feat: 任务发布到企微客户群（企业群发）与资源库单条新增

- 任务中心新增「发布到企微群」：同步客户群清单（wecom_group_chats）、按群主分组创建企业群发任务（add_msg_template）、发送记录落库 wecom_group_pushes，迁移 mysql/0009；lib/wecom-client.ts 补 listCustomerGroupChats/createGroupMsgTemplate - KOC 资源库支持单条新增：app/api/resources-insert + lib/resource-write，拆出资源写入公共逻辑供导入复用；0008 补合作方外部联系人字段 - 环境变量示例补企微凭证与 SEED_DEMO_DATA；next.config 增加 allowedDevOrigins；CLAUDE.md 补充项目说明 - .gitignore 排除 .codegraph/ 与 .ipynb_checkpoints/
Merge pull request 'feat: 接入企业微信通知（临期催办 + 群机器人汇总）' (#3 ) from feat/account-tags into main
2026-08-20 15:16:18 +08:00 · 2026-08-18 09:12:59 +00:00 · 2026-08-18 17:05:16 +08:00 · 2026-08-16 22:22:50 +08:00 · 2026-08-15 03:57:50 +08:00 · 2026-08-15 03:53:09 +08:00
85 changed files with 10939 additions and 844 deletions
--- a/.dev.vars.example
+++ b/.dev.vars.example
@@ -11,3 +11,14 @@ AI_TOOL_CENTER_MCP_KEY=replace-with-mcp-key
 # Feishu custom app credentials. Keep the secret out of source control.
 FEISHU_APP_ID=cli_xxxxxxxxxxxxxxxxx
 FEISHU_APP_SECRET=replace-with-feishu-app-secret
+
+# 企业微信 · 全部选填；不填则临期催办与汇总都跳过，不影响其他功能
+WECOM_CORP_ID=
+WECOM_AGENT_ID=
+WECOM_SECRET=
+WECOM_ROBOT_WEBHOOK=
+WECOM_NOTIFY_DUE_DAYS=3
+WECOM_NOTIFY_ENABLED=true
+
+# 本地测试用：让 /api/bootstrap 自动 seed 测试 partner + distribution；上线设为 false
+SEED_DEMO_DATA=false
--- a/.env.self-hosted.example
+++ b/.env.self-hosted.example
@@ -24,6 +24,15 @@ FEISHU_APP_SECRET=
 AI_TOOL_CENTER_MCP_URL=
 AI_TOOL_CENTER_MCP_KEY=

+# 企业微信通知（临期催办 + 管理员汇总）。群机器人只需 webhook；KOC 侧催办还需 corp/agent/secret，
+# 并在后台 partners 编辑里把 wecom_external_user_id 填好。
+WECOM_ROBOT_WEBHOOK=
+WECOM_CORP_ID=
+WECOM_AGENT_ID=
+WECOM_SECRET=
+WECOM_NOTIFY_DUE_DAYS=3
+WECOM_NOTIFY_ENABLED=true
+
 # 每天北京时间 09:00 自动执行采集计划。
 ENABLE_SCHEDULER=true
 SEED_DEMO_DATA=false
--- a/.gitignore
+++ b/.gitignore
@@ -22,6 +22,8 @@
 # misc
 .DS_Store
 *.pem
+.codegraph/
+.ipynb_checkpoints/

 # debug
 npm-debug.log*
@@ -39,6 +41,7 @@ yarn-error.log*

 # typescript
 next-env.d.ts
+*.tsbuildinfo
 /dist/
 /.wrangler/
 /outputs/
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -0,0 +1,132 @@
+# CLAUDE.md
+
+This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
+
+## What This Project Does
+
+KOC LOOP is a content distribution & data collection platform for KOC (Key Opinion Consumer) operations teams. It runs on **vinext** (Vite-based Next.js 16 on Cloudflare Workers + Pages) with Cloudflare D1 (SQLite) and R2 storage.
+
+The platform manages: task creation from Feishu (飞书) spreadsheets, content distribution to KOC partners, publishing to Xiaohongshu (小红书), and automated data collection (likes/comments/collects) via MCP-based metrics scraping.
+
+## Two Applications
+
+### 1. Admin App (`app/`) — Main dashboard
+- Server-rendered admin UI at root (`/`), protected by ChatGPT Sign-In + admin email check
+- Heavy client-side SPA in `app/admin-app.tsx` (a single ~59KB component with all dashboard state/views)
+- API routes under `app/api/` for CRUD, Feishu import, data collection, image upload
+
+### 2. KOC Portal (`koc-portal/`) — External task portal
+- Independent Next.js app (separate `package.json`) for external KOC collaborators
+- KOCs can claim tasks, view assigned content, and submit publish URLs/screenshots
+- Does NOT connect directly to the database — uses the admin app's API
+
+## Tech Stack
+
+- **Framework**: Next.js 16 + React 19 + TypeScript
+- **Build/Runtime**: vinext 0.0.50 (Vite 8 plugin → Cloudflare Workers/Pages)
+- **Database**: Cloudflare D1 (SQLite via Drizzle ORM 0.45)
+- **Storage**: Cloudflare R2 for uploads/screenshots
+- **Styling**: Tailwind CSS 4
+- **Cron**: Cloudflare Workers cron (daily 02:00 UTC / 10:00 CST)
+
+## Database Schema (7 tables in `db/schema.ts`)
+
+| Table | Purpose |
+|-------|---------|
+| `partners` | KOC partners/groups (name, owner, stats) |
+| `tasks` | Campaign tasks (brand, quantity, due date, Feishu source) |
+| `contents` | Content items (title, body, images, linked to task) |
+| `accounts` | Xiaohongshu accounts scraped from publish links |
+| `claims` | Partner claims on task content |
+| `delegation_bundles` | Delegation bundles with share tokens |
+| `distributions` | Content-to-partner assignments (publish URL, metrics, collection status) |
+| `collection_runs` | Scheduled metrics collection run log |
+
+Schema is maintained both via Drizzle (`db/schema.ts`) and imperative migrations in `lib/mvp-db.ts:ensureSchema()`. The imperative path is the source of truth for production — Drizzle migrations are optional.
+
+## Key Libraries
+
+- `lib/mvp-db.ts` — Raw D1 helpers, schema bootstrapping, seed data, `getDashboardData()`
+- `lib/collection-service.ts` — Scheduled metrics collection orchestration
+- `lib/mcp-collection-client.ts` — MCP-based client for scraping Xiaohongshu public metrics
+- `lib/feishu-client.ts` — Feishu API client (spreadsheet/wiki/doc reading)
+- `lib/account-enrichment-service.ts` — Backfill account profiles from Xiaohongshu
+- `lib/admin-auth.ts` — Admin email/token authentication
+- `lib/partner-utils.ts` / `lib/publish-url.ts` — URL extraction helpers
+- `lib/date-utils.ts` — Date formatting (Shanghai timezone)
+
+## Worker (`worker/index.ts`)
+
+The Cloudflare Worker entry point handles:
+1. `fetch` — Image optimization proxy at `/_vinext/image`, delegates everything else to vinext app router
+2. `scheduled` — Daily cron: ensures DB schema, runs scheduled collections, backfills account profiles
+
+> Note: 在私有化部署（Node + MySQL）里实际调度走 `lib/scheduler.ts` 的 `node-cron`，每天 09:00 Asia/Shanghai 跑同一套 `runDailyJob`（采集 + 账号资料补全 + 企业微信催办）。Cloudflare Worker 入口仅用于原线上版本。
+
+## API Routes (`app/api/`)
+
+- `action/route.ts` — Central admin action endpoint: create task from Feishu, dashboard data, collect metrics, batch operations, account backfill, 企业微信测试 / 状态查询 / 绑定外部联系人 / 客户群同步与企业群发（半自动，群主确认后送达）
+- `partner/route.ts` — Partner-facing API (claim, delegation, distribution)
+- `bootstrap/route.ts` — Seed database with demo data
+- `upload/route.ts` — Generic file upload to R2
+- `partner-upload/route.ts` — Partner screenshot upload (CORS-enabled)
+- `partner-image/route.ts` — Partner image serving (CORS-enabled)
+- `content-image-upload/route.ts` — Content image upload
+- `creator-screenshot/route.ts` — Creator screenshot upload
+- `resources-import/route.ts` — 批量导入 KOC 账号资源（Excel/CSV）
+- `resources-insert/route.ts` — 单条新增 KOC 账号资源（表单）
+
+## Environment Variables (`.dev.vars`)
+
+```
+KOC_PORTAL_URL         — URL of the koc-portal app
+ADMIN_ALLOWED_EMAIL    — ChatGPT email allowed for admin access
+ADMIN_INTERNAL_TOKEN   — Shared secret for API-to-API auth
+AI_TOOL_CENTER_MCP_URL — MCP endpoint for XHS data collection
+AI_TOOL_CENTER_MCP_KEY — MCP API key
+FEISHU_APP_ID          — Feishu custom app credentials
+FEISHU_APP_SECRET      — Feishu app secret
+WECOM_CORP_ID          — 企业微信企业 ID（自建应用消息推送用，可选）
+WECOM_AGENT_ID         — 企业微信自建应用 AgentId（可选）
+WECOM_SECRET           — 企业微信自建应用 Secret（可选）
+WECOM_ROBOT_WEBHOOK    — 企业微信群机器人 webhook（当日催办汇总，可选）
+WECOM_NOTIFY_DUE_DAYS  — 临期阈值，默认 3，取值 1–30（可选）
+WECOM_NOTIFY_ENABLED   — 企业微信通知总开关，默认 true（可选）
+```
+
+## Commands
+
+```bash
+# Admin app (root of repo)
+npm run dev              # Start local dev server
+npm run build            # Build for production
+npm test                 # Build + run integration tests
+npm run lint             # ESLint check
+npm run db:generate      # Generate Drizzle SQL migration
+
+# KOC Portal (koc-portal/)
+cd koc-portal && npm run dev -- --port 3000  # Start on port 3000
+cd koc-portal && npm run build
+cd koc-portal && npm test
+
+# Tests use `node --test` (Node built-in test runner), live in tests/
+node --test tests/date-utils.test.mjs         # Run a single test
+```
+
+## Build System
+
+- `vinext` replaces the Next.js build pipeline with Vite
+- `@cloudflare/vite-plugin` provides D1/R2/cron local bindings
+- Vite config at root resolves bindings from `.openai/hosting.json`
+- Custom `build/sites-vite-plugin.ts` copies `.openai/` and `drizzle/` into `dist/` for deployment
+- The `koc-portal/` subdirectory has its own identical build setup (separate `vite.config.ts`, `build/sites-vite-plugin.ts`)
+
+<!-- BEGIN:nextjs-agent-rules -->
+
+# This is NOT the Next.js you know
+
+This version has breaking changes — APIs, conventions, and file structure may all differ from your training data. Read the relevant guide in `node_modules/next/dist/docs/` (resolved from this file's directory; in monorepos the `next` package may not be visible from the repo root) before writing any code. Heed deprecation notices.
+
+This block is written and re-added by `next dev` — verify at `node_modules/next/dist/server/lib/generate-agent-files.js`. Removing it from a diff only re-creates the uncommitted change; committing it with your work keeps the tree clean.
+
+<!-- END:nextjs-agent-rules -->
--- a/README.md
+++ b/README.md
@@ -1,6 +1,6 @@
 # KOC LOOP

-KOC 内容分发与数据回收闭环。当前私有化分支运行于标准 Next.js Node.js、MySQL 8、Nginx 和本地持久化文件存储。
+KOC 内容分发与数据回收闭环。`main` 分支运行于标准 Next.js Node.js、MySQL 8、Nginx 和本地持久化文件存储。

 ## Prerequisites

@@ -45,13 +45,21 @@ npm run build

 ## KOC 资源导入

-超级管理员和管理员可在“KOC资源”页面下载标准 Excel 模板，批量导入已有资源。模板只需填写小红书账号主页，合作来源可选填；上传后系统自动解析账号名称、小红书号、IP属地和粉丝数。
+超级管理员和管理员可在“KOC资源”页面下载标准 Excel 模板，批量导入已有资源。模板只需填写账号主页；账号名称、账号 ID、IP 属地、粉丝数、性别、简介、标签和合作来源均可选填。多个标签使用逗号分隔，每个账号最多 5 个标签。

- 单次最多导入 100 个账号，支持 `.xlsx` 和 `.csv`，文件不超过 5MB。
- 当前自动解析支持小红书账号主页；上传后先展示新增、更新和异常数据，存在异常时不会写入数据库。
+- 单次最多导入 10,000 个账号，支持 `.xlsx` 和 `.csv`，文件不超过 20MB。
+- 当前自动解析支持小红书账号主页；上传后先展示新增、更新和异常数据。异常行会跳过，其余有效账号可以正常导入。
 - 按“平台 + 账号主页”去重；解析出相同小红书号时也会更新已有账号。
 - 重复账号更新公开资料和合作来源，不产生两份资源。
- 导入的合作来源会进入现有资源搜索、筛选和导出结果。
+- 大批量导入会先写入资源库，再在后台逐步补全缺失的公开资料。
+- KOC 使用手机号或微信号领取任务后，系统会把该值写入“当前联系人”；原“合作来源”继续保留渠道信息。
+- 导入的标签、当前联系人和合作来源会进入资源搜索或导出结果。
+
+## KOC 批量回填 Excel
+
+KOC 领取端支持导出和上传批量回填表。视频任务只生成“序号、标题、笔记内容、视频、发布链接、笔记截图、数据分析截图”列，不生成“图片”列。视频链接通过当前公网域名生成，下载接口返回可播放的 `.mp4` 附件。
+
+反向代理部署必须正确传递 `Host`、`X-Forwarded-Host` 和 `X-Forwarded-Proto`，并把 `APP_ORIGIN` 配置为实际公网地址；不要填写 `localhost` 或容器内部地址。

 ## Agent MCP

--- a/app/admin-app.tsx
+++ b/app/admin-app.tsx
--- a/app/api/action/route.ts
+++ b/app/api/action/route.ts
@@ -2,7 +2,10 @@ import { getRuntimeEnv } from "../../../lib/runtime-env";
 import { runInBackground } from "../../../lib/background";

 const env = getRuntimeEnv();
-import { backfillAccountProfiles } from "../../../lib/account-enrichment-service";
+import {
+  backfillAccountProfiles,
+  enrichDistributionAccount,
+} from "../../../lib/account-enrichment-service";
 import {
  ensureSchema,
  getDashboardData,
@@ -34,7 +37,23 @@ import {
  DistributionReleaseError,
  releaseUnfinishedDistribution,
 } from "../../../lib/distribution-release-service";
+import {
+  resolveWecomConfig,
+  sendWecomAppMessage,
+  sendWecomRobotMessage,
+  listExternalContacts,
+  WecomClientError,
+  type WecomBindings,
+} from "../../../lib/wecom-client";
+import { runDueSoonWecomNotifications } from "../../../lib/wecom-notifier-service";
+import {
+  listGroupChatRows,
+  listGroupPushes,
+  pushTaskToGroupChats,
+  syncGroupChats,
+} from "../../../lib/wecom-group-push-service";
 import { isManagerRequest } from "../../../lib/user-auth";
+import { extractPublishUrl } from "../../../lib/publish-url";

