { useState, useCallback } from 'react'
+import { AppIcon } from '@/components/ui/icons'
+
+interface AiWriteModalProps {
+  open: boolean
+  loading: boolean
+  onClose: () => void
+  onStart: (prompt: string) => void
+  t: (key: string) => string
+}
+
+export default function AiWriteModal({
+  open,
+  loading,
+  onClose,
+  onStart,
+  t,
+}: AiWriteModalProps) {
+  const [promptText, setPromptText] = useState('')
+
+  const handleClose = useCallback(() => {
+    if (loading) return
+    setPromptText('')
+    onClose()
+  }, [loading, onClose])
+
+  const handleStart = useCallback(() => {
+    if (!promptText.trim() || loading) return
+    onStart(promptText.trim())
+  }, [promptText, loading, onStart])
+
+  if (!open) return null
+
+  return (
+    <div
+      className="fixed inset-0 glass-overlay flex items-center justify-center z-50 backdrop-blur-sm"
+      onClick={handleClose}
+    >
+      <div
+        className="w-full max-w-lg mx-4 relative"
+        onClick={(e) => e.stopPropagation()}
+      >
+        {/* 模态框容器 */}
+        <div className="glass-surface-modal rounded-2xl p-6 space-y-5">
+          {/* 头部 */}
+          <div className="flex items-center justify-between">
+            <div className="flex items-center gap-3">
+              <div
+                className="w-10 h-10 rounded-xl flex items-center justify-center flex-shrink-0"
+                style={{ background: 'linear-gradient(135deg, rgba(59,130,246,0.15), rgba(139,92,246,0.15))' }}
+              >
+                <AppIcon name="sparkles" className="w-5 h-5 text-[#7c3aed]" />
+              </div>
+              <div>
+                <h3 className="text-lg font-bold text-[var(--glass-text-primary)]">
+                  {t('modalTitle')}
+                </h3>
+                <p className="text-xs text-[var(--glass-text-tertiary)]">
+                  {t('modalSubtitle')}
+                </p>
+              </div>
+            </div>
+            <button
+              onClick={handleClose}
+              className="glass-icon-btn-sm"
+              disabled={loading}
+            >
+              <AppIcon name="close" className="w-4 h-4" />
+            </button>
+          </div>
+
+          {/* 输入区域 */}
+          <div>
+            <label className="text-sm font-medium text-[var(--glass-text-secondary)] mb-2 block">
+              {t('inputLabel')}
+            </label>
+            <textarea
+              value={promptText}
+              onChange={(e) => setPromptText(e.target.value)}
+              placeholder={t('placeholder')}
+              className="glass-textarea-base custom-scrollbar h-36 px-4 py-3 text-sm resize-none placeholder:text-[var(--glass-text-tertiary)]"
+              disabled={loading}
+              autoFocus
+            />
+          </div>
+
+          {/* 提示文案 */}
+          <div
+            className="px-3 py-2 rounded-lg text-xs text-[var(--glass-text-tertiary)] leading-relaxed"
+            style={{ background: 'linear-gradient(135deg, rgba(59,130,246,0.06), rgba(139,92,246,0.06))' }}
+          >
+            {t('hint')}
+          </div>
+
+          {/* 按钮区域 */}
+          <div className="flex items-center gap-3">
+            <button
+              onClick={handleClose}
+              disabled={loading}
+              className="flex-1 py-2.5 text-sm text-[var(--glass-text-tertiary)] hover:text-[var(--glass-text-secondary)] transition-colors rounded-xl"
+            >
+              {t('cancel')}
+            </button>
+            <button
+              onClick={handleStart}
+              disabled={!promptText.trim() || loading}
+              className="flex-1 py-3 rounded-xl text-white font-semibold text-sm flex items-center justify-center gap-2 transition-all hover:opacity-90 active:scale-[0.98] disabled:opacity-50"
+              style={{ background: 'linear-gradient(135deg, #3b82f6, #7c3aed)' }}
+            >
+              <AppIcon name="sparkles" className="w-4 h-4" />
+              <span>{loading ? '...' : t('startAiWrite')}</span>
+            </button>
+          </div>
+        </div>
+      </div>
+    </div>
+  )
+}

src/lib/api-errors.ts

@@ -31,6 +31,7 @@ const GENERATION_OPERATION_PATTERNS = [
   /\/story-to-script(?:-|\/|$)/,
   /\/script-to-storyboard(?:-|\/|$)/,
   /\/screenplay-conversion(?:\/|$)/,
+  /\/ai-story-expand(?:\/|$)/,
   /\/voice-(?:analyze|design|generate)(?:\/|$)/,
   /\/ai-(?:create|modify)-/,
   /\/modify-(?:asset|storyboard)-image(?:\/|$)/,

src/lib/billing/task-policy.ts

@@ -36,6 +36,7 @@ const BILLABLE_TASK_TYPES = new Set<TaskType>([
   TASK_TYPE.SCREENPLAY_CONVERT,
   TASK_TYPE.VOICE_ANALYZE,
   TASK_TYPE.ANALYZE_GLOBAL,
+  TASK_TYPE.AI_STORY_EXPAND,
   TASK_TYPE.AI_MODIFY_APPEARANCE,
   TASK_TYPE.AI_MODIFY_LOCATION,
   TASK_TYPE.AI_MODIFY_SHOT_PROMPT,
@@ -277,6 +278,7 @@ export function buildDefaultTaskBillingInfo(taskType: TaskType, payload: AnyPayl
     case TASK_TYPE.SCREENPLAY_CONVERT:
     case TASK_TYPE.VOICE_ANALYZE:
     case TASK_TYPE.ANALYZE_GLOBAL:
+    case TASK_TYPE.AI_STORY_EXPAND:
     case TASK_TYPE.AI_MODIFY_APPEARANCE:
     case TASK_TYPE.AI_MODIFY_LOCATION:
     case TASK_TYPE.AI_MODIFY_SHOT_PROMPT:

src/lib/home/ai-story-expand.ts

@@ -0,0 +1,41 @@
+import { resolveTaskResponse } from '@/lib/task/client'
+
+interface ApiFetchLike {
+  (input: string, init?: RequestInit): Promise<Response>
+}
+
+interface ExpandHomeStoryPayload {
+  expandedText?: string
+}
+
+export interface ExpandHomeStoryParams {
+  apiFetch: ApiFetchLike
+  prompt: string
+}
+
+export interface ExpandHomeStoryResult {
+  expandedText: string
+}
+
+export async function expandHomeStory({
+  apiFetch,
+  prompt,
+}: ExpandHomeStoryParams): Promise<ExpandHomeStoryResult> {
+  const response = await apiFetch('/api/user/ai-story-expand', {
+    method: 'POST',
+    headers: { 'Content-Type': 'application/json' },
+    body: JSON.stringify({
+      prompt,
+    }),
+  })
+
+  const result = await resolveTaskResponse<ExpandHomeStoryPayload>(response)
+  const expandedText = typeof result.expandedText === 'string' ? result.expandedText.trim() : ''
+  if (!expandedText) {
+    throw new Error('AI story expand response missing expandedText')
+  }
+
+  return {
+    expandedText,
+  }
+}

src/lib/home/create-project-launch.ts

@@ -22,7 +22,7 @@ export interface HomeWorkspaceLaunchTarget {
   pathname: string
   query: {
     episode: string
-    autoRun: 'storyToScript'
+    autoRun?: 'storyToScript'
   }
 }
 

src/lib/llm-observe/task-policy.ts

@@ -48,6 +48,7 @@ const POLICY_BY_TASK_TYPE: Partial<Record<TaskType, LLMTaskPolicy>> = {
   [TASK_TYPE.SCREENPLAY_CONVERT]: LONG_FLOW_HIGH_POLICY,
   [TASK_TYPE.VOICE_ANALYZE]: LLM_STANDARD_POLICY,
   [TASK_TYPE.ANALYZE_GLOBAL]: LLM_STANDARD_POLICY,
+  [TASK_TYPE.AI_STORY_EXPAND]: LLM_STANDARD_POLICY,
   [TASK_TYPE.AI_MODIFY_APPEARANCE]: LLM_STANDARD_POLICY,
   [TASK_TYPE.AI_MODIFY_LOCATION]: LLM_STANDARD_POLICY,
   [TASK_TYPE.AI_MODIFY_SHOT_PROMPT]: LLM_STANDARD_POLICY,

src/lib/prompt-i18n/catalog.ts

@@ -81,6 +81,10 @@ export const PROMPT_CATALOG: Record<PromptId, PromptCatalogEntry> = {
       'clip_content',
     ],
   },
+  [PROMPT_IDS.NP_AI_STORY_EXPAND]: {
+    pathStem: 'novel-promotion/ai_story_expand',
+    variableKeys: ['input'],
+  },
   [PROMPT_IDS.NP_CHARACTER_CREATE]: {
     pathStem: 'novel-promotion/character_create',
     variableKeys: ['user_input'],