 type ActionBody = {
  action?: string;
@@ -63,6 +82,14 @@ export async function POST(request: Request) {
        sheetName: source.sheetName,
        syncedAt: source.syncedAt,
        rowCount: source.rows.length,
+        imageCount: source.rows.reduce(
+          (total, row) => total + row.images.length,
+          0,
+        ),
+        videoCount: source.rows.reduce(
+          (total, row) => total + row.videos.length,
+          0,
+        ),
        columns: source.columns,
        preview: source.rows.slice(0, 3),
      });
@@ -72,6 +99,8 @@ export async function POST(request: Request) {
      const name = String(body.name ?? "").trim();
      const brand = String(body.brand ?? "").trim();
      const dueAt = String(body.dueAt ?? "").trim();
+      const platform = body.platform === "抖音" ? "抖音" : "小红书";
+      const contentFormat = body.contentFormat === "video" ? "video" : "image_text";
      if (!name || !brand || !dueAt) {
        return Response.json(
          { error: "请补全任务名称、品牌和截止日期" },
@@ -84,6 +113,8 @@ export async function POST(request: Request) {
          name,
          brand,
          dueAt,
+          platform,
+          contentFormat,
        },
        env as unknown as FeishuBindings,
      );
@@ -146,6 +177,159 @@ export async function POST(request: Request) {
        db,
        String(body.distributionId ?? "").trim(),
      );
+    } else if (body.action === "update_distribution_publish_url") {
+      if (!(await isManagerRequest(request))) return adminForbidden();
+      const distributionId = String(body.distributionId ?? "").trim();
+      if (!distributionId) {
+        return Response.json(
+          { error: "作品记录不存在" },
+          { status: 400 },
+        );
+      }
+      const current = await db
+        .prepare(
+          `SELECT d.id, d.task_id, d.partner_id, d.publish_url,
+                  t.task_type, t.platform, t.collection_start_date, t.collection_days,
+                  COALESCE(a.nickname, '待识别账号') AS account_nickname
+           FROM distributions d
+           JOIN tasks t ON t.id = d.task_id
+           LEFT JOIN accounts a ON a.id = d.account_id
+           WHERE d.id = ?`,
+        )
+        .bind(distributionId)
+        .first<{
+          id: string;
+          task_id: string;
+          partner_id: string;
+          publish_url: string | null;
+          task_type?: string | null;
+          platform: string;
+          collection_start_date: string | null;
+          collection_days: string;
+          account_nickname: string;
+        }>();
+      if (!current) {
+        return Response.json({ error: "作品记录不存在" }, { status: 404 });
+      }
+      if (current.task_type === "screenshot_collect") {
+        return Response.json(
+          { error: "截图回收任务不需要填写发布链接" },
+          { status: 400 },
+        );
+      }
+      const platform = current.platform === "抖音" ? "抖音" : "小红书";
+      const publishUrl = extractPublishUrl(
+        String(body.publishUrl ?? "").trim(),
+        platform,
+      );
+      if (!publishUrl) {
+        return Response.json(
+          { error: `请填写包含${platform}作品链接的发布内容` },
+          { status: 400 },
+        );
+      }
+      if (current.publish_url === publishUrl) {
+        return Response.json(await getDashboardData());
+      }
+      let collectionDays: number[] = [];
+      try {
+        const parsed = JSON.parse(current.collection_days || "[]");
+        if (Array.isArray(parsed)) {
+          collectionDays = [...new Set(parsed.map(Number))]
+            .filter(
+              (day) =>
+                Number.isInteger(day) && day >= 1 && day <= 7,
+            )
+            .sort((a, b) => a - b);
+        }
+      } catch {
+        collectionDays = [];
+      }
+      const isScheduled = Boolean(
+        current.collection_start_date && collectionDays.length > 0,
+      );
+      const statements = [
+        db
+          .prepare(
+            `UPDATE distributions SET
+              publish_url = ?,
+              publish_time = CURRENT_TIMESTAMP,
+              status = 'published',
+              d2_likes = NULL,
+              d2_comments = NULL,
+              d2_collects = NULL,
+              d5_likes = NULL,
+              d5_comments = NULL,
+              d5_collects = NULL,
+              d7_likes = NULL,
+              d7_comments = NULL,
+              d7_collects = NULL,
+              latest_likes = NULL,
+              latest_comments = NULL,
+              latest_collects = NULL,
+              latest_shares = NULL,
+              collection_status = ?,
+              collection_status_description = ?,
+              collection_updated_at = NULL,
+              last_collection_day = NULL,
+              updated_at = CURRENT_TIMESTAMP
+             WHERE id = ?`,
+          )
+          .bind(
+            publishUrl,
+            isScheduled ? "scheduled" : "pending",
+            isScheduled
+              ? `管理员已更新链接，等待${collectionDays.length}个采集日`
+              : "管理员已更新链接，等待设置采集计划",
+            distributionId,
+          ),
+        db
+          .prepare("DELETE FROM collection_runs WHERE distribution_id = ?")
+          .bind(distributionId),
+      ];
+      if (!current.publish_url) {
+        statements.push(
+          db
+            .prepare(
+              `UPDATE partners SET completed_total = completed_total + 1
+               WHERE id = ?`,
+            )
+            .bind(current.partner_id),
+        );
+      }
+      await db.batch(statements);
+      if (isScheduled && current.collection_start_date) {
+        await createCollectionRunTasks(
+          db,
+          current.task_id,
+          current.collection_start_date,
+          collectionDays,
+        );
+        runInBackground(
+          runDueScheduledCollections(
+            db,
+            Date.now(),
+            resolveCollectionMcpConfig(
+              env as unknown as CollectionMcpBindings,
+            ),
+            "catchup",
+            current.task_id,
+          ).catch(() => undefined),
+          "collection catchup after publish URL update",
+        );
+      }
+      runInBackground(
+        enrichDistributionAccount(
+          db,
+          distributionId,
+          publishUrl,
+          current.account_nickname,
+          resolveCollectionMcpConfig(
+            env as unknown as CollectionMcpBindings,
+          ),
+        ).catch(() => undefined),
+        "account enrichment after publish URL update",
+      );
    } else if (body.action === "save_collection_schedule") {
      const taskId = String(body.taskId ?? "").trim();
      const startDate = String(body.startDate ?? "").trim();
@@ -365,6 +549,143 @@ export async function POST(request: Request) {
        )
        .bind(exposure, views, distributionId)
        .run();
+    } else if (body.action === "bind_wecom_external_id") {
+      const partnerId = String(body.partnerId ?? "").trim().slice(0, 80);
+      const externalId = String(body.wecomExternalUserId ?? "")
+        .trim()
+        .slice(0, 128);
+      const wecomName = String(body.wecomName ?? "").trim().slice(0, 120);
+      if (!partnerId) {
+        return Response.json(
+          { error: "缺少 partnerId" },
+          { status: 400 },
+        );
+      }
+      if (wecomName) {
+        await db
+          .prepare(
+            "UPDATE partners SET wecom_external_user_id = ?, wecom_name = ? WHERE id = ?",
+          )
+          .bind(externalId || null, wecomName, partnerId)
+          .run();
+      } else {
+        await db
+          .prepare(
+            "UPDATE partners SET wecom_external_user_id = ? WHERE id = ?",
+          )
+          .bind(externalId || null, partnerId)
+          .run();
+      }
+      return Response.json({
+        partnerId,
+        wecomExternalUserId: externalId || null,
+        wecomName,
+      });
+    } else if (body.action === "send_test_wecom") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      const partnerId = String(body.partnerId ?? "").trim();
+      let partnerExternalId: string | null = null;
+      if (partnerId) {
+        const row = await db
+          .prepare(
+            "SELECT wecom_external_user_id FROM partners WHERE id = ?",
+          )
+          .bind(partnerId)
+          .first<{ wecom_external_user_id: string | null }>();
+        partnerExternalId = row?.wecom_external_user_id ?? null;
+      }
+      const testContent = `[KOC LOOP 测试] 群机器人连通性正常，时间 ${new Date().toISOString()}`;
+      let robotStatus: "ok" | "skipped" = "skipped";
+      if (wecomConfig.robotWebhook) {
+        await sendWecomRobotMessage(testContent, wecomConfig);
+        robotStatus = "ok";
+      }
+      let appStatus: "ok" | "skipped" | "failed" = "skipped";
+      if (
+        partnerExternalId &&
+        wecomConfig.corpId &&
+        wecomConfig.agentId &&
+        wecomConfig.secret
+      ) {
+        const result = await sendWecomAppMessage(
+          [partnerExternalId],
+          testContent,
+          wecomConfig,
+        );
+        appStatus = result.failed > 0 ? "failed" : "ok";
+      }
+      return Response.json({
+        robot: robotStatus,
+        app: appStatus,
+      });
+    } else if (body.action === "wecom_status") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      return Response.json({
+        robotConfigured: Boolean(wecomConfig.robotWebhook),
+        appConfigured: Boolean(
+          wecomConfig.corpId && wecomConfig.agentId && wecomConfig.secret,
+        ),
+        dueDays: wecomConfig.dueDays,
+      });
+    } else if (body.action === "trigger_wecom_due_soon") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      const summary = await runDueSoonWecomNotifications(
+        db,
+        wecomConfig,
+      );
+      return Response.json(summary);
+    } else if (body.action === "wecom_list_external_contacts") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      const contacts = await listExternalContacts(wecomConfig);
+      return Response.json({ contacts });
+    } else if (body.action === "wecom_sync_group_chats") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      const groups = await syncGroupChats(db, wecomConfig);
+      return Response.json({ groups, syncedCount: groups.length });
+    } else if (body.action === "wecom_push_task_to_groups") {
+      const wecomConfig = resolveWecomConfig(
+        env as unknown as WecomBindings,
+      );
+      const summary = await pushTaskToGroupChats(
+        db,
+        {
+          taskId: String(body.taskId ?? ""),
+          chatIds: body.chatIds,
+          text: body.text,
+        },
+        wecomConfig,
+      );
+      const failCount = summary.results.reduce(
+        (total, item) => total + item.failList.length,
+        0,
+      );
+      const groupCount = summary.results.reduce(
+        (total, item) => total + item.chatCount,
+        0,
+      );
+      return Response.json({
+        results: summary.results,
+        groupCount,
+        failCount,
+        hint: "群发任务已创建，群主需在企微客户端「群发助手」点击发送后，消息才会送达客户群",
+      });
+    } else if (body.action === "wecom_task_group_pushes") {
+      const taskId = String(body.taskId ?? "").trim();
+      const pushes = await listGroupPushes(db, taskId || undefined);
+      return Response.json({ pushes });
+    } else if (body.action === "wecom_group_chats") {
+      const groups = await listGroupChatRows(db);
+      return Response.json({ groups });
    } else {
      return Response.json({ error: "不支持的操作" }, { status: 400 });
    }
@@ -376,7 +697,8 @@ export async function POST(request: Request) {
      {
        status:
          error instanceof FeishuSourceError ||
-          error instanceof DistributionReleaseError
+          error instanceof DistributionReleaseError ||
+          error instanceof WecomClientError
            ? error.status
            : 500,
      },
--- a/app/api/mcp/route.ts
+++ b/app/api/mcp/route.ts
@@ -32,6 +32,8 @@ const toolOutputSchema = z.object({
  due_date: z.string(),
  sheet_name: z.string(),
  note_count: z.number().int().nonnegative(),
+  platform: z.enum(["小红书", "抖音"]),
+  content_format: z.enum(["image_text", "video"]),
  claim_url: z.string().url(),
 });

@@ -57,7 +59,7 @@ function createServer(context: McpRequestContext) {
    {
      title: "创建 KOC 分发任务",
      description:
-        "读取飞书电子表格中的标题、正文和配图，在 KOC LOOP 创建分发任务，并返回可直接发给 KOC 的领取链接。飞书表格若有多个工作表，链接必须包含目标 sheet 参数。",
+        "读取飞书电子表格中的标题、正文及图片或视频，在 KOC LOOP 创建小红书/抖音分发任务，并返回可直接发给 KOC 的领取链接。飞书表格若有多个工作表，链接必须包含目标 sheet 参数。",
      inputSchema: z.object({
        feishu_url: z
          .string()
@@ -74,6 +76,14 @@ function createServer(context: McpRequestContext) {
          .max(100)
          .optional()
          .describe("品牌或项目名称；未提供时系统记录为“未设置项目”"),
+        platform: z
+          .enum(["小红书", "抖音"])
+          .optional()
+          .describe("发布平台，默认小红书"),
+        content_format: z
+          .enum(["image_text", "video"])
+          .optional()
+          .describe("内容形式：image_text 图文，video 视频；默认图文"),
      }),
      outputSchema: toolOutputSchema,
      annotations: {
@@ -83,7 +93,7 @@ function createServer(context: McpRequestContext) {
        openWorldHint: true,
      },
    },
-    async ({ feishu_url, task_name, due_date, brand_project }) => {
+    async ({ feishu_url, task_name, due_date, brand_project, platform, content_format }) => {
      try {
        const bindings = getBindings();
        const portalUrl = String(bindings.KOC_PORTAL_URL ?? "").trim();
@@ -96,6 +106,8 @@ function createServer(context: McpRequestContext) {
            name: task_name,
            brand: brand_project?.trim() || "未设置项目",
            dueAt: due_date,
+            platform: platform ?? "小红书",
+            contentFormat: content_format ?? "image_text",
          },
          bindings,
          { deduplicate: true },
@@ -108,6 +120,8 @@ function createServer(context: McpRequestContext) {
          due_date: result.dueAt,
          sheet_name: result.sheetName,
          note_count: result.noteCount,
+          platform: result.platform,
+          content_format: result.contentFormat,
          claim_url: buildClaimUrl(portalUrl, result.shareToken),
        };
        const actionText = result.created ? "已创建" : "已找到相同任务";
@@ -115,7 +129,7 @@ function createServer(context: McpRequestContext) {
          content: [
            {
              type: "text",
-              text: `${actionText}“${result.name}”，共 ${result.noteCount} 篇笔记。领取链接：${output.claim_url}`,
+              text: `${actionText}“${result.name}”，平台：${result.platform}，内容形式：${result.contentFormat === "video" ? "视频" : "图文"}，共 ${result.noteCount} 篇。领取链接：${output.claim_url}`,
            },
          ],
          structuredContent: output,
--- a/app/api/partner-batch-workbook/route.ts
+++ b/app/api/partner-batch-workbook/route.ts
@@ -0,0 +1,750 @@
+import { getRuntimeEnv } from "../../../lib/runtime-env";
+import { runInBackground } from "../../../lib/background";
+import type { DatabaseStatement } from "../../../lib/database";
+import {
+  downloadFeishuMedia,
+  type FeishuBindings,
+} from "../../../lib/feishu-client";
+import {
+  ensureSchema,
+  getRawDb,
+  getUploadBucket,
+  hashText,
+  uid,
+} from "../../../lib/mvp-db";
+import {
+  PARTNER_BATCH_MAX_BYTES,
+  buildPartnerBatchWorkbookColumns,
+  parsePartnerBatchWorkbook,
+  resolvePartnerWorkbookOrigin,
+} from "../../../lib/partner-batch-workbook";
+import {
+  accountFromPublishLink,
+} from "../../../lib/partner-utils";
+import { extractPublishUrl } from "../../../lib/publish-url";
+import {
+  partnerOptions,
+  withPartnerCors,
+} from "../../../lib/partner-cors";
+import {
+  buildRecoveryWorkbook,
+  type RecoveryWorkbookImage,
+  type RecoveryWorkbookRow,
+} from "../../../lib/recovery-workbook";
+import {
+  resolveCollectionMcpConfig,
+  type CollectionMcpBindings,
+} from "../../../lib/mcp-collection-client";
+import { enrichDistributionAccount } from "../../../lib/account-enrichment-service";
+import {
+  createCollectionRunTasks,
+  runDueScheduledCollections,
+} from "../../../lib/collection-service";
+import { normalizeWorkbookImage } from "../../../lib/workbook-image";
+
+const env = getRuntimeEnv();
+
+type StoredAsset = {
+  index: number;
+  key: string;
+  fileToken?: string;
+  width?: number | null;
+  height?: number | null;
+};
+
+type BatchRow = {
+  distribution_id: string;
+  title: string;
+  body: string;
+  source_row: number | null;
+  image_assets: string;
+  video_assets: string;
+  publish_url: string | null;
+  publish_screenshot_key: string | null;
+  screenshot_key: string | null;
+  partner_id: string;
+  account_id: string | null;
+  claimant_name: string;
+};
+
+type TaskRow = {
+  id: string;
+  name: string;
+  task_type: string;
+  collection_start_date: string | null;
+  collection_days: string;
+  platform: "小红书" | "抖音";
+  content_format: "image_text" | "video";
+};
+
+type BatchAccess = {
+  task: TaskRow;
+  rows: BatchRow[];
+};
+
+function textValue(value: string | null, maxLength = 100) {
+  return String(value ?? "").trim().slice(0, maxLength);
+}
+
+function safeFileName(value: string) {
+  return value.replace(/[\\/:*?"<>|]/g, " ").trim().slice(0, 60) || "领取笔记";
+}
+
+function exactArrayBuffer(bytes: Uint8Array) {
+  const copy = new Uint8Array(bytes.byteLength);
+  copy.set(bytes);
+  return copy.buffer;
+}
+
+function contentTypeFromObject(object: { writeHttpMetadata(headers: Headers): void }) {
+  const headers = new Headers();
+  object.writeHttpMetadata(headers);
+  return headers.get("Content-Type") || "application/octet-stream";
+}
+
+function parseAssets(
+  value: string,
+  prefixes = ["content-assets/", "task-assets/"],
+) {
+  try {
+    const assets = JSON.parse(value || "[]") as StoredAsset[];
+    return Array.isArray(assets)
+      ? assets
+          .filter(
+            (asset) =>
+              Number.isInteger(Number(asset.index)) &&
+              Number(asset.index) > 0 &&
+              typeof asset.key === "string" &&
+              prefixes.some((prefix) => asset.key.startsWith(prefix)),
+          )
+          .map((asset) => ({ ...asset, index: Number(asset.index) }))
+          .sort((left, right) => left.index - right.index)
+      : [];
+  } catch {
+    return [];
+  }
+}
+
+async function loadImage(
+  key: string,
+  description: string,
+  fileToken?: string,
+  width?: number | null,
+  height?: number | null,
+  compactSource = false,
+) {
+  const bucket = getUploadBucket();
+  let object = await bucket.get(key);
+  if (!object && fileToken) {
+    const media = await downloadFeishuMedia(
+      fileToken,
+      env as unknown as FeishuBindings,
+    );
+    await bucket.put(key, media.bytes, {
+      httpMetadata: { contentType: media.contentType },
+      customMetadata: { source: "feishu-api" },
+    });
+    object = await bucket.get(key);
+  }
+  if (!object) return null;
+  return normalizeWorkbookImage(
+    {
+      bytes: new Uint8Array(await object.arrayBuffer()),
+      contentType: contentTypeFromObject(object),
+      description,
+      width,
+      height,
+    } satisfies RecoveryWorkbookImage,
+    compactSource
+      ? { maxDimension: 1600, outputFormat: "jpeg", jpegQuality: 82 }
+      : undefined,
+  );
+}
+
+async function findAccess(
+  taskToken: string,
+  claimToken: string,
+  delegationToken: string,
+): Promise<BatchAccess | null> {
+  const db = getRawDb();
+  const task = delegationToken
+    ? await db
+        .prepare(
+          `SELECT t.id, t.name, t.task_type, t.collection_start_date, t.collection_days,
+                  t.platform, t.content_format
+           FROM tasks t
+           JOIN delegation_bundles b ON b.task_id = t.id
+           WHERE b.share_token = ? AND b.status = 'active'`,
+        )
+        .bind(delegationToken)
+        .first<TaskRow>()
+    : await db
+        .prepare(
+          `SELECT t.id, t.name, t.task_type, t.collection_start_date, t.collection_days,
+                  t.platform, t.content_format
+           FROM tasks t
+           JOIN claims cl ON cl.task_id = t.id
+           WHERE t.share_token = ? AND cl.claim_token = ?`,
+        )
+        .bind(taskToken, claimToken)
+        .first<TaskRow>();
+  if (!task) return null;
+  const rows = delegationToken
+    ? await db
+        .prepare(
+          `SELECT d.id AS distribution_id, d.partner_id, d.account_id,
+                  d.publish_url, d.publish_screenshot_key, d.screenshot_key,
+                  c.title, c.body, c.source_row, c.image_assets, c.video_assets,
+                  cl.claimant_name
+           FROM distributions d
+           JOIN contents c ON c.id = d.content_id
+           JOIN delegation_bundles b ON b.id = d.delegation_bundle_id
+           JOIN claims cl ON cl.id = d.claim_id
+           WHERE b.share_token = ? AND b.task_id = ? AND b.status = 'active'
+           ORDER BY d.claimed_at, d.id`,
+        )
+        .bind(delegationToken, task.id)
+        .all<BatchRow>()
+    : await db
+        .prepare(
+          `SELECT d.id AS distribution_id, d.partner_id, d.account_id,
+                  d.publish_url, d.publish_screenshot_key, d.screenshot_key,
+                  c.title, c.body, c.source_row, c.image_assets, c.video_assets,
+                  cl.claimant_name
+           FROM distributions d
+           JOIN contents c ON c.id = d.content_id
+           JOIN claims cl ON cl.id = d.claim_id
+           WHERE cl.claim_token = ? AND cl.task_id = ?
+           ORDER BY d.claimed_at, d.id`,
+        )
+        .bind(claimToken, task.id)
+        .all<BatchRow>();
+  return { task, rows: rows.results };
+}
+
+async function handleGet(request: Request) {
+  try {
+    await ensureSchema();
+    const url = new URL(request.url);
+    const taskToken = textValue(url.searchParams.get("task"));
+    const claimToken = textValue(url.searchParams.get("claim"));
+    const delegationToken = textValue(url.searchParams.get("share"));
+    if (!delegationToken && (!taskToken || !claimToken)) {
+      return Response.json({ error: "领取凭证不完整" }, { status: 400 });
+    }
+    const access = await findAccess(taskToken, claimToken, delegationToken);
+    if (!access) return Response.json({ error: "领取凭证无效" }, { status: 403 });
+    if (access.task.task_type !== "content_publish") {
+      return Response.json({ error: "当前任务不支持笔记批量回填" }, { status: 400 });
+    }
+    const maxSourceImages = Math.max(
+      0,
+      ...access.rows.map((row) => parseAssets(row.image_assets).length),
+    );
+    const maxSourceVideos = Math.max(
+      0,
+      ...access.rows.map(
+        (row) => parseAssets(row.video_assets, ["content-videos/"]).length,
+      ),
+    );
+    const columns = buildPartnerBatchWorkbookColumns({
+      contentFormat: access.task.content_format,
+      maxSourceImages,
+      maxSourceVideos,
+    });
+    const downloadOrigin = resolvePartnerWorkbookOrigin(
+      request,
+      env.APP_ORIGIN,
+    );
+    const workbookRows: RecoveryWorkbookRow[] = [];
+    for (let rowIndex = 0; rowIndex < access.rows.length; rowIndex += 1) {
+      const row = access.rows[rowIndex];
+      const images: RecoveryWorkbookRow["images"] = [];
+      const hyperlinks: NonNullable<RecoveryWorkbookRow["hyperlinks"]> = [];
+      const assets =
+        columns.sourceImageCount > 0 ? parseAssets(row.image_assets) : [];
+      for (let index = 0; index < assets.length; index += 1) {
+        const asset = assets[index];
+        const image = await loadImage(
+          asset.key,
+          `${row.title} 原图${asset.index}`,
+          asset.fileToken,
+          asset.width,
+          asset.height,
+          true,
+        );
+        if (image) {
+          images.push({
+            column: columns.sourceImageStartColumn + index,
+            image,
+            maxWidth: 160,
+            maxHeight: 118,
+          });
+        }
+      }
+      if (row.publish_screenshot_key) {
+        const image = await loadImage(
+          row.publish_screenshot_key,
+          `${row.title} 笔记截图`,
+        );
+        if (image) {
+          images.push({ column: columns.publishScreenshotColumn, image });
+        }
+      }
+      if (row.screenshot_key) {
+        const image = await loadImage(
+          row.screenshot_key,
+          `${row.title} 数据分析截图`,
+        );
+        if (image) {
+          images.push({ column: columns.creatorScreenshotColumn, image });
+        }
+      }
+      const videoAssets = parseAssets(row.video_assets, ["content-videos/"]);
+      for (let index = 0; index < videoAssets.length; index += 1) {
+        const params = new URLSearchParams({
+          distribution: row.distribution_id,
+          index: String(videoAssets[index].index),
+          kind: "video",
+          download: "1",
+        });
+        if (delegationToken) params.set("share", delegationToken);
+        else {
+          params.set("task", taskToken);
+          params.set("claim", claimToken);
+        }
+        hyperlinks.push({
+          column: columns.sourceVideoStartColumn + index,
+          url: `${downloadOrigin}/api/partner-image?${params}`,
+        });
+      }
+      workbookRows.push({
+        cells: [
+          rowIndex + 1,
+          row.title,
+          row.body,
+          ...Array.from({ length: columns.sourceImageCount }, () => ""),
+          ...Array.from(
+            { length: columns.sourceVideoCount },
+            (_, index) => (index < videoAssets.length ? `下载视频${index + 1}` : ""),
+          ),
+          row.publish_url || "",
+          "",
+          "",
+          row.distribution_id,
+          row.publish_screenshot_key || "",
+          row.screenshot_key || "",
+        ],
+        images,
+        hyperlinks: [
+          ...hyperlinks,
+          ...(row.publish_url
+            ? [{ column: columns.publishUrlColumn, url: row.publish_url }]
+            : []),
+        ],
+      });
+    }
+    const workbook = buildRecoveryWorkbook({
+      sheetName: "批量回填",
+      headers: columns.headers,
+      columnWidths: columns.columnWidths,
+      rows: workbookRows,
+      hiddenColumns: [
+        columns.systemColumn,
+        columns.systemColumn + 1,
+        columns.systemColumn + 2,
+      ],
+    });
+    const fileName = `${safeFileName(access.task.name)}-批量回填.xlsx`;
+    return new Response(exactArrayBuffer(workbook), {
+      headers: {
+        "Cache-Control": "private, no-store",
+        "Content-Type":
+          "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        "Content-Disposition": `attachment; filename="koc-batch.xlsx"; filename*=UTF-8''${encodeURIComponent(fileName)}`,
+        "X-Content-Type-Options": "nosniff",
+      },
+    });
+  } catch (error) {
+    return Response.json(
+      { error: error instanceof Error ? error.message : "导出失败" },
+      { status: 500 },
+    );
+  }
+}
+
+function extensionForImage(image: { fileName: string; contentType: string }) {
+  const fromName = image.fileName.split(".").at(-1)?.replace(/[^a-zA-Z0-9]/g, "");
+  if (fromName) return fromName.toLowerCase() === "jpeg" ? "jpg" : fromName;
+  return image.contentType.includes("png")
+    ? "png"
+    : image.contentType.includes("webp")
+      ? "webp"
+      : image.contentType.includes("gif")
+        ? "gif"
+        : "jpg";
+}
+
+async function storeImportedImage(
+  kind: "publish" | "creator",
+  distributionId: string,
+  image: { bytes: Uint8Array; contentType: string; fileName: string },
+) {
+  const prefix = kind === "publish" ? "publish-evidence" : "creator-center";
+  const key = `${prefix}/${distributionId}/${uid("sheet")}.${extensionForImage(image)}`;
+  await getUploadBucket().put(key, image.bytes, {
+    httpMetadata: { contentType: image.contentType },
+    customMetadata: { source: "partner-batch-workbook" },
+  });
+  return key;
+}
+
+async function isDifferentFromStoredImage(
+  existingKey: string | null,
+  image: { bytes: Uint8Array },
+) {
+  if (!existingKey) return true;
+  const stored = await getUploadBucket().get(existingKey);
+  if (!stored) return true;
+  const storedBytes = new Uint8Array(await stored.arrayBuffer());
+  if (storedBytes.byteLength !== image.bytes.byteLength) return true;
+  for (let index = 0; index < storedBytes.byteLength; index += 1) {
+    if (storedBytes[index] !== image.bytes[index]) return true;
+  }
+  return false;
+}
+
+async function handlePost(request: Request) {
+  try {
+    await ensureSchema();
+    const form = await request.formData();
+    const file = form.get("file");
+    if (!(file instanceof File) || file.size === 0) {
+      return Response.json({ error: "请选择填写完成的Excel表" }, { status: 400 });
+    }
+    if (file.size > PARTNER_BATCH_MAX_BYTES) {
+      return Response.json({ error: "批量回填表不能超过80MB" }, { status: 400 });
+    }
+    if (!/\.xlsx$/i.test(file.name)) {
+      return Response.json({ error: "仅支持系统导出的 .xlsx 表格" }, { status: 400 });
+    }
+    const taskToken = textValue(String(form.get("taskToken") ?? ""));
+    const claimToken = textValue(String(form.get("claimToken") ?? ""));
+    const delegationToken = textValue(String(form.get("delegationToken") ?? ""));
+    const access = await findAccess(taskToken, claimToken, delegationToken);
+    if (!access) return Response.json({ error: "领取凭证无效" }, { status: 403 });
+    if (access.task.task_type !== "content_publish") {
+      return Response.json({ error: "当前任务不支持笔记批量回填" }, { status: 400 });
+    }
+    const importedRows = parsePartnerBatchWorkbook(await file.arrayBuffer());
+    const assignmentById = new Map(
+      access.rows.map((row) => [row.distribution_id, row]),
+    );
+    const seen = new Set<string>();
+    const errors: string[] = [];
+    const prepared = importedRows.map((row) => {
+      const assignment = assignmentById.get(row.distributionId);
+      if (!assignment) {
+        errors.push(`第${row.spreadsheetRow}行不属于当前领取批次，请重新导出表格`);
+      } else if (seen.has(row.distributionId)) {
+        errors.push(`第${row.spreadsheetRow}行笔记重复`);
+      } else if (row.title && row.title !== assignment.title) {
+        errors.push(`第${row.spreadsheetRow}行标题已被修改，请重新导出表格`);
+      }
+      seen.add(row.distributionId);
+      const publishUrl = row.publishUrl
+        ? extractPublishUrl(row.publishUrl, access.task.platform)
+        : assignment?.publish_url || "";
+      if (row.publishUrl && !publishUrl) {
+        errors.push(`第${row.spreadsheetRow}行发布链接不是有效的${access.task.platform}作品链接`);
+      }
+      const hasPublishScreenshot = Boolean(
+        assignment?.publish_screenshot_key || row.publishScreenshot,
+      );
+      if (publishUrl && !hasPublishScreenshot) {
+        errors.push(`第${row.spreadsheetRow}行填写了发布链接，请同时插入笔记截图`);
+      }
+      if (row.publishScreenshot && !publishUrl) {
+        errors.push(`第${row.spreadsheetRow}行插入了笔记截图，请同时填写发布链接`);
+      }
+      if (row.creatorScreenshot && !publishUrl) {
+        errors.push(`第${row.spreadsheetRow}行需先回填发布链接，再补数据分析截图`);
+      }
+      return {
+        imported: row,
+        assignment,
+        publishUrl,
+      };
+    });
+    if (errors.length > 0) {
+      return Response.json(
+        { error: errors.slice(0, 8).join("；"), errors },
+        { status: 400 },
+      );
+    }
+
+    const db = getRawDb();
+    let updatedRows = 0;
+    let publishedCount = 0;
+    let analysisScreenshotCount = 0;
+    let noteScreenshotCount = 0;
+    let publishUrlChangedCount = 0;
+    const enrichments: Array<{ id: string; url: string; nickname: string }> = [];
+    for (const item of prepared) {
+      const assignment = item.assignment!;
+      let publishScreenshotKey = assignment.publish_screenshot_key;
+      let creatorScreenshotKey = assignment.screenshot_key;
+      const hasNewPublishScreenshot = item.imported.publishScreenshot
+        ? await isDifferentFromStoredImage(
+            assignment.publish_screenshot_key,
+            item.imported.publishScreenshot,
+          )
+        : false;
+      const hasNewCreatorScreenshot = item.imported.creatorScreenshot
+        ? await isDifferentFromStoredImage(
+            assignment.screenshot_key,
+            item.imported.creatorScreenshot,
+          )
+        : false;
+      if (hasNewPublishScreenshot && item.imported.publishScreenshot) {
+        publishScreenshotKey = await storeImportedImage(
+          "publish",
+          assignment.distribution_id,
+          item.imported.publishScreenshot,
+        );
+        noteScreenshotCount += 1;
+      }
+      if (hasNewCreatorScreenshot && item.imported.creatorScreenshot) {
+        creatorScreenshotKey = await storeImportedImage(
+          "creator",
+          assignment.distribution_id,
+          item.imported.creatorScreenshot,
+        );
+        analysisScreenshotCount += 1;
+      }
+      const statements: DatabaseStatement[] = [];
+      if (publishScreenshotKey !== assignment.publish_screenshot_key) {
+        statements.push(
+          db
+            .prepare(
+              `UPDATE distributions SET publish_screenshot_key = ?, updated_at = CURRENT_TIMESTAMP WHERE id = ?`,
+            )
+            .bind(publishScreenshotKey, assignment.distribution_id),
+        );
+      }
+      if (creatorScreenshotKey !== assignment.screenshot_key) {
+        statements.push(
+          db
+            .prepare(
+              `UPDATE distributions SET screenshot_key = ?,
+                 ocr_status = CASE WHEN exposure IS NOT NULL AND views IS NOT NULL THEN ocr_status ELSE 'uploaded' END,
+                 updated_at = CURRENT_TIMESTAMP WHERE id = ?`,
+            )
+            .bind(creatorScreenshotKey, assignment.distribution_id),
+        );
+      }
+      if (item.publishUrl && item.publishUrl !== assignment.publish_url) {
+        const isReplacement = Boolean(assignment.publish_url);
+        const account = accountFromPublishLink(
+          item.publishUrl,
+          access.task.platform,
+        );
+        if (!account) {
+          return Response.json(
+            { error: `第${item.imported.spreadsheetRow}行发布链接格式不正确` },
+            { status: 400 },
+          );
+        }
+        const matchedAccount = await db
+          .prepare(
+            `SELECT id FROM accounts
+             WHERE platform = ? AND platform_uid = ?
+             LIMIT 1`,
+          )
+          .bind(account.platform, account.platformUid)
+          .first<{ id: string }>();
+        const accountId =
+          matchedAccount?.id ||
+          `account-${hashText(`${account.platform}:${account.platformUid}`)}`;
+        statements.push(
+          ...(isReplacement
+            ? [
+                db
+                  .prepare(
+                    `UPDATE distributions SET
+                      d2_likes = NULL, d2_comments = NULL, d2_collects = NULL,
+                      d5_likes = NULL, d5_comments = NULL, d5_collects = NULL,
+                      d7_likes = NULL, d7_comments = NULL, d7_collects = NULL,
+                      latest_likes = NULL, latest_comments = NULL,
+                      latest_collects = NULL, latest_shares = NULL,
+                      collection_status = 'pending',
+                      collection_status_description = '批量回填已更新链接，等待重新采集',
+                      collection_updated_at = NULL, last_collection_day = NULL,
+                      updated_at = CURRENT_TIMESTAMP
+                     WHERE id = ?`,
+                  )
+                  .bind(assignment.distribution_id),
+                db
+                  .prepare("DELETE FROM collection_runs WHERE distribution_id = ?")
+                  .bind(assignment.distribution_id),
+              ]
+            : []),
+          db
+            .prepare(
+              `INSERT INTO accounts
+                (id, platform, platform_uid, nickname, profile_url,
+                 current_contact, post_count)
+               VALUES (?, ?, ?, ?, ?, ?, 1)
+               ON CONFLICT(platform, platform_uid) DO UPDATE SET
+                 nickname = excluded.nickname,
+                 profile_url = excluded.profile_url,
+                 current_contact = CASE
+                   WHEN excluded.current_contact != ''
+                     THEN excluded.current_contact
+                   ELSE accounts.current_contact
+                 END,
+                 last_seen_at = CURRENT_TIMESTAMP`,
+            )
+            .bind(
+              accountId,
+              account.platform,
+              account.platformUid,
+              account.nickname,
+              account.profileUrl,
+              assignment.claimant_name,
+            ),
+          db
+            .prepare(
+              `UPDATE distributions SET account_id = ?, publish_url = ?,
+                 publish_time = COALESCE(publish_time, CURRENT_TIMESTAMP),
+                 status = 'published', updated_at = CURRENT_TIMESTAMP
+               WHERE id = ?`,
+            )
+            .bind(accountId, item.publishUrl, assignment.distribution_id),
+        );
+        statements.push(
+          db
+            .prepare(
+              `UPDATE accounts SET post_count = (
+                 SELECT COUNT(*) FROM distributions WHERE account_id = ?
+               ) WHERE id = ?`,
+            )
+            .bind(accountId, accountId),
+        );
+        if (assignment.account_id && assignment.account_id !== accountId) {
+          statements.push(
+            db
+              .prepare(
+                `UPDATE accounts SET post_count = (
+                   SELECT COUNT(*) FROM distributions WHERE account_id = ?
+                 ) WHERE id = ?`,
+              )
+              .bind(assignment.account_id, assignment.account_id),
+          );
+        }
+        if (isReplacement) publishUrlChangedCount += 1;
+        if (!assignment.publish_url) {
+          statements.push(
+            db
+              .prepare(
+                `UPDATE partners SET completed_total = completed_total + 1 WHERE id = ?`,
+              )
+              .bind(assignment.partner_id),
+          );
+          publishedCount += 1;
+        }
+        enrichments.push({
+          id: assignment.distribution_id,
+          url: item.publishUrl,
+          nickname: account.nickname,
+        });
+      }
+      if (statements.length > 0) {
+        await db.batch(statements);
+        updatedRows += 1;
+      }
+    }
+    if (
+      (publishedCount > 0 || publishUrlChangedCount > 0) &&
+      access.task.collection_start_date &&
+      access.task.collection_days !== "[]"
+    ) {
+      let collectionDays: number[] = [];
+      try {
+        const parsed = JSON.parse(access.task.collection_days);
+        if (Array.isArray(parsed)) collectionDays = parsed.map(Number);
+      } catch {
+        collectionDays = [];
+      }
+      if (collectionDays.length > 0) {
+        await db
+          .prepare(
+            `UPDATE distributions SET collection_status = 'scheduled',
+               collection_status_description = ?
+             WHERE task_id = ? AND publish_url IS NOT NULL AND publish_url != ''`,
+          )
+          .bind(
+            `已安排${collectionDays.length}个采集日，每日09:00执行`,
+            access.task.id,
+          )
+          .run();
+        await createCollectionRunTasks(
+          db,
+          access.task.id,
+          access.task.collection_start_date,
+          collectionDays,
+        );
+        runInBackground(
+          runDueScheduledCollections(
+            db,
+            Date.now(),
+            resolveCollectionMcpConfig(
+              env as unknown as CollectionMcpBindings,
+            ),
+            "catchup",
+            access.task.id,
+          ).catch(() => undefined),
+          "collection catchup after batch publish update",
+        );
+      }
+    }
+    for (const enrichment of enrichments) {
+      runInBackground(
+        enrichDistributionAccount(
+          db,
+          enrichment.id,
+          enrichment.url,
+          enrichment.nickname,
+          resolveCollectionMcpConfig(env as unknown as CollectionMcpBindings),
+        ).catch(() => undefined),
+        "batch distribution account enrichment",
+      );
+    }
+    return Response.json({
+      imported: true,
+      updatedRows,
+      publishedCount,
+      publishUrlChangedCount,
+      noteScreenshotCount,
+      analysisScreenshotCount,
+      skippedRows: importedRows.length - updatedRows,
+    });
+  } catch (error) {
+    return Response.json(
+      { error: error instanceof Error ? error.message : "批量回填失败" },
+      { status: 500 },
+    );
+  }
+}
+
+export async function GET(request: Request) {
+  return withPartnerCors(request, await handleGet(request));
+}
+
+export async function POST(request: Request) {
+  return withPartnerCors(request, await handlePost(request));
+}
+
+export async function OPTIONS(request: Request) {
+  return partnerOptions(request);
+}
--- a/app/api/partner-image/route.ts
+++ b/app/api/partner-image/route.ts
@@ -13,6 +13,7 @@ import {
  withPartnerCors,
 } from "../../../lib/partner-cors";
 import { parseResultScreenshotKeys } from "../../../lib/result-screenshots";
+import { hasMp4FileSignature } from "../../../lib/video-file";