src/lib/prompt-i18n/prompt-ids.ts

@@ -11,6 +11,7 @@ export const PROMPT_IDS = {
   NP_AGENT_STORYBOARD_DETAIL: 'np_agent_storyboard_detail',
   NP_AGENT_STORYBOARD_INSERT: 'np_agent_storyboard_insert',
   NP_AGENT_STORYBOARD_PLAN: 'np_agent_storyboard_plan',
+  NP_AI_STORY_EXPAND: 'np_ai_story_expand',
   NP_CHARACTER_CREATE: 'np_character_create',
   NP_CHARACTER_DESCRIPTION_UPDATE: 'np_character_description_update',
   NP_CHARACTER_MODIFY: 'np_character_modify',

src/lib/task/intent.ts

@@ -44,6 +44,7 @@ const TASK_INTENT_BY_TYPE: Record<TaskType, TaskIntent> = {
   [TASK_TYPE.SCREENPLAY_CONVERT]: 'convert',
   [TASK_TYPE.VOICE_ANALYZE]: 'analyze',
   [TASK_TYPE.ANALYZE_GLOBAL]: 'analyze',
+  [TASK_TYPE.AI_STORY_EXPAND]: 'generate',
   [TASK_TYPE.AI_MODIFY_APPEARANCE]: 'modify',
   [TASK_TYPE.AI_MODIFY_LOCATION]: 'modify',
   [TASK_TYPE.AI_MODIFY_SHOT_PROMPT]: 'modify',

src/lib/task/progress-message.ts

@@ -23,6 +23,7 @@ const TASK_TYPE_LABELS: Record<string, string> = {
   [TASK_TYPE.SCREENPLAY_CONVERT]: 'progress.taskType.screenplayConvert',
   [TASK_TYPE.VOICE_ANALYZE]: 'progress.taskType.voiceAnalyze',
   [TASK_TYPE.ANALYZE_GLOBAL]: 'progress.taskType.analyzeGlobal',
+  [TASK_TYPE.AI_STORY_EXPAND]: 'progress.taskType.aiStoryExpand',
   [TASK_TYPE.AI_MODIFY_APPEARANCE]: 'progress.taskType.aiModifyAppearance',
   [TASK_TYPE.AI_MODIFY_LOCATION]: 'progress.taskType.aiModifyLocation',
   [TASK_TYPE.AI_MODIFY_SHOT_PROMPT]: 'progress.taskType.aiModifyShotPrompt',
@@ -63,6 +64,8 @@ const STAGE_LABELS: Record<string, string> = {
   script_to_storyboard_step: 'progress.stage.scriptToStoryboardStep',
   script_to_storyboard_persist: 'progress.stage.scriptToStoryboardPersist',
   script_to_storyboard_persist_done: 'progress.stage.scriptToStoryboardPersistDone',
+  ai_story_expand_prepare: 'progress.stage.aiStoryExpandPrepare',
+  ai_story_expand_done: 'progress.stage.aiStoryExpandDone',
   insert_panel_generate_text: 'progress.stage.insertPanelGenerateText',
   insert_panel_persist: 'progress.stage.insertPanelPersist',
   polling_external: 'progress.stage.pollingExternal',

src/lib/task/types.ts

@@ -60,6 +60,7 @@ export const TASK_TYPE = {
   SCREENPLAY_CONVERT: 'screenplay_convert',
   VOICE_ANALYZE: 'voice_analyze',
   ANALYZE_GLOBAL: 'analyze_global',
+  AI_STORY_EXPAND: 'ai_story_expand',
   AI_MODIFY_APPEARANCE: 'ai_modify_appearance',
   AI_MODIFY_LOCATION: 'ai_modify_location',
   AI_MODIFY_SHOT_PROMPT: 'ai_modify_shot_prompt',