 const env = getRuntimeEnv();

@@ -26,7 +27,11 @@ function textValue(value: string | null, maxLength = 100) {
  return String(value ?? "").trim().slice(0, maxLength);
 }

-function findAsset(value: string, imageIndex: number) {
+function findAsset(
+  value: string,
+  imageIndex: number,
+  prefixes = ["content-assets/", "task-assets/"],
+) {
  try {
    const assets = JSON.parse(value) as StoredAsset[];
    return Array.isArray(assets)
@@ -34,8 +39,7 @@ function findAsset(value: string, imageIndex: number) {
          (asset) =>
            asset.index === imageIndex &&
            typeof asset.key === "string" &&
-            (asset.key.startsWith("content-assets/") ||
-              asset.key.startsWith("task-assets/")) &&
+            prefixes.some((prefix) => asset.key.startsWith(prefix)) &&
            (asset.fileToken === undefined ||
              typeof asset.fileToken === "string"),
        )
@@ -55,18 +59,20 @@ async function handleGet(request: Request) {
    const distributionId = textValue(url.searchParams.get("distribution"));
    const imageIndex = Number(url.searchParams.get("index"));
    const imageKind = textValue(url.searchParams.get("kind"), 20);
+    const downloadRequested = url.searchParams.get("download") === "1";
    if (
      (!delegationToken && (!taskToken || !claimToken)) ||
      !distributionId ||
      !Number.isInteger(imageIndex) ||
      imageIndex < 1
    ) {
-      return Response.json({ error: "图片链接不完整" }, { status: 400 });
+      return Response.json({ error: "素材链接不完整" }, { status: 400 });
    }
    const row = delegationToken
      ? await getRawDb()
          .prepare(
            `SELECT c.image_assets,
+                    c.video_assets,
                    d.result_screenshot_key,
                    d.publish_screenshot_key,
                    d.screenshot_key
@@ -80,6 +86,7 @@ async function handleGet(request: Request) {
          .bind(distributionId, delegationToken)
          .first<{
            image_assets: string;
+            video_assets: string;
            result_screenshot_key: string | null;
            publish_screenshot_key: string | null;
            screenshot_key: string | null;
@@ -87,6 +94,7 @@ async function handleGet(request: Request) {
      : await getRawDb()
          .prepare(
            `SELECT c.image_assets,
+                    c.video_assets,
                    d.result_screenshot_key,
                    d.publish_screenshot_key,
                    d.screenshot_key
@@ -102,6 +110,7 @@ async function handleGet(request: Request) {
          .bind(distributionId, claimToken, taskToken)
          .first<{
            image_assets: string;
+            video_assets: string;
            result_screenshot_key: string | null;
            publish_screenshot_key: string | null;
            screenshot_key: string | null;
@@ -124,36 +133,82 @@ async function handleGet(request: Request) {
          ? row?.screenshot_key?.startsWith("creator-center/")
            ? { index: 1, key: row.screenshot_key }
            : undefined
+        : imageKind === "video"
+          ? row
+            ? findAsset(row.video_assets, imageIndex, ["content-videos/"])
+            : undefined
        : row
          ? findAsset(row.image_assets, imageIndex)
          : undefined;
    if (!asset?.key) {
-      return Response.json({ error: "没有找到这张图片" }, { status: 404 });
+      return Response.json({ error: "没有找到这个素材" }, { status: 404 });
    }
    const bucket = getUploadBucket();
    let object = await bucket.get(asset.key);
-    if (!object && asset.fileToken) {
+    let objectBytes = object ? await object.arrayBuffer() : null;
+    const invalidStoredVideo =
+      imageKind === "video" &&
+      objectBytes !== null &&
+      !hasMp4FileSignature(objectBytes);
+    if ((!object || invalidStoredVideo) && asset.fileToken) {
      const media = await downloadFeishuMedia(
        asset.fileToken,
        env as unknown as FeishuBindings,
+        fetch,
+        {
+          maxBytes: imageKind === "video" ? 500 * 1024 * 1024 : undefined,
+          label: imageKind === "video" ? "视频" : "图片",
+        },
      );
+      if (imageKind === "video" && !hasMp4FileSignature(media.bytes)) {
+        return Response.json(
+          { error: "源视频不是可下载的 MP4 文件，请重新上传 MP4 视频" },
+          { status: 422 },
+        );
+      }
      await bucket.put(asset.key, media.bytes, {
-        httpMetadata: { contentType: media.contentType },
+        httpMetadata: {
+          contentType: imageKind === "video" ? "video/mp4" : media.contentType,
+        },
        customMetadata: { source: "feishu-api" },
      });
      object = await bucket.get(asset.key);
+      objectBytes = object ? await object.arrayBuffer() : media.bytes;
    }
-    if (!object) {
-      return Response.json({ error: "图片同步失败，请稍后重试" }, { status: 404 });
+    if (!object || !objectBytes) {
+      return Response.json({ error: "素材同步失败，请稍后重试" }, { status: 404 });
+    }
+    if (imageKind === "video" && !hasMp4FileSignature(objectBytes)) {
+      return Response.json(
+        { error: "已存储的视频不是有效的 MP4 文件，请联系运营重新同步" },
+        { status: 422 },
+      );
    }
    const headers = new Headers();
    object.writeHttpMetadata(headers);
-    headers.set("Cache-Control", "private, max-age=3600");
-    headers.set("Content-Disposition", `inline; filename="note-${imageIndex}"`);
-    return new Response(await object.arrayBuffer(), { headers });
+    const isMutableEvidence =
+      imageKind === "publish" || imageKind === "creator";
+    headers.set(
+      "Cache-Control",
+      isMutableEvidence ? "private, no-store" : "private, max-age=3600",
+    );
+    if (imageKind === "video") {
+      headers.set("Content-Type", "video/mp4");
+    }
+    headers.set("Content-Length", String(objectBytes.byteLength));
+    headers.set("X-Content-Type-Options", "nosniff");
+    const fileName =
+      imageKind === "video"
+        ? `video-${imageIndex}.mp4`
+        : `image-${imageIndex}`;
+    headers.set(
+      "Content-Disposition",
+      `${downloadRequested ? "attachment" : "inline"}; filename="${fileName}"; filename*=UTF-8''${encodeURIComponent(fileName)}`,
+    );
+    return new Response(objectBytes, { headers });
  } catch (error) {
    return Response.json(
-      { error: error instanceof Error ? error.message : "图片读取失败" },
+      { error: error instanceof Error ? error.message : "素材读取失败" },
      { status: 500 },
    );
  }
--- a/app/api/partner/route.ts
+++ b/app/api/partner/route.ts
@@ -4,7 +4,10 @@ import type { DatabaseStatement } from "../../../lib/database";

 const env = getRuntimeEnv();
 import { enrichDistributionAccount } from "../../../lib/account-enrichment-service";
-import { createCollectionRunTasks } from "../../../lib/collection-service";
+import {
+  createCollectionRunTasks,
+  runDueScheduledCollections,
+} from "../../../lib/collection-service";
 import {
  resolveCollectionMcpConfig,
  type CollectionMcpBindings,
@@ -17,8 +20,8 @@ import {
 } from "../../../lib/mvp-db";
 import {
  accountFromPublishLink,
-  extractXhsPublishUrl,
 } from "../../../lib/partner-utils";
+import { extractPublishUrl } from "../../../lib/publish-url";
 import { parseClaimantIdentifier } from "../../../lib/claimant-identifier";
 import {
  partnerOptions,
@@ -101,23 +104,28 @@ function publicImageAssets(value: unknown): ImageAsset[] {
  }
 }

+type PartnerTask = {
+  id: string;
+  name: string;
+  brand: string;
+  quantity: number;
+  claimed_quantity: number;
+  due_at: string;
+  status: string;
+  task_type: string;
+  platform: string;
+  content_format: string;
+};
+
 async function findTask(taskToken: string) {
  return getRawDb()
    .prepare(
-      `SELECT id, name, brand, quantity, claimed_quantity, due_at, status, task_type
+      `SELECT id, name, brand, quantity, claimed_quantity, due_at, status,
+              task_type, platform, content_format
       FROM tasks WHERE share_token = ?`,
    )
    .bind(taskToken)
-    .first<{
-      id: string;
-      name: string;
-      brand: string;
-      quantity: number;
-      claimed_quantity: number;
-      due_at: string;
-      status: string;
-      task_type: string;
-    }>();
+    .first<PartnerTask>();
 }

 async function findDelegationAccess(delegationToken: string) {
@@ -132,6 +140,8 @@ async function findDelegationAccess(delegationToken: string) {
        t.due_at,
        t.status,
        t.task_type,
+        t.platform,
+        t.content_format,
        b.id AS bundle_id,
        b.label AS bundle_label,
        b.quantity AS bundle_quantity,
@@ -141,15 +151,7 @@ async function findDelegationAccess(delegationToken: string) {
       WHERE b.share_token = ? AND b.status = 'active'`,
    )
    .bind(delegationToken)
-    .first<{
-      id: string;
-      name: string;
-      brand: string;
-      quantity: number;
-      claimed_quantity: number;
-      due_at: string;
-      status: string;
-      task_type: string;
+    .first<PartnerTask & {
      bundle_id: string;
      bundle_label: string;
      bundle_quantity: number;
@@ -172,13 +174,15 @@ async function findAccessibleAssignment(
      d.publish_screenshot_key,
      d.screenshot_key,
      d.result_screenshot_key,
-      d.result_submitted_at`;
+      d.result_submitted_at,
+      c.claimant_name`;
  if (delegationToken) {
    return db
      .prepare(
        `${select}
         FROM distributions d
         JOIN delegation_bundles b ON b.id = d.delegation_bundle_id
+         JOIN claims c ON c.id = d.claim_id
         WHERE d.id = ?
           AND b.share_token = ?
           AND b.task_id = ?
@@ -194,6 +198,7 @@ async function findAccessibleAssignment(
        screenshot_key: string | null;
        result_screenshot_key: string | null;
        result_submitted_at: string | null;
+        claimant_name: string;
      }>();
  }
  if (!claimToken) return null;
@@ -214,6 +219,7 @@ async function findAccessibleAssignment(
      screenshot_key: string | null;
      result_screenshot_key: string | null;
      result_submitted_at: string | null;
+      claimant_name: string;
    }>();
 }

@@ -284,6 +290,7 @@ async function handleGet(request: Request) {
            c.body,
            c.source_row,
            c.image_assets,
+            c.video_assets,
            a.nickname AS account_nickname,
            b.id AS delegation_bundle_id,
            b.label AS delegation_label
@@ -326,7 +333,9 @@ async function handleGet(request: Request) {
        assignments: assignments.results.map((assignment) => ({
          ...assignment,
          images: publicImageAssets(assignment.image_assets),
+          videos: publicImageAssets(assignment.video_assets),
          image_assets: undefined,
+          video_assets: undefined,
        })),
        delegations: delegations.results,
      };
@@ -349,6 +358,7 @@ async function handleGet(request: Request) {
            c.body,
            c.source_row,
            c.image_assets,
+            c.video_assets,
            a.nickname AS account_nickname
           FROM distributions d
           JOIN contents c ON c.id = d.content_id
@@ -366,7 +376,9 @@ async function handleGet(request: Request) {
        assignments: assignments.results.map((assignment) => ({
          ...assignment,
          images: publicImageAssets(assignment.image_assets),
+          videos: publicImageAssets(assignment.video_assets),
          image_assets: undefined,
+          video_assets: undefined,
        })),
      };
    }
@@ -379,6 +391,8 @@ async function handleGet(request: Request) {
            dueAt: task.due_at,
            status: task.status,
            type: task.task_type,
+            platform: task.platform,
+            contentFormat: task.content_format,
          }
        : {
            name: task.name,
@@ -388,6 +402,8 @@ async function handleGet(request: Request) {
            dueAt: task.due_at,
            status: task.status,
            type: task.task_type,
+            platform: task.platform,
+            contentFormat: task.content_format,
            availableQuantity: available?.count ?? 0,
          },
      claim,
@@ -798,14 +814,15 @@ async function handlePost(request: Request) {
          { status: 400 },
        );
      }
-      const publishUrl = extractXhsPublishUrl(publishInput);
+      const platform = task.platform === "抖音" ? "抖音" : "小红书";
+      const publishUrl = extractPublishUrl(publishInput, platform);
      if (!publishUrl) {
        return Response.json(
-          { error: "请粘贴包含小红书长链或短链的分享内容" },
+          { error: `请粘贴包含${platform}作品链接的分享内容` },
          { status: 400 },
        );
      }
-      const account = accountFromPublishLink(publishUrl);
+      const account = accountFromPublishLink(publishUrl, platform);
      if (!account) {
        return Response.json({ error: "发布链接格式不正确" }, { status: 400 });
      }
@@ -816,28 +833,69 @@ async function handlePost(request: Request) {
        delegationToken,
      );
      if (!assignment) {
-        return Response.json({ error: "笔记与领取凭证不匹配" }, { status: 403 });
+        return Response.json({ error: "作品与领取凭证不匹配" }, { status: 403 });
      }
      if (!assignment.publish_screenshot_key) {
        return Response.json({ error: "请先上传发布截图" }, { status: 400 });
      }
      const reuseExistingAccount =
        assignment.publish_url === publishUrl && assignment.account_id;
+      const matchedAccount = reuseExistingAccount
+        ? null
+        : await db
+            .prepare(
+              `SELECT id FROM accounts
+               WHERE platform = ? AND platform_uid = ?
+               LIMIT 1`,
+            )
+            .bind(account.platform, account.platformUid)
+            .first<{ id: string }>();
      const accountId =
        reuseExistingAccount ||
+        matchedAccount?.id ||
        `account-${hashText(`${account.platform}:${account.platformUid}`)}`;
      const statements: DatabaseStatement[] = [];
+      const publishUrlChanged = Boolean(
+        assignment.publish_url && assignment.publish_url !== publishUrl,
+      );
+      if (publishUrlChanged) {
+        statements.push(
+          db
+            .prepare(
+              `UPDATE distributions SET
+                d2_likes = NULL, d2_comments = NULL, d2_collects = NULL,
+                d5_likes = NULL, d5_comments = NULL, d5_collects = NULL,
+                d7_likes = NULL, d7_comments = NULL, d7_collects = NULL,
+                latest_likes = NULL, latest_comments = NULL,
+                latest_collects = NULL, latest_shares = NULL,
+                collection_status = 'pending',
+                collection_status_description = '发布链接已更新，等待重新采集',
+                collection_updated_at = NULL, last_collection_day = NULL,
+                updated_at = CURRENT_TIMESTAMP
+               WHERE id = ?`,
+            )
+            .bind(assignment.id),
+          db
+            .prepare("DELETE FROM collection_runs WHERE distribution_id = ?")
+            .bind(assignment.id),
+        );
+      }
      if (!reuseExistingAccount) {
        statements.push(
          db
          .prepare(
            `INSERT INTO accounts
-              (id, platform, platform_uid, nickname, profile_url, post_count)
-             VALUES (?, ?, ?, ?, ?, 1)
+              (id, platform, platform_uid, nickname, profile_url,
+               current_contact, post_count)
+             VALUES (?, ?, ?, ?, ?, ?, 1)
             ON CONFLICT(platform, platform_uid) DO UPDATE SET
               nickname = excluded.nickname,
               profile_url = excluded.profile_url,
-               post_count = accounts.post_count + ?,
+               current_contact = CASE
+                 WHEN excluded.current_contact != ''
+                   THEN excluded.current_contact
+                 ELSE accounts.current_contact
+               END,
               last_seen_at = CURRENT_TIMESTAMP`,
          )
          .bind(
@@ -846,7 +904,7 @@ async function handlePost(request: Request) {
            account.platformUid,
            account.nickname,
            account.profileUrl,
-            assignment.publish_url ? 0 : 1,
+            assignment.claimant_name,
          ),
        );
      }
@@ -863,6 +921,26 @@ async function handlePost(request: Request) {
          )
          .bind(accountId, publishUrl, assignment.id),
      );
+      statements.push(
+        db
+          .prepare(
+            `UPDATE accounts SET post_count = (
+               SELECT COUNT(*) FROM distributions WHERE account_id = ?
+             ) WHERE id = ?`,
+          )
+          .bind(accountId, accountId),
+      );
+      if (assignment.account_id && assignment.account_id !== accountId) {
+        statements.push(
+          db
+            .prepare(
+              `UPDATE accounts SET post_count = (
+                 SELECT COUNT(*) FROM distributions WHERE account_id = ?
+               ) WHERE id = ?`,
+            )
+            .bind(assignment.account_id, assignment.account_id),
+        );
+      }
      if (!assignment.publish_url) {
        statements.push(
          db
@@ -896,26 +974,46 @@ async function handlePost(request: Request) {
          collectionDays = [];
        }
        if (collectionDays.length > 0) {
+          await db
+            .prepare(
+              `UPDATE distributions SET collection_status = 'scheduled',
+                 collection_status_description = ? WHERE id = ?`,
+            )
+            .bind(
+              `已安排${collectionDays.length}个采集日，每日09:00执行`,
+              assignment.id,
+            )
+            .run();
          await createCollectionRunTasks(
            db,
            task.id,
            collectionSchedule.collection_start_date,
            collectionDays,
          );
+          runInBackground(
+            runDueScheduledCollections(
+              db,
+              Date.now(),
+              resolveCollectionMcpConfig(
+                env as unknown as CollectionMcpBindings,
+              ),
+              "catchup",
+              task.id,
+            ).catch(() => undefined),
+            "collection catchup after partner publish update",
+          );
        }
      }
-      if (account.platform === "小红书") {
-        const enrichment = enrichDistributionAccount(
-          db,
-          assignment.id,
-          publishUrl,
-          account.nickname,
-          resolveCollectionMcpConfig(
-            env as unknown as CollectionMcpBindings,
-          ),
-        ).catch(() => undefined);
-        runInBackground(enrichment, "distribution account enrichment");
-      }
+      const enrichment = enrichDistributionAccount(
+        db,
+        assignment.id,
+        publishUrl,
+        account.nickname,
+        resolveCollectionMcpConfig(
+          env as unknown as CollectionMcpBindings,
+        ),
+      ).catch(() => undefined);
+      runInBackground(enrichment, "distribution account enrichment");
      return Response.json({ ok: true });
    }

--- a/app/api/recovery-export/route.ts
+++ b/app/api/recovery-export/route.ts
@@ -17,6 +17,7 @@ import {
  type RecoveryWorkbookRow,
 } from "../../../lib/recovery-workbook";
 import { consumeMcpExportToken } from "../../../lib/mcp-export-token";
+import { normalizeWorkbookImage } from "../../../lib/workbook-image";

 const env = getRuntimeEnv();

@@ -53,6 +54,7 @@ type ExportRow = {
  latest_likes: number | null;
  latest_comments: number | null;
  latest_collects: number | null;
+  latest_shares: number | null;
  collection_status: string | null;
  collection_status_description: string | null;
  collection_updated_at: string | null;
@@ -121,12 +123,16 @@ function latestMetrics(row: ExportRow) {
  const likes = row.latest_likes ?? legacyLikes;
  const comments = row.latest_comments ?? legacyComments;
  const collects = row.latest_collects ?? legacyCollects;
+  const shares = row.latest_shares;
  return {
    likes,
    comments,
    collects,
+    shares,
    total:
-      likes === null ? null : likes + (comments ?? 0) + (collects ?? 0),
+      likes === null
+        ? null
+        : likes + (comments ?? 0) + (collects ?? 0) + (shares ?? 0),
  };
 }

@@ -181,13 +187,13 @@ async function loadImage(reference: ImageReference) {
    object = await bucket.get(reference.key);
  }
  if (!object) return null;
-  return {
+  return normalizeWorkbookImage({
    bytes: new Uint8Array(await object.arrayBuffer()),
    contentType: contentTypeFromObject(object),
    width: reference.width,
    height: reference.height,
    description: reference.description,
-  } satisfies RecoveryWorkbookImage;
+  } satisfies RecoveryWorkbookImage);
 }

 async function loadImages(references: ImageReference[]) {
@@ -236,9 +242,9 @@ export async function GET(request: Request) {
    }
    const db = getRawDb();
    const task = await db
-      .prepare("SELECT id, name, brand FROM tasks WHERE id = ?")
+      .prepare("SELECT id, name, brand, platform FROM tasks WHERE id = ?")
      .bind(taskId)
-      .first<{ id: string; name: string; brand: string }>();
+      .first<{ id: string; name: string; brand: string; platform: string }>();
    if (!task) {
      return Response.json({ error: "没有找到这个任务" }, { status: 404 });
    }
@@ -269,6 +275,7 @@ export async function GET(request: Request) {
          d.latest_likes,
          d.latest_comments,
          d.latest_collects,
+          d.latest_shares,
          d.collection_status,
          d.collection_status_description,
          d.collection_updated_at,
@@ -318,18 +325,19 @@ export async function GET(request: Request) {
      }
    });
    const loadedImages = await loadImages(references);
+    const isDouyin = task.platform === "抖音";
+    const metricHeaders = isDouyin
+      ? ["点赞", "收藏", "转发", "评论", "总互动"]
+      : ["点赞", "收藏", "评论", "总互动"];
    const headers = [
      "序号（不能改）",
      "标题",
      "笔记内容（正文+话题）",
      ...Array.from({ length: maxContentImages }, (_, index) => `图片${index + 1}`),
-      "小红书昵称",
+      `${task.platform}昵称`,
      "发布链接",
      "发布时间",
-      "点赞",
-      "收藏",
-      "评论",
-      "总互动",
+      ...metricHeaders,
      "曝光量-实际（第7天）",
      "阅读量-实际（第7天）",
      "数据分析截图（单篇笔记数据分析截图）",
@@ -341,7 +349,7 @@ export async function GET(request: Request) {
    ];
    const originalImageStart = 3;
    const accountColumn = originalImageStart + maxContentImages;
-    const creatorScreenshotColumn = accountColumn + 9;
+    const creatorScreenshotColumn = accountColumn + (isDouyin ? 10 : 9);
    const publishScreenshotColumn = creatorScreenshotColumn + 1;
    const workbookRows: RecoveryWorkbookRow[] = rows.map((row, rowIndex) => {
      const metrics = latestMetrics(row);
@@ -350,7 +358,7 @@ export async function GET(request: Request) {
        { length: maxContentImages },
        (_, index) => {
          const asset = contentAssets.find((item) => item.index === index + 1);
-          return asset && loadedImages.get(asset.key) ? "见图" : asset ? "图片读取失败" : "";
+          return "";
        },
      );
      const creatorImage = row.screenshot_key
@@ -369,12 +377,13 @@ export async function GET(request: Request) {
        formatExportDate(row.publish_time),
        metrics.likes,
        metrics.collects,
+        ...(isDouyin ? [metrics.shares] : []),
        metrics.comments,
        metrics.total,
        row.exposure,
        row.views,
-        creatorImage ? "见图" : row.screenshot_key ? "截图读取失败" : "",
-        publishImage ? "见图" : row.publish_screenshot_key ? "截图读取失败" : "",
+        "",
+        "",
        formatExportDate(row.collection_updated_at || row.updated_at),
        row.distribution_id ? collectionLabel(row) : "未领取",
        row.partner_name || "",
@@ -403,6 +412,7 @@ export async function GET(request: Request) {
      20,
      11,
      11,
+      ...(isDouyin ? [11] : []),
      11,
      11,
      18,
--- a/app/api/resources-export/route.ts
+++ b/app/api/resources-export/route.ts
@@ -15,8 +15,12 @@ type AccountRow = {
  profile_url: string;
  ip_location: string;
  followers: number;
+  gender: string;
+  bio: string;
+  tags: string;
  post_count: number;
  cooperation_source: string;
+  current_contact: string;
  first_seen_at: string;
  last_seen_at: string;
 };
@@ -24,6 +28,7 @@ type AccountRow = {
 type CooperationRow = {
  account_id: string;
  partner_name: string;
+  claimant_name: string | null;
  delegation_bundle_id: string | null;
 };

@@ -74,9 +79,11 @@ async function exportAccounts(accountIds: string[]) {
          `SELECT
            d.account_id,
            p.name AS partner_name,
+            cl.claimant_name,
            d.delegation_bundle_id
           FROM distributions d
           JOIN partners p ON p.id = d.partner_id
+           LEFT JOIN claims cl ON cl.id = d.claim_id
           WHERE d.account_id IS NOT NULL`,
        )
        .all<CooperationRow>(),
@@ -109,8 +116,12 @@ async function exportAccounts(accountIds: string[]) {
      "账号主页",
      "IP地",
      "粉丝数",
+      "性别",
+      "简介",
+      "标签",
      "合作发布数",
      "历史合作来源",
+      "当前联系人",
      "资源归属",
      "首次合作时间",
      "最近合作时间",
@@ -119,7 +130,13 @@ async function exportAccounts(accountIds: string[]) {
      const cooperation = cooperationByAccount.get(account.id) ?? [];
      const sources = [
        ...new Set([
-          ...cooperation.map((item) => item.partner_name),
+          ...cooperation
+            .filter(
+              (item) =>
+                !item.claimant_name ||
+                item.partner_name !== item.claimant_name,
+            )
+            .map((item) => item.partner_name),
          ...(account.cooperation_source || "")
            .split(/[、,，;；|]/)
            .map((item) => item.trim())
@@ -138,8 +155,12 @@ async function exportAccounts(accountIds: string[]) {
          account.profile_url || "",
          account.ip_location || "待识别",
          account.followers,
+          account.gender || "",
+          account.bio || "",
+          account.tags || "",
          account.post_count,
          sources.join("、"),
+          account.current_contact || "",
          partnerManagedOnly ? "合作社资源 · 不可直联" : "可直联",
          formatExportDate(account.first_seen_at),
          formatExportDate(account.last_seen_at),
@@ -161,9 +182,13 @@ async function exportAccounts(accountIds: string[]) {
        44,
        14,
        14,
+        10,
+        36,
+        32,
        14,
        32,
        22,
+        22,
        21,
        21,
      ],
--- a/app/api/resources-import/route.ts
+++ b/app/api/resources-import/route.ts
@@ -1,214 +1,24 @@
 import { isManagerRequest, managerForbidden } from "../../../lib/user-auth";
-import { ensureSchema, getRawDb } from "../../../lib/mvp-db";
+import { runInBackground } from "../../../lib/background";
 import {
-  resolveCollectionMcpConfig,
-  resolveXhsProfileDetailsFromMcp,
-  resolveXhsPublicAccountDetails,
-  type CollectionMcpBindings,
-} from "../../../lib/mcp-collection-client";
-import { getRuntimeEnv } from "../../../lib/runtime-env";
-import {
-  mergeCooperationSources,
-  normalizeProfileUrl,
  parseResourceImportFile,
  RESOURCE_IMPORT_MAX_BYTES,
-  resourcePlatformUid,
+  RESOURCE_IMPORT_MAX_ROWS,
  resourceImportMissingFields,
  type ResourceImportRow,
 } from "../../../lib/resource-import";
+import {
+  analyzeRows,
+  enrichRows,
+  loadAccounts,
+  mergeExistingFields,
+  writeAnalyzedRows,
+  RESOURCE_IMPORT_SYNC_ENRICH_ROWS,
+  type AnalyzedRow,
+  type AccountRow,
+} from "../../../lib/resource-write";