src/lib/workers/handlers/ai-story-expand.ts

@@ -0,0 +1,79 @@
+import type { Job } from 'bullmq'
+import { executeAiTextStep } from '@/lib/ai-runtime'
+import { withInternalLLMStreamCallbacks } from '@/lib/llm-observe/internal-stream-context'
+import { buildPrompt, PROMPT_IDS } from '@/lib/prompt-i18n'
+import type { TaskJobData } from '@/lib/task/types'
+import { reportTaskProgress } from '@/lib/workers/shared'
+import { assertTaskActive } from '@/lib/workers/utils'
+import { createWorkerLLMStreamCallbacks, createWorkerLLMStreamContext } from './llm-stream'
+
+function readText(value: unknown): string {
+  return typeof value === 'string' ? value : ''
+}
+
+export async function handleAiStoryExpandTask(job: Job<TaskJobData>) {
+  const payload = (job.data.payload || {}) as Record<string, unknown>
+  const promptInput = readText(payload.prompt).trim()
+  const analysisModel = readText(payload.analysisModel).trim()
+
+  if (!promptInput) {
+    throw new Error('prompt is required')
+  }
+  if (!analysisModel) {
+    throw new Error('analysisModel is required')
+  }
+
+  const prompt = buildPrompt({
+    promptId: PROMPT_IDS.NP_AI_STORY_EXPAND,
+    locale: job.data.locale,
+    variables: {
+      input: promptInput,
+    },
+  })
+
+  await reportTaskProgress(job, 25, {
+    stage: 'ai_story_expand_prepare',
+    stageLabel: '准备故事扩写参数',
+    displayMode: 'loading',
+  })
+  await assertTaskActive(job, 'ai_story_expand_prepare')
+
+  const streamContext = createWorkerLLMStreamContext(job, 'ai_story_expand')
+  const streamCallbacks = createWorkerLLMStreamCallbacks(job, streamContext)
+
+  const completion = await withInternalLLMStreamCallbacks(
+    streamCallbacks,
+    async () =>
+      await executeAiTextStep({
+        userId: job.data.userId,
+        model: analysisModel,
+        messages: [{ role: 'user', content: prompt }],
+        temperature: 0.7,
+        projectId: job.data.projectId || 'home-ai-write',
+        action: 'ai_story_expand',
+        meta: {
+          stepId: 'ai_story_expand',
+          stepTitle: '故事扩写',
+          stepIndex: 1,
+          stepTotal: 1,
+        },
+      }),
+  )
+  await streamCallbacks.flush()
+  await assertTaskActive(job, 'ai_story_expand_persist')
+
+  const expandedText = completion.text.trim()
+  if (!expandedText) {
+    throw new Error('AI story expand response is empty')
+  }
+
+  await reportTaskProgress(job, 96, {
+    stage: 'ai_story_expand_done',
+    stageLabel: '故事扩写已完成',
+    displayMode: 'loading',
+  })
+
+  return {
+    expandedText,
+  }
+}

src/lib/workers/text.worker.ts

@@ -25,6 +25,7 @@ import { handleStoryToScriptTask } from './handlers/story-to-script'
 import { handleScriptToStoryboardTask } from './handlers/script-to-storyboard'
 import { handleVoiceAnalyzeTask } from './handlers/voice-analyze'
 import { handleAssetHubAIDesignTask } from './handlers/asset-hub-ai-design'
+import { handleAiStoryExpandTask } from './handlers/ai-story-expand'
 import { handleClipsBuildTask } from './handlers/clips-build'
 import { handleAnalyzeNovelTask } from './handlers/analyze-novel'
 import { handleScreenplayConvertTask } from './handlers/screenplay-convert'
@@ -663,6 +664,8 @@ async function processTextTask(job: Job<TaskJobData>) {
       return await handleVoiceAnalyzeTask(job)
     case TASK_TYPE.ANALYZE_NOVEL:
       return await handleAnalyzeNovelTask(job)
+    case TASK_TYPE.AI_STORY_EXPAND:
+      return await handleAiStoryExpandTask(job)
     case TASK_TYPE.CLIPS_BUILD:
       return await handleClipsBuildTask(job)
     case TASK_TYPE.SCREENPLAY_CONVERT:

tests/contracts/route-catalog.ts

@@ -163,6 +163,7 @@ const ROUTE_FILES = [
   'src/app/api/user/balance/route.ts',
   'src/app/api/user/costs/details/route.ts',
   'src/app/api/user/costs/route.ts',
+  'src/app/api/user/ai-story-expand/route.ts',
   'src/app/api/user/models/route.ts',
   'src/app/api/user/transactions/route.ts',
 ] as const