-type AccountRow = {
-  id: string;
-  platform: string;
-  platform_uid: string;
-  public_account_id: string;
-  nickname: string;
-  profile_url: string;
-  ip_location: string;
-  followers: number;
-  cooperation_source: string;
-};
-
-type AnalyzedRow = ResourceImportRow & {
-  action: "create" | "update" | "error";
-  accountId: string;
-  platformUid: string;
-  cooperationSource: string;
-};
-
-function identityKey(platform: string, value: string) {
-  return `${platform.trim().toLocaleLowerCase("zh-CN")}|${value
-    .trim()
-    .toLocaleLowerCase("zh-CN")}`;
-}
-
-async function loadAccounts() {
-  return getRawDb()
-    .prepare(
-      `SELECT id, platform, platform_uid, public_account_id, nickname,
-              profile_url, ip_location, followers, cooperation_source
-       FROM accounts`,
-    )
-    .all<AccountRow>();
-}
-
-async function mapConcurrent<T, R>(
-  items: T[],
-  limit: number,
-  worker: (item: T) => Promise<R>,
-) {
-  const results = new Array<R>(items.length);
-  let cursor = 0;
-  await Promise.all(
-    Array.from({ length: Math.min(limit, items.length) }, async () => {
-      while (cursor < items.length) {
-        const index = cursor;
-        cursor += 1;
-        results[index] = await worker(items[index]);
-      }
-    }),
-  );
-  return results;
-}
-
-async function enrichRows(rows: ResourceImportRow[], accounts: AccountRow[]) {
-  const existingByProfile = new Map<string, AccountRow>();
-  for (const account of accounts) {
-    const profileUrl = normalizeProfileUrl(account.profile_url || "");
-    if (profileUrl) {
-      existingByProfile.set(identityKey(account.platform, profileUrl), account);
-    }
-  }
-  const mcpConfig = resolveCollectionMcpConfig(
-    getRuntimeEnv() as unknown as CollectionMcpBindings,
-  );
-  return mapConcurrent(rows, 4, async (row) => {
-    if (row.errors.length > 0 || !row.profileUrl || row.platform !== "小红书") {
-      return row;
-    }
-    const existing = existingByProfile.get(identityKey(row.platform, row.profileUrl));
-    const existingIpLocation =
-      existing?.ip_location && existing.ip_location !== "待识别"
-        ? existing.ip_location
-        : "";
-    const baseline: ResourceImportRow = {
-      ...row,
-      nickname: row.nickname || existing?.nickname || "",
-      publicAccountId: row.publicAccountId || existing?.public_account_id || "",
-      ipLocation: row.ipLocation || existingIpLocation,
-      followers: row.followersResolved
-        ? row.followers
-        : Number(existing?.followers || 0),
-      followersResolved:
-        row.followersResolved || Number(existing?.followers || 0) > 0,
-    };
-    if (resourceImportMissingFields(baseline).length === 0) {
-      return baseline;
-    }
-
-    let details: {
-      nickname: string | null;
-      redId: string | null;
-      followers: number | null;
-      ipLocation: string | null;
-    } = await resolveXhsProfileDetailsFromMcp(row.profileUrl, mcpConfig).catch(
-      () => ({ nickname: null, redId: null, followers: null, ipLocation: null }),
-    );
-    const mcpResult = {
-      nickname: baseline.nickname || details.nickname?.trim() || "",
-      publicAccountId: baseline.publicAccountId || details.redId?.trim() || "",
-      ipLocation: baseline.ipLocation || details.ipLocation?.trim() || "",
-      followersResolved: baseline.followersResolved || details.followers !== null,
-    };
-    if (resourceImportMissingFields(mcpResult).length > 0) {
-      const publicDetails = await resolveXhsPublicAccountDetails(row.profileUrl).catch(
-        () => ({ nickname: null, redId: null, followers: null, ipLocation: null }),
-      );
-      details = {
-        nickname: details.nickname || publicDetails.nickname,
-        redId: details.redId || publicDetails.redId,
-        followers: details.followers ?? publicDetails.followers,
-        ipLocation: details.ipLocation || publicDetails.ipLocation,
-      };
-    }
-    return {
-      ...baseline,
-      nickname: baseline.nickname || details.nickname?.trim() || "待识别账号",
-      publicAccountId: baseline.publicAccountId || details.redId?.trim() || "",
-      ipLocation: baseline.ipLocation || details.ipLocation?.trim() || "待识别",
-      followers: baseline.followersResolved
-        ? baseline.followers
-        : (details.followers ?? 0),
-      followersResolved:
-        baseline.followersResolved || details.followers !== null,
-    };
-  });
-}
-
-function analyzeRows(rows: ResourceImportRow[], accountRows: AccountRow[]) {
-  const profileMap = new Map<string, AccountRow>();
-  const publicIdMap = new Map<string, AccountRow>();
-  const platformUidMap = new Map<string, AccountRow>();
-  for (const account of accountRows) {
-    const profileUrl = normalizeProfileUrl(account.profile_url || "");
-    if (profileUrl) profileMap.set(identityKey(account.platform, profileUrl), account);
-    if (account.public_account_id) {
-      publicIdMap.set(identityKey(account.platform, account.public_account_id), account);
-    }
-    platformUidMap.set(identityKey(account.platform, account.platform_uid), account);
-  }
-
-  return rows.map<AnalyzedRow>((row) => {
-    const platformUid = resourcePlatformUid(row);
-    const profileMatch = row.profileUrl
-      ? profileMap.get(identityKey(row.platform, row.profileUrl))
-      : undefined;
-    const publicIdMatch = row.publicAccountId
-      ? publicIdMap.get(identityKey(row.platform, row.publicAccountId))
-      : undefined;
-    const uidMatch = platformUidMap.get(identityKey(row.platform, platformUid));
-    const matches = [profileMatch, publicIdMatch, uidMatch].filter(
-      (account): account is AccountRow => Boolean(account),
-    );
-    const matchedIds = [...new Set(matches.map((account) => account.id))];
-    const errors = [...row.errors];
-    if (matchedIds.length > 1) {
-      errors.push("账号主页和账号ID匹配到不同的现有账号，请先核对");
-    }
-    const existing = matchedIds.length === 1 ? matches[0] : undefined;
-    const accountId = existing?.id ?? `account-${crypto.randomUUID().slice(0, 12)}`;
-    const analyzed: AnalyzedRow = {
-      ...row,
-      errors,
-      action: errors.length > 0 ? "error" : existing ? "update" : "create",
-      accountId,
-      platformUid: existing?.platform_uid ?? platformUid,
-      cooperationSource: mergeCooperationSources(
-        existing?.cooperation_source ?? "",
-        row.cooperationSource,
-      ),
-    };
-    if (analyzed.action !== "error") {
-      const virtual: AccountRow = {
-        id: accountId,
-        platform: row.platform,
-        platform_uid: analyzed.platformUid,
-        public_account_id: row.publicAccountId || existing?.public_account_id || "",
-        nickname: row.nickname,
-        profile_url: row.profileUrl || existing?.profile_url || "",
-        ip_location: row.ipLocation || existing?.ip_location || "待识别",
-        followers: row.followers || existing?.followers || 0,
-        cooperation_source: analyzed.cooperationSource,
-      };
-      if (row.profileUrl) profileMap.set(identityKey(row.platform, row.profileUrl), virtual);
-      if (row.publicAccountId) {
-        publicIdMap.set(identityKey(row.platform, row.publicAccountId), virtual);
-      }
-      platformUidMap.set(identityKey(row.platform, analyzed.platformUid), virtual);
-    }
-    return analyzed;
-  });
-}
+const RESOURCE_IMPORT_PREVIEW_ROWS = 100;

 function summarize(rows: AnalyzedRow[]) {
  return {
@@ -219,10 +29,51 @@ function summarize(rows: AnalyzedRow[]) {
  };
 }

+function previewAnalyzedRows(rows: AnalyzedRow[]) {
+  const errorRows = rows.filter((row) => row.action === "error");
+  if (errorRows.length === 0) {
+    return rows.slice(0, RESOURCE_IMPORT_PREVIEW_ROWS);
+  }
+  const importableRows = rows.filter((row) => row.action !== "error");
+  return [
+    ...errorRows.slice(0, RESOURCE_IMPORT_PREVIEW_ROWS),
+    ...importableRows.slice(
+      0,
+      Math.max(0, RESOURCE_IMPORT_PREVIEW_ROWS - errorRows.length),
+    ),
+  ];
+}
+
+function deferredEnrichmentCount(rows: ResourceImportRow[]) {
+  return rows.filter(
+    (row) =>
+      row.errors.length === 0 &&
+      row.profileUrl &&
+      resourceImportMissingFields(row).length > 0,
+  ).length;
+}
+
+async function enrichImportedRowsInBackground(rows: ResourceImportRow[]) {
+  const accounts = await loadAccounts();
+  const baselineRows = mergeExistingFields(rows, accounts.results);
+  const missingRows = baselineRows.filter(
+    (row) =>
+      row.errors.length === 0 &&
+      row.profileUrl &&
+      resourceImportMissingFields(row).length > 0,
+  );
+  if (missingRows.length === 0) return;
+  const enriched = await enrichRows(missingRows, accounts.results);
+  const latestAccounts = await loadAccounts();
+  const analyzed = analyzeRows(enriched, latestAccounts.results).filter(
+    (row) => row.action !== "error",
+  );
+  await writeAnalyzedRows(analyzed);
+}
+
 export async function POST(request: Request) {
  if (!(await isManagerRequest(request))) return managerForbidden();
  try {
-    await ensureSchema();
    const form = await request.formData();
    const file = form.get("file");
    const mode = String(form.get("mode") ?? "preview");
@@ -230,17 +81,31 @@ export async function POST(request: Request) {
      return Response.json({ error: "请选择需要导入的 Excel 文件" }, { status: 400 });
    }
    if (file.size <= 0 || file.size > RESOURCE_IMPORT_MAX_BYTES) {
-      return Response.json({ error: "文件不能为空，且不能超过 5MB" }, { status: 400 });
+      return Response.json({ error: "文件不能为空，且不能超过 20MB" }, { status: 400 });
    }
    const rows = parseResourceImportFile(file.name, new Uint8Array(await file.arrayBuffer()));
    const accounts = await loadAccounts();
-    const enriched = await enrichRows(rows, accounts.results);
-    const analyzed = analyzeRows(enriched, accounts.results);
+    const accountRows: AccountRow[] = accounts.results;
+    const shouldEnrichSynchronously = rows.length <= RESOURCE_IMPORT_SYNC_ENRICH_ROWS;
+    const preparedRows = shouldEnrichSynchronously
+      ? await enrichRows(rows, accountRows)
+      : mergeExistingFields(rows, accountRows);
+    const analyzed = analyzeRows(preparedRows, accountRows);
    const summary = summarize(analyzed);
+    const importableRows = analyzed.filter((row) => row.action !== "error");
+    const importableRowNumbers = new Set(
+      importableRows.map((row) => row.rowNumber),
+    );
+    const importablePreparedRows = preparedRows.filter((row) =>
+      importableRowNumbers.has(row.rowNumber),
+    );
+    const deferredEnrichment = shouldEnrichSynchronously
+      ? 0
+      : deferredEnrichmentCount(importablePreparedRows);
    if (mode !== "commit") {
      return Response.json({
        summary,
-        rows: analyzed.slice(0, 100).map((row) => ({
+        rows: previewAnalyzedRows(analyzed).map((row) => ({
          rowNumber: row.rowNumber,
          platform: row.platform,
          nickname: row.nickname,
@@ -248,74 +113,44 @@ export async function POST(request: Request) {
          profileUrl: row.profileUrl,
          ipLocation: row.ipLocation,
          followers: row.followers,
+          gender: row.gender,
+          bio: row.bio,
+          tags: row.tags,
          cooperationSource: row.cooperationSource,
          action: row.action,
          errors: row.errors,
        })),
-        truncated: analyzed.length > 100,
+        truncated: analyzed.length > RESOURCE_IMPORT_PREVIEW_ROWS,
+        deferredEnrichment,
+        maxRows: RESOURCE_IMPORT_MAX_ROWS,
      });
    }
-    if (summary.error > 0) {
+    if (importableRows.length === 0) {
      return Response.json(
-        { error: `有 ${summary.error} 行数据未通过校验，请修正后重新上传`, summary },
+        { error: "没有可导入的有效账号，请修正异常数据后重新上传", summary },
        { status: 400 },
      );
    }
-
-    const db = getRawDb();
-    const statements = analyzed.map((row) =>
-      row.action === "update"
-        ? db
-            .prepare(
-              `UPDATE accounts SET
-                 nickname = ?,
-                 public_account_id = CASE WHEN ? != '' THEN ? ELSE public_account_id END,
-                 profile_url = CASE WHEN ? != '' THEN ? ELSE profile_url END,
-                 ip_location = CASE
-                   WHEN ? != '' AND ? != '待识别' THEN ? ELSE ip_location END,
-                 followers = CASE WHEN ? = 1 THEN ? ELSE followers END,
-                 cooperation_source = ?,
-                 last_seen_at = CURRENT_TIMESTAMP
-               WHERE id = ?`,
-            )
-            .bind(
-              row.nickname,
-              row.publicAccountId,
-              row.publicAccountId,
-              row.profileUrl,
-              row.profileUrl,
-              row.ipLocation,
-              row.ipLocation,
-              row.ipLocation,
-              row.followersResolved ? 1 : 0,
-              row.followers,
-              row.cooperationSource,
-              row.accountId,
-            )
-        : db
-            .prepare(
-              `INSERT INTO accounts
-                (id, platform, platform_uid, public_account_id, nickname,
-                 profile_url, ip_location, followers, post_count, avg_views,
-                 cooperation_source)
-               VALUES (?, ?, ?, ?, ?, ?, ?, ?, 0, 0, ?)`,
-            )
-            .bind(
-              row.accountId,
-              row.platform,
-              row.platformUid,
-              row.publicAccountId,
-              row.nickname,
-              row.profileUrl,
-              row.ipLocation,
-              row.followers,
-              row.cooperationSource,
-            ),
-    );
-    if (statements.length > 0) await db.batch(statements);
+    await writeAnalyzedRows(importableRows);
+    if (deferredEnrichment > 0) {
+      const importableSourceRows = rows.filter((row) =>
+        importableRowNumbers.has(row.rowNumber),
+      );
+      runInBackground(
+        enrichImportedRowsInBackground(importableSourceRows),
+        "bulk resource profile enrichment",
+      );
+    }
    return Response.json({
      summary,
-      message: `已导入 ${summary.create} 个新账号，更新 ${summary.update} 个已有账号`,
+      deferredEnrichment,
+      message: `已导入 ${summary.create} 个新账号，更新 ${summary.update} 个已有账号${
+        summary.error > 0 ? `；跳过 ${summary.error} 条异常数据` : ""
+      }${
+        deferredEnrichment > 0
+          ? `；${deferredEnrichment} 个账号的缺失公开资料将在后台补全`
+          : ""
+      }`,
    });
  } catch (error) {
    const message = error instanceof Error ? error.message : "导入失败";
--- a/app/api/resources-insert/route.ts
+++ b/app/api/resources-insert/route.ts
@@ -0,0 +1,64 @@
+import { isManagerRequest, managerForbidden } from "../../../lib/user-auth";
+import { buildManualRow, type ManualResourceInput } from "../../../lib/resource-import";
+import {
+  analyzeRows,
+  enrichRows,
+  loadAccounts,
+  writeAnalyzedRows,
+} from "../../../lib/resource-write";
+
+type InsertResult = {
+  action: "create" | "update";
+  message: string;
+};
+
+export async function POST(request: Request) {
+  if (!(await isManagerRequest(request))) return managerForbidden();
+  try {
+    const payload = (await request.json()) as Partial<ManualResourceInput> & {
+      mode?: string;
+    };
+    const row = buildManualRow({
+      profileUrl: payload.profileUrl ?? "",
+      nickname: payload.nickname ?? "",
+      publicAccountId: payload.publicAccountId ?? "",
+      ipLocation: payload.ipLocation ?? "",
+      followers: payload.followers ?? "",
+      gender: payload.gender ?? "",
+      bio: payload.bio ?? "",
+      tags: payload.tags ?? "",
+      cooperationSource: payload.cooperationSource ?? "",
+    });
+    if (row.errors.length > 0) {
+      return Response.json(
+        { error: row.errors[0], errors: row.errors },
+        { status: 400 },
+      );
+    }
+
+    const accounts = await loadAccounts();
+    const enriched = await enrichRows([row], accounts.results);
+    const analyzed = analyzeRows(enriched, accounts.results).filter(
+      (item) => item.action !== "error",
+    );
+    if (analyzed.length === 0) {
+      return Response.json(
+        { error: "账号数据校验未通过，无法保存" },
+        { status: 400 },
+      );
+    }
+    await writeAnalyzedRows(analyzed);
+    const result = analyzed[0];
+    const outcome: InsertResult = {
+      action: result.action === "update" ? "update" : "create",
+      message:
+        result.action === "update"
+          ? `已更新账号 ${row.nickname || row.publicAccountId || "资料"}`
+          : `已新增账号 ${row.nickname || row.publicAccountId || "资料"}`,
+    };
+    return Response.json(outcome);
+  } catch (error) {
+    const message = error instanceof Error ? error.message : "保存失败";
+    return Response.json({ error: message }, { status: 400 });
+  }
+}
--- a/app/globals.css
+++ b/app/globals.css
--- a/db/schema.ts
+++ b/db/schema.ts
@@ -21,6 +21,7 @@ export const partners = mysqlTable("partners", {
  owner: varchar("owner", { length: 255 }).notNull().default("运营组"),
  claimedTotal: int("claimed_total").notNull().default(0),
  completedTotal: int("completed_total").notNull().default(0),
+  wecomExternalUserId: varchar("wecom_external_user_id", { length: 128 }),
  createdAt: timestamp("created_at"),
 });

@@ -37,6 +38,10 @@ export const tasks = mysqlTable(
    taskType: varchar("task_type", { length: 32 })
      .notNull()
      .default("content_publish"),
+    platform: varchar("platform", { length: 32 }).notNull().default("小红书"),
+    contentFormat: varchar("content_format", { length: 32 })
+      .notNull()
+      .default("image_text"),
    sourceUrl: text("source_url").notNull(),
    sourceSheetId: varchar("source_sheet_id", { length: 255 }).notNull().default(""),
    sourceSheetName: varchar("source_sheet_name", { length: 255 }).notNull().default(""),
@@ -59,6 +64,7 @@ export const contents = mysqlTable("contents", {
  title: text("title").notNull(),
  body: text("body").notNull(),
  imageAssets: text("image_assets").notNull(),
+  videoAssets: text("video_assets").notNull(),
  status: varchar("status", { length: 32 }).notNull().default("available"),
  source: varchar("source", { length: 255 }).notNull().default("飞书内容表"),
  sourceRow: int("source_row"),
@@ -76,11 +82,17 @@ export const accounts = mysqlTable(
    profileUrl: text("profile_url").notNull(),
    ipLocation: varchar("ip_location", { length: 255 }).notNull().default("待识别"),
    followers: int("followers").notNull().default(0),
+    gender: varchar("gender", { length: 16 }).notNull().default(""),
+    bio: text("bio").notNull().default(""),
+    tags: varchar("tags", { length: 500 }).notNull().default(""),
    postCount: int("post_count").notNull().default(0),
    avgViews: int("avg_views").notNull().default(0),
    cooperationSource: varchar("cooperation_source", { length: 500 })
      .notNull()
      .default(""),
+    currentContact: varchar("current_contact", { length: 255 })
+      .notNull()
+      .default(""),
    firstSeenAt: timestamp("first_seen_at"),
    lastSeenAt: timestamp("last_seen_at"),
  },
@@ -162,6 +174,7 @@ export const distributions = mysqlTable("distributions", {
  latestLikes: int("latest_likes"),
  latestComments: int("latest_comments"),
  latestCollects: int("latest_collects"),
+  latestShares: int("latest_shares"),
  collectionStatus: varchar("collection_status", { length: 32 })
    .notNull()
    .default("pending"),
@@ -184,6 +197,7 @@ export const collectionRuns = mysqlTable(
    likes: int("likes"),
    comments: int("comments"),
    collects: int("collects"),
+    shares: int("shares"),
    statusDescription: text("status_description"),
    startedAt: datetime("started_at", { mode: "string", fsp: 3 }),
    completedAt: datetime("completed_at", { mode: "string", fsp: 3 }),
@@ -239,6 +253,35 @@ export const mcpExportTokens = mysqlTable(
  (table) => [index("mcp_export_tokens_expires_at_idx").on(table.expiresAt)],
 );

+export const wecomGroupChats = mysqlTable(
+  "wecom_group_chats",
+  {
+    chatId: varchar("chat_id", { length: 128 }).primaryKey(),
+    name: varchar("name", { length: 512 }).notNull().default(""),
+    ownerUserId: varchar("owner_user_id", { length: 128 }).notNull(),
+    memberCount: int("member_count").notNull().default(0),
+    status: int("status").notNull().default(0),
+    syncedAt: timestamp("synced_at"),
+  },
+  (table) => [index("wecom_group_chats_owner_idx").on(table.ownerUserId)],
+);
+
+export const wecomGroupPushes = mysqlTable(
+  "wecom_group_pushes",
+  {
+    id: varchar("id", { length: 64 }).primaryKey(),
+    taskId: varchar("task_id", { length: 64 }).notNull(),
+    msgid: varchar("msgid", { length: 128 }).notNull().default(""),
+    sender: varchar("sender", { length: 128 }).notNull(),
+    chatIds: text("chat_ids").notNull(),
+    textContent: text("text_content").notNull(),
+    linkUrl: text("link_url").notNull(),
+    failList: text("fail_list").notNull(),
+    createdAt: timestamp("created_at"),
+  },
+  (table) => [index("wecom_group_pushes_task_idx").on(table.taskId)],
+);
+
 export const backgroundJobs = mysqlTable(
  "background_jobs",
  {
--- a/deploy/nginx/koc-loop.conf
+++ b/deploy/nginx/koc-loop.conf
@@ -2,7 +2,8 @@ server {
  listen 80;
  server_name _;

-  client_max_body_size 10m;
+  # 批量回填 Excel 会内嵌多篇笔记原图和截图。
+  client_max_body_size 85m;

  location = /koc {
    return 301 /koc/$is_args$args;
@@ -21,7 +22,8 @@ server {
    proxy_buffering off;
    proxy_read_timeout 3600s;
    proxy_send_timeout 3600s;
-    proxy_set_header Host $host;
+    proxy_set_header Host $http_host;
+    proxy_set_header X-Forwarded-Host $http_host;
    proxy_set_header X-Real-IP $remote_addr;
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    proxy_set_header X-Forwarded-Proto $scheme;
@@ -30,7 +32,8 @@ server {
  location / {
    proxy_pass http://app:3000;
    proxy_http_version 1.1;
-    proxy_set_header Host $host;
+    proxy_set_header Host $http_host;
+    proxy_set_header X-Forwarded-Host $http_host;
    proxy_set_header X-Real-IP $remote_addr;
    proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
    proxy_set_header X-Forwarded-Proto $scheme;
--- a/design-qa-comparison.png
+++ b/design-qa-comparison.png
--- a/design-qa-current-filter.png
+++ b/design-qa-current-filter.png
--- a/design-qa-filter-comparison.png
+++ b/design-qa-filter-comparison.png
--- a/design-qa-implementation.png
+++ b/design-qa-implementation.png
--- a/design-qa-inline-filter-comparison.png
+++ b/design-qa-inline-filter-comparison.png
--- a/design-qa-inline-filter.png
+++ b/design-qa-inline-filter.png
--- a/design-qa-resource-card-alignment-comparison.png
+++ b/design-qa-resource-card-alignment-comparison.png
--- a/design-qa-resource-card-alignment-final.png
+++ b/design-qa-resource-card-alignment-final.png
--- a/design-qa-resource-card-comparison.png
+++ b/design-qa-resource-card-comparison.png
--- a/design-qa-resource-card-metrics-alignment-comparison.png
+++ b/design-qa-resource-card-metrics-alignment-comparison.png
--- a/design-qa-resource-card-metrics-alignment-final.png
+++ b/design-qa-resource-card-metrics-alignment-final.png
--- a/design-qa-resource-card-metrics-alignment-two-column.png
+++ b/design-qa-resource-card-metrics-alignment-two-column.png
--- a/design-qa-resource-cards-final.png
+++ b/design-qa-resource-cards-final.png
--- a/design-qa-resource-cards-v1.png
+++ b/design-qa-resource-cards-v1.png
--- a/design-qa-resource-cards-v2.png
+++ b/design-qa-resource-cards-v2.png
--- a/design-qa.md
+++ b/design-qa.md
@@ -45,3 +45,249 @@
 3. 重新构建后，浏览器识别到 1 条可点击标题；点击实际打开对应小红书页面，未回填行仍不可点击。

 final result: passed
+
+---
+
+# KOC LOOP 任务筛选常驻搜索框设计 QA
+
+## 验证对象
+
+- 用户目标截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-67b0b080-8ea8-4504-82eb-348e8cf4ec19.png`
+- 浏览器实现截图：`/private/tmp/koc-distributions-direct-search.png`
+- CSS 视口：842 × 778，设备像素比 2
+- 本地页面：`http://localhost:8080/?nav=distributions`
+
+## 调整结果
+
+1. 品牌/项目、内容类型、平台不再使用“先点下拉、再输入搜索”的两层结构，筛选栏直接展示三个可输入的搜索框。
+2. 输入内容时立即进行模糊筛选；聚焦或输入后，下方仅展示匹配候选项，不再重复显示第二个搜索框。
+3. 宽屏下四个搜索框保持同一行；当前窄视口下任务名称独占一行，三个分类搜索保持在下一行，不互相遮挡。
+4. 保留键盘与无障碍语义，三个分类搜索均使用 `combobox`，候选项使用 `listbox` / `option`。
+
+## 功能验证
+
+- 默认展示 6 个任务。
+- 品牌/项目直接输入“美团”后，候选项仅显示“美团”和“美团医美”，任务结果同步缩小为 2 个。
+- 清空输入后恢复 6 个任务。
+- 静态验收测试 14 项全部通过，正式构建通过；本地 Docker 应用已重新构建并加载新页面。
+
+final result: passed
+
+---
+
+# KOC LOOP 任务筛选下拉遮挡设计 QA
+
+## 验证对象
+
+- 用户问题截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-bd2d14b7-472f-43a5-b997-1b6ff9528ba1.png`
+- 浏览器实现截图：`/Users/wufp/Documents/koc-loop/design-qa-inline-filter.png`
+- 修复前后并排对照：`/Users/wufp/Documents/koc-loop/design-qa-inline-filter-comparison.png`
+- CSS 视口：1280 × 720，设备像素比 2
+- 本地页面：`http://localhost:8080/?nav=distributions`
+
+## 问题与调整
+
+1. 之前仍使用浏览器原生 `select`，系统级下拉层会悬浮在页面上，因此即使搜索框本身布局正常，展开菜单仍会遮住搜索区或任务卡片。
+2. 品牌/项目、内容类型、平台三个筛选器改为页面内选择面板，面板使用正常文档流布局，展开时把任务卡片向下推。
+3. 同一时间只展开一个筛选器；选择选项后自动收起，并保留键盘焦点、`aria-expanded`、`listbox` 和 `option` 语义。
+
+## 布局与功能验证
+
+- 内容类型面板展开后：面板 `position: static`，搜索框底部为 379px，面板顶部为 388px，两者无重叠。
+- 第一张任务卡片顶部为 480px，面板底部为 452px，任务卡片位于面板下方，未被覆盖。
+- 选择“视频”后面板自动收起，任务数由 6 个缩小为 1 个；清空筛选后恢复 6 个任务。
+- 浏览器实测搜索框、三个筛选按钮和任务卡片均可正常操作，交互过程中未出现页面报错。
+- TypeScript 构建、静态验收及共 65 项测试全部通过；本地 Docker 服务已重新构建并通过健康检查。
+
+final result: passed
+
+---
+
+# KOC LOOP 任务分发筛选栏设计 QA
+
+## 验证对象
+
+- 用户问题截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-4c32c7e3-58e0-4262-b625-c02b20ec4490.png`
+- 浏览器实现截图：`/Users/wufp/Documents/koc-loop/design-qa-implementation.png`
+- 修复前后并排对照：`/Users/wufp/Documents/koc-loop/design-qa-comparison.png`
+- CSS 视口：1280 × 720
+- 本地页面：`http://localhost:8080/?nav=distributions`
+
+## 问题与调整
+
+1. 全局 `label` 布局把搜索图标和输入框纵向堆叠，导致搜索框被拆成上下两层。
+2. 搜索框现在显式使用横向布局，并清除继承的外边距；图标使用固定宽度居中对齐。
+3. 品牌/项目下拉框加宽，选中值和下拉菜单锚点保持稳定，不再挤压搜索区域。
+
+## 功能验证
+
+- 输入“视频”后，结果从 6 个任务缩小为 1 个任务，筛选栏没有发生错位。
+- 清空搜索后恢复展示 6 个任务。
+- 浏览器控制台无错误。
+- TypeScript、静态验收测试和正式构建均通过。
+- 本地 Docker 服务已重建并通过健康检查。
+
+final result: passed
+
+---
+
+# KOC LOOP 可搜索任务筛选浮层设计 QA
+
+## 验证对象
+
+- 参考截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-fc7ce949-d8ea-4a3b-afa6-04a778fdb8dc.png`
+- 浏览器实现截图：`/Users/wufp/Documents/koc-loop/design-qa-current-filter.png`
+- 参考与实现并排对照：`/Users/wufp/Documents/koc-loop/design-qa-filter-comparison.png`
+- 本地页面：`http://localhost:8080/?nav=distributions`
+
+## 布局与交互验证
+
+- 品牌/项目、内容类型、平台均改为紧凑的页面内浮层，不再使用系统原生下拉，也不会把任务卡片整体向下推。
+- 浮层锚定在当前筛选按钮下方，宽度与筛选控件协调，未遮挡任务名称搜索框及其他筛选按钮。
+- 浮层顶部支持输入模糊查询；输入“医”后只显示“美团医美”，可继续点击完成筛选。
+- 同一时间只展开一个筛选器，支持点击页面其他位置或按 Esc 收起。
+- 浏览器实测筛选栏、任务卡片和按钮均保持稳定，没有布局跳动。
+
+## 数据修复验证
+
+- 抖音作品重新采集后获得点赞 5,684、收藏 565、转发 6,878、评论 332。
+- 抖音主页链接改用短链跳转携带的真实 `sec_uid`，不再用数字内部用户 ID 拼接主页。
+- MCP 的公开主页解析对当前账号仍返回资源不存在，因此公开抖音号、粉丝数和 IP 暂时保持待识别，不再写入错误数据。
+- 相关自动化测试共 67 项全部通过，本地 Docker 服务已重建并通过健康检查。
+
+final result: passed
+# KOC LOOP 任务分发横向搜索框设计 QA
+
+## 验证对象
+
+- 用户问题截图：`codex-clipboard-252bf483-c3d1-4d70-abfb-d54587a300a2.png`
+- 浏览器实现截图：`/private/tmp/koc-loop-distribution-filters-fixed.png`
+- 本地页面：`http://localhost:8080/?nav=distributions`
+
+## 问题与调整
+
+1. 品牌/项目、内容类型、平台三个筛选框使用通用表单标签，继承了纵向排列样式，导致图标和提示文字上下分离。
+2. 筛选框改为独立容器，并显式使用横向排列和垂直居中。
+3. 保留原有模糊搜索、筛选逻辑及整体视觉规范。
+
+## 布局与功能验证
+
+- 三个筛选框内图标、输入文字与容器的垂直中心误差均为 0px。
+- 四个搜索框保持同排展示，无相互遮挡、无异常换行、无额外浮层。
+- 项目构建通过；页面渲染测试 14/14 通过；本地 Docker 运行版本已更新。
+
+最终结果：通过。
+
+---
+
+# KOC LOOP KOC资源卡片密度优化设计 QA
+
+## 验证对象
+
+- 参考卡片：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-615400a6-0bd1-4110-b280-f7fdc29bf705.png`
+- 原页面参考：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-5bf2bfb5-a50e-4c08-a407-192fdcf20b58.png`
+- 第一轮实现截图：`/Users/wufp/Documents/koc-loop/design-qa-resource-cards-v2.png`
+- 最终浏览器截图：`/Users/wufp/Documents/koc-loop/design-qa-resource-cards-final.png`
+- 聚焦卡片并排对照：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-comparison.png`
+- 本地页面：`http://localhost:8080/`，KOC资源状态
+
+## 环境与归一化
+
+- CSS 视口：1280 × 720；设备像素比 2；浏览器截图按 1280 × 720 CSS 像素输出。
+- 参考卡片像素：478 × 700；最终完整页面截图：1280 × 720。
+- 聚焦对照将实现中的 304px 宽卡片等比放大到 478px，并与参考卡片并排查看；没有把两张独立截图当作同一对比证据。
+- 桌面状态为三列卡片；同时验证 960px 两列、640px 一列。
+
+## 完整画面对比
+
+- 信息架构采用参考图的“头像 → 身份信息 → 简介 → 标签”顺序，同时保留 KOC LOOP 原有的粉丝、合作发布、合作来源和主页入口。
+- 卡片高度由旧版的大块分区缩短到首屏实测约 274—278px；三列宽度均为 304px，页面没有横向溢出。
+- 小红书使用粉红顶部识别线和柔和粉色标签，抖音使用深色顶部识别线，继续沿用现有平台标识。
+
+## 聚焦区域检查
+
+- 字体与层级：昵称 14px 并提高字重；账号号、属地、标签和数据从第一轮偏小的 8px 提升至 9—11px，信息仍紧凑但可读性更好。
+- 间距与布局：圆形头像、昵称与性别保持同一视觉组；账号号、平台和属地收进头像右侧；粉丝、合作发布、最近合作合并为一条浅底数据栏。
+- 颜色与视觉标记：标签颜色与平台呼应，状态对比足够；卡片 hover 只使用轻微位移和阴影，不改变布局。
+- 图片与资产：当前账号数据没有头像 URL，因此保留现有首字母头像作为明确的数据缺失状态，没有伪造真人头像；平台标识继续使用项目已有资产。
+- 文案与内容：真实简介最多展示三行；“未填写简介”“还没有简介”等占位内容统一折叠为“暂无简介”；无标签显示“待打标”；合作来源和主页入口合并到底部同一行。
+
+## 交互与响应式验证
+
+- 输入标签“美食探店”后，结果正确缩小为 3 个账号；清空后恢复完整列表。
+- 主页入口保持可见并保留现有跳转目标；首屏六张卡片均检测到有效入口文案。
+- 960px 视口为两列，640px 视口为一列，两个断点均无横向溢出。
+- 浏览器控制台无 error；本地应用、MySQL、Nginx 均正常运行。
+- 正式构建及完整自动化测试通过，共 82 项，无失败。
+
+## 迭代记录
+
+1. 第一轮已完成资料卡层级和高度压缩，但聚焦截图显示辅助文字偏小，无简介账号的底部留白仍较明显，记为 P2。
+2. 第二轮提高辅助文字字号，移除固定最小高度，并把合作来源与主页入口合并为同一底栏。
+3. 复查后卡片高度稳定在约 274—278px，关键内容可读，桌面与移动断点无溢出；先前 P2 已解决。
+
+## 结论
+
+- 没有遗留 P0、P1 或 P2 问题。
+- P3 后续项：如果 MCP 未来提供可靠头像 URL，可将首字母头像替换成真实头像，进一步接近参考图。
+
+final result: passed
+
+---
+
+# KOC LOOP KOC资源卡片底栏对齐设计 QA
+
+## 验证对象
+
+- 用户标注截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-fa2c031d-52ba-4ef5-bd70-2bf74adfa52d.png`
+- 最终浏览器截图：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-alignment-final.png`
+- 归一化并排对照：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-alignment-comparison.png`
+- 本地页面：`http://localhost:8080/?nav=resources`
+
+## 问题与调整
+
+1. 不同账号的简介和标签数量不一致时，合作来源与主页入口会跟随前方内容上下浮动，导致同一排卡片的底部结构不齐。
+2. 卡片保持纵向弹性布局，底栏改为自动占用剩余空间并贴住卡片底部；不增加固定高度，不改变数据或交互逻辑。
+3. 三列桌面视口下实测前三张卡片高度均为 267.9px，底栏顶部均为 333.4px，底栏底部均为 365.9px，误差为 0px。
+
+## 验证结果
+
+- CSS 视口：1280 × 720；三列卡片状态。
+- 不同简介、标签数量下，合作来源、来源标签和主页入口保持同一条水平基线。
+- 浏览器控制台无 error；页面 hover 位移不会改变静止状态的布局基线。
+- 页面渲染测试与 TypeScript 检查通过；本地 Docker 已重建。
+
+final result: passed
+
+---
+
+# KOC LOOP KOC资源卡片数据栏对齐设计 QA
+
+## 验证对象
+
+- 用户标注截图：`/var/folders/0g/8yrmts9s7v5g_xc60sxx0__80000gn/T/codex-clipboard-78335120-710a-4a7b-96f9-b1046191c788.png`
+- 双列实现截图：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-metrics-alignment-two-column.png`
+- 三列实现截图：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-metrics-alignment-final.png`
+- 归一化并排对照：`/Users/wufp/Documents/koc-loop/design-qa-resource-card-metrics-alignment-comparison.png`
+- 本地页面：`http://localhost:8080/?nav=resources`
+
+## 环境与归一化
+
+- 用户截图为 1674 × 1180px；双列实现截图为 837 × 591px。
+- 并排对照将用户截图归一化为 837 × 591px，与实现截图使用同一双列宽度和页面状态进行聚焦比较。
+- 同时在 1280 × 720 三列桌面视口复测，确认调整不依赖双列断点。
+
+## 问题与调整
+
+1. 先前仅把合作来源底栏贴底，数据栏仍跟随简介和标签数量上下浮动，形成 P2 对齐问题。
+2. 将弹性留白移动到标签区之后，数据栏与合作来源底栏作为一个完整结构贴住卡片底部；数据内容、标签和交互均未改动。
+
+## 验证结果
+
+- 双列第一排两张卡片的数据栏顶部均为 233.4px、底部均为 267.9px；底栏顶部均为 277.9px、底部均为 310.4px。
+- 双列第二排两张卡片的数据栏顶部均为 525.3px、底部均为 559.8px；底栏顶部均为 569.8px、底部均为 602.3px。
+- 三列前三张卡片的数据栏与底栏上下边界误差均为 0px。
+- 字体、颜色、图片资产和文案未发生变化；浏览器控制台无 error。
+- 页面渲染测试 14/14、TypeScript 检查和 Docker 构建均通过。
+
+final result: passed
--- a/dev.vars_exp
+++ b/dev.vars_exp
@@ -0,0 +1,32 @@
+KOC_PORTAL_URL=http://localhost:3000
+ADMIN_ALLOWED_EMAIL=operator@example.com
+ADMIN_INTERNAL_TOKEN=replace-with-a-random-secret
+KOC_DEV=1
+
+# Optional override. The production key must be stored as a runtime secret.
+AI_TOOL_CENTER_MCP_URL=https://middle-aitool.gbotai.cn/mcp
+AI_TOOL_CENTER_MCP_KEY=replace-with-mcp-key
+
+# Feishu custom app credentials. Keep the secret out of source control.
+FEISHU_APP_ID=cli_xxxxxxxxxxxxxxxxx
+FEISHU_APP_SECRET=replace-with-feishu-app-secret
+
+# 企业微信 · 全部选填；不填则临期催办与汇总都跳过，不影响其他功能
+# 管理后台 https://work.weixin.qq.com →「我的企业 → 企业信息 → 企业 ID」
+WECOM_CORP_ID=
+# 管理后台 →「应用管理 → 自建」→ 创建应用「KOC LOOP 催办」后，应用详情页顶部显示
+WECOM_AGENT_ID=
+# 同一应用详情页 →「Secret」→ 点「发送」→ 企业微信里收到 64 位字符串
+# 注意：Secret 只能发送一次，收到立刻保存；丢了只能点「重置 Secret」重新生成
+WECOM_SECRET=
+# 测试群 → 群设置 → 群机器人 → 添加机器人 → 完整 webhook URL
+# 形如 https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=xxxxxxxx
+WECOM_ROBOT_WEBHOOK=
+# 临期阈值，默认 3；超过则不在催办。取值 1-30
+WECOM_NOTIFY_DUE_DAYS=3
+# 企业微信通知总开关，默认 true；临时关闭设为 false
+WECOM_NOTIFY_ENABLED=true
+
+# 本地测试：让 /api/bootstrap 自动 seed 3 个测试 partner + 5 条待发布 distribution
+# 上线后改为 false
+SEED_DEMO_DATA=true
--- a/私有化部署指南.md
+++ b/私有化部署指南.md
@@ -1,6 +1,6 @@
 # KOC LOOP 私有化部署指南