tests/contracts/task-type-catalog.ts

@@ -31,6 +31,7 @@ const TASK_TYPE_OWNER_MAP = {
   [TASK_TYPE.SCREENPLAY_CONVERT]: 'tests/unit/worker/screenplay-convert.test.ts',
   [TASK_TYPE.VOICE_ANALYZE]: 'tests/unit/worker/voice-analyze.test.ts',
   [TASK_TYPE.ANALYZE_GLOBAL]: 'tests/unit/worker/analyze-global.test.ts',
+  [TASK_TYPE.AI_STORY_EXPAND]: 'tests/unit/worker/ai-story-expand.test.ts',
   [TASK_TYPE.AI_MODIFY_APPEARANCE]: 'tests/unit/worker/shot-ai-prompt-appearance.test.ts',
   [TASK_TYPE.AI_MODIFY_LOCATION]: 'tests/unit/worker/shot-ai-prompt-location.test.ts',
   [TASK_TYPE.AI_MODIFY_SHOT_PROMPT]: 'tests/unit/worker/shot-ai-prompt-shot.test.ts',

tests/integration/api/contract/llm-observe-routes.test.ts

@@ -171,6 +171,13 @@ const ROUTE_CASES: ReadonlyArray<LLMRouteCase> = [
     expectedTargetType: 'NovelPromotionLocationDesign',
     expectedProjectId: 'project-1',
   },
+  {
+    routeFile: 'src/app/api/user/ai-story-expand/route.ts',
+    body: { prompt: '宫廷复仇女主回京' },
+    expectedTaskType: TASK_TYPE.AI_STORY_EXPAND,
+    expectedTargetType: 'HomeAiStoryExpand',
+    expectedProjectId: 'home-ai-write',
+  },
   {
     routeFile: 'src/app/api/novel-promotion/[projectId]/ai-modify-appearance/route.ts',
     body: {
@@ -336,7 +343,7 @@ describe('api contract - llm observe routes (behavior)', () => {
   })
 
   it('keeps expected coverage size', () => {
-    expect(ROUTE_CASES.length).toBe(22)
+    expect(ROUTE_CASES.length).toBe(23)
   })
 
   for (const routeCase of ROUTE_CASES) {

tests/unit/home/ai-story-expand.test.ts

@@ -0,0 +1,50 @@
+import { describe, expect, it, vi } from 'vitest'
+import { expandHomeStory } from '@/lib/home/ai-story-expand'
+
+vi.mock('@/lib/task/client', () => ({
+  resolveTaskResponse: vi.fn(),
+}))
+
+import { resolveTaskResponse } from '@/lib/task/client'
+
+function buildJsonResponse(body: unknown, status = 200): Response {
+  return new Response(JSON.stringify(body), {
+    status,
+    headers: { 'Content-Type': 'application/json' },
+  })
+}
+
+describe('expandHomeStory', () => {
+  it('posts the prompt to the user ai-story-expand route and returns expanded text', async () => {
+    const apiFetch = vi.fn(async () => buildJsonResponse({ async: true, taskId: 'task-1' }))
+    vi.mocked(resolveTaskResponse).mockResolvedValue({
+      expandedText: '扩写后的故事正文',
+    })
+
+    const result = await expandHomeStory({
+      apiFetch,
+      prompt: '宫廷复仇女主回京',
+    })
+
+    expect(apiFetch).toHaveBeenCalledWith('/api/user/ai-story-expand', {
+      method: 'POST',
+      headers: { 'Content-Type': 'application/json' },
+      body: JSON.stringify({
+        prompt: '宫廷复仇女主回京',
+      }),
+    })
+    expect(result).toEqual({
+      expandedText: '扩写后的故事正文',
+    })
+  })
+
+  it('fails explicitly when the route does not return expandedText', async () => {
+    const apiFetch = vi.fn(async () => buildJsonResponse({ async: true, taskId: 'task-1' }))
+    vi.mocked(resolveTaskResponse).mockResolvedValue({})
+
+    await expect(expandHomeStory({
+      apiFetch,
+      prompt: '宫廷复仇女主回京',
+    })).rejects.toThrow('AI story expand response missing expandedText')
+  })
+})

tests/unit/home/quick-start-textarea.test.ts

@@ -59,6 +59,10 @@ vi.mock('@/lib/home/create-project-launch', () => ({
   createHomeProjectLaunch: vi.fn(),
 }))
 
+vi.mock('@/lib/home/ai-story-expand', () => ({
+  expandHomeStory: vi.fn(),
+}))
+
 describe('resolveTextareaTargetHeight', () => {
   it('keeps the home quick-start input at least three rows tall', () => {
     expect(resolveTextareaTargetHeight({

tests/unit/task/intent.test.ts

@@ -7,6 +7,7 @@ describe('resolveTaskIntent', () => {
     expect(resolveTaskIntent(TASK_TYPE.IMAGE_CHARACTER)).toBe('generate')
     expect(resolveTaskIntent(TASK_TYPE.IMAGE_LOCATION)).toBe('generate')
     expect(resolveTaskIntent(TASK_TYPE.VIDEO_PANEL)).toBe('generate')
+    expect(resolveTaskIntent(TASK_TYPE.AI_STORY_EXPAND)).toBe('generate')
   })
 
   it('maps regenerate and modify task types', () => {

tests/unit/worker/ai-story-expand.test.ts

@@ -0,0 +1,84 @@
+import type { Job } from 'bullmq'
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+import { TASK_TYPE, type TaskJobData } from '@/lib/task/types'
+
+const aiRuntimeMock = vi.hoisted(() => ({
+  executeAiTextStep: vi.fn(async () => ({
+    text: '扩写后的完整故事内容',
+    reasoning: '',
+  })),
+}))
+
+const workerMock = vi.hoisted(() => ({
+  reportTaskProgress: vi.fn(async () => undefined),
+  assertTaskActive: vi.fn(async () => undefined),
+}))
+
+vi.mock('@/lib/ai-runtime', () => aiRuntimeMock)
+vi.mock('@/lib/llm-observe/internal-stream-context', () => ({
+  withInternalLLMStreamCallbacks: vi.fn(async (_callbacks: unknown, fn: () => Promise<unknown>) => await fn()),
+}))
+vi.mock('@/lib/prompt-i18n', () => ({
+  PROMPT_IDS: { NP_AI_STORY_EXPAND: 'np_ai_story_expand' },
+  buildPrompt: vi.fn(() => 'story-expand-prompt'),
+}))
+vi.mock('@/lib/workers/shared', () => ({ reportTaskProgress: workerMock.reportTaskProgress }))
+vi.mock('@/lib/workers/utils', () => ({ assertTaskActive: workerMock.assertTaskActive }))
+vi.mock('@/lib/workers/handlers/llm-stream', () => ({
+  createWorkerLLMStreamContext: vi.fn(() => ({ streamRunId: 'run-1', nextSeqByStepLane: {} })),
+  createWorkerLLMStreamCallbacks: vi.fn(() => ({
+    onStage: vi.fn(),
+    onChunk: vi.fn(),
+    onComplete: vi.fn(),
+    onError: vi.fn(),
+    flush: vi.fn(async () => undefined),
+  })),
+}))
+
+import { handleAiStoryExpandTask } from '@/lib/workers/handlers/ai-story-expand'
+
+function buildJob(payload: Record<string, unknown>): Job<TaskJobData> {
+  return {
+    data: {
+      taskId: 'task-ai-story-expand-1',
+      type: TASK_TYPE.AI_STORY_EXPAND,
+      locale: 'zh',
+      projectId: 'home-ai-write',
+      targetType: 'HomeAiStoryExpand',
+      targetId: 'user-1',
+      payload,
+      userId: 'user-1',
+    },
+  } as unknown as Job<TaskJobData>
+}
+
+describe('worker ai-story-expand behavior', () => {
+  beforeEach(() => {
+    vi.clearAllMocks()
+  })
+
+  it('missing prompt -> explicit error', async () => {
+    const job = buildJob({ prompt: '   ', analysisModel: 'provider::analysis-model' })
+    await expect(handleAiStoryExpandTask(job)).rejects.toThrow('prompt is required')
+  })
+
+  it('missing analysis model -> explicit error', async () => {
+    const job = buildJob({ prompt: '宫廷复仇女主回京' })
+    await expect(handleAiStoryExpandTask(job)).rejects.toThrow('analysisModel is required')
+  })
+
+  it('success path -> returns expanded text without touching episode persistence', async () => {
+    const job = buildJob({ prompt: '宫廷复仇女主回京', analysisModel: 'provider::analysis-model' })
+    const result = await handleAiStoryExpandTask(job)
+
+    expect(result).toEqual({
+      expandedText: '扩写后的完整故事内容',
+    })
+    expect(aiRuntimeMock.executeAiTextStep).toHaveBeenCalledWith(expect.objectContaining({
+      userId: 'user-1',
+      model: 'provider::analysis-model',
+      projectId: 'home-ai-write',
+      action: 'ai_story_expand',
+    }))
+  })
+})