-本文适用于 `codex/self-hosted-mysql` 分支。目标架构是运维提出的：Nginx 代理 + KOC 服务 + MySQL 数据库，并让后台、外部 KOC 领取页和 Agent MCP 都能通过一个公网域名访问。
+本文适用于 `main` 分支。目标架构是运维提出的：Nginx 代理 + KOC 服务 + MySQL 数据库，并让后台、外部 KOC 领取页和 Agent MCP 都能通过一个公网域名访问。

 ## 1. 部署形态

@@ -35,7 +35,7 @@
 ```bash
 git clone ssh://git@gta.gbotai.cn:42001/wufengping/koc-loop.git
 cd koc-loop
-git checkout codex/self-hosted-mysql
+git checkout main
 cp .env.self-hosted.example .env.self-hosted
 ```

@@ -54,9 +54,17 @@ cp .env.self-hosted.example .env.self-hosted
 | `FEISHU_APP_ID` / `FEISHU_APP_SECRET` | 读取飞书内容表和配图 |
 | `AI_TOOL_CENTER_MCP_URL` / `AI_TOOL_CENTER_MCP_KEY` | 小红书公开数据采集服务 |
 | `ENABLE_SCHEDULER` | 是否启用每天 09:00 自动采集，生产保持 `true` |
+| `WECOM_CORP_ID` / `WECOM_AGENT_ID` / `WECOM_SECRET` | 企业微信自建应用，用于临期催办的应用消息推送；三项需同时填写，全部留空则只走群机器人 |
+| `WECOM_ROBOT_WEBHOOK` | 企业微信群机器人 webhook，用于当日催办汇总 |
+| `WECOM_NOTIFY_DUE_DAYS` | 临期阈值，默认 3，超过则不在催办；可填 1–30 |
+| `WECOM_NOTIFY_ENABLED` | 企业微信通知总开关，默认 `true`；临时关闭设为 `false` |
+
+企业微信变量全部选填：不填则临期催办与汇总都跳过，不影响其他功能。所有变量都可以在「合作方」页面顶部「企业微信连通性」面板查看就绪状态，并用「测试群机器人」「测试发送」按钮验证。

 密钥必须由密码管理器生成，禁止提交到 Git、聊天、部署日志或 URL。三个业务密钥 `ADMIN_INTERNAL_TOKEN`、`KOC_MCP_API_KEY`、`AI_TOOL_CENTER_MCP_KEY` 不得复用。

+`APP_ORIGIN` 必须填写用户实际访问的 HTTPS 公网地址，不能填写 `localhost`、`app:3000` 或其他容器内部地址。Excel 中的视频下载链接会优先使用这个地址；前置网关还必须把原始 `Host`、`X-Forwarded-Host` 和 `X-Forwarded-Proto` 传给仓库内的 Nginx。
+
 ## 4. 首次启动

 ```bash
@@ -88,7 +96,9 @@ curl -fsS http://127.0.0.1:${HTTP_PORT:-80}/api/health
 3. 80 端口只做 301 跳转；
 4. 保留 `/koc/` 静态规则、`/api/mcp` 长连接规则和 `/` 反向代理规则。

-MCP 路由已经关闭代理缓冲并将超时时间延长到 1 小时，避免 Agent 的长调用被 Nginx 提前截断。
+MCP 路由已经关闭代理缓冲并将超时时间延长到 1 小时，避免 Agent 的长调用被 Nginx 提前截断。仓库内 Nginx 同时将上传限制设为 85 MB，用于接收最多 80 MB 的批量回填 Excel；公司网关或负载均衡的请求体限制也必须不低于 85 MB。
+
+视频下载接口必须经过 `/api/partner-image` 反向代理，正常响应应包含 `Content-Type: video/mp4` 和带 `.mp4` 文件名的 `Content-Disposition: attachment`。不要在网关层改写该响应类型或移除附件响应头。

 ## 6. 迁移原 Sites 数据

@@ -124,12 +134,13 @@ npm run db:import-json -- /backup/koc-d1-export.json

 导入脚本按业务依赖顺序写入，并使用主键/唯一键安全更新已有记录。正式迁移前先在测试库演练并核对任务数、笔记数、领取数、发布数和账号数。

-### 6.2 R2 图片导入
+### 6.2 R2 媒体文件导入

 把 R2 按原对象 key 导出到一个目录，目录层级必须保留，例如：

 ```text
 content-assets/...
+content-videos/...
 publish-evidence/...
 creator-center/...
 ```
@@ -141,7 +152,7 @@ UPLOAD_DIR=/data/koc/uploads \
  npm run storage:import -- /backup/koc-r2-export
 ```

-脚本会复制文件，并为缺少元数据的图片生成 Content-Type 元数据。容器部署时也可以在宿主机临时挂载 `upload_data` 卷后执行。
+脚本会复制文件，并为缺少元数据的媒体文件生成 Content-Type 元数据。容器部署时也可以在宿主机临时挂载 `upload_data` 卷后执行。

 ## 7. 上线验收

@@ -155,9 +166,11 @@ UPLOAD_DIR=/data/koc/uploads \
 6. 上传创作者截图并填写曝光量、阅读量；
 7. 后台立即采集一篇笔记成功；
 8. 保存次日采集计划，确认数据库产生 `collection_runs`；
-9. 导出的 Excel 内能直接看到原图和截图；
-10. Agent 用 `KOC_MCP_API_KEY` 调用 `/api/mcp` 能发现全部工具；
-11. 重启全部容器后数据与图片不丢失。
+9. 图文任务导出的 Excel 内能直接看到原图和截图；
+10. 视频任务导出的 Excel 不含“图片”列，包含“视频”列，点击链接能下载扩展名为 `.mp4` 且可正常播放的文件；
+11. 批量回填 Excel 可以上传，发布链接、笔记截图和单篇笔记数据分析截图均能正确回写；
+12. Agent 用 `KOC_MCP_API_KEY` 调用 `/api/mcp` 能发现全部工具；
+13. 重启全部容器后数据、图片和视频不丢失。

 ## 8. 备份与恢复

@@ -180,6 +193,8 @@ docker compose --env-file .env.self-hosted \
  -f docker-compose.self-hosted.yml up -d --build
 ```

+应用容器每次启动都会按文件名顺序执行尚未应用的 `mysql/*.sql`。本次版本包含平台/视频字段、账号性别/简介/标签以及“当前联系人”字段的增量迁移；升级后应检查容器日志确认 `0005`、`0006`、`0007` 已执行或已被识别为历史迁移。
+
 数据库迁移只允许向前追加新的 `mysql/*.sql` 文件，禁止修改已经在生产执行过的迁移。应用回滚到旧镜像前，要确认旧代码兼容当前数据库结构；涉及不可逆结构变化时，必须同时准备数据库恢复方案。

 ## 10. 运维排查
@@ -189,8 +204,65 @@ docker compose --env-file .env.self-hosted \
 | `/api/health` 返回 503 | 查看 MySQL 容器健康状态与应用数据库变量 |
 | 登录页可开但登录失败 | 确认迁移完成、超级管理员变量仅用于初始化 |
 | 配图或截图 404 | 检查 `upload_data` 卷和 `UPLOAD_DIR=/data/koc/uploads` |
+| Excel 视频链接出现 localhost 或无法访问 | 检查 `APP_ORIGIN`、公网域名和网关转发的 Host/Proto 请求头 |
+| 视频下载后不是 MP4 或无法播放 | 检查 `/api/partner-image` 是否经过应用代理、文件是否完整，以及网关是否保留 Content-Type/Content-Disposition |
+| 批量回填表上传返回 413 | 将公司网关、负载均衡和 Nginx 的请求体限制统一提高到至少 85 MB |
 | 09:00 未自动采集 | 检查 `ENABLE_SCHEDULER=true`、服务器日志和采集 MCP 网络 |
 | MCP 401 | 检查请求头是否为 `Authorization: Bearer <KOC_MCP_API_KEY>` |
 | 飞书读取失败 | 检查应用权限、文档授权和服务器到飞书 OpenAPI 的网络 |

 生产日志不得打印数据库密码、飞书 Secret、MCP key 或完整带 key 的采集服务 URL。
+
+## 11. 企业微信自建应用与群机器人配置
+
+KOC LOOP 的企业微信通知是**单向外发**：每天 09:00 与定时采集一同触发，给合作方绑定的外部联系人推送临期催办，再向群机器人发送当日汇总。不需要 OAuth 回调，也不需要拉通讯录。
+
+### 11.1 群机器人（用于当日催办汇总）
+
+1. 在企业微信里建立一个用于催办的群；
+2. 群设置 → 群机器人 → 添加机器人 → 复制 webhook 地址，形如 `https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=...`；
+3. 把整个 webhook 写入 `WECOM_ROBOT_WEBHOOK`。
+
+只要这一项就绪，每天 09:00 的催办汇总就会推到该群。
+
+### 11.2 自建应用（用于给单个合作方发应用消息）
+
+1. 登录企业微信管理后台 → 应用管理 → 自建 → 创建应用，名称建议 `KOC LOOP 催办`；
+2. 应用详情页记下 `AgentId`，填入 `WECOM_AGENT_ID`；
+3. 我的企业 → 企业信息 → 企业 ID，填入 `WECOM_CORP_ID`；
+4. 应用详情页 → Secret → 发送 Secret 到管理端，填入 `WECOM_SECRET`；
+5. 应用「可见范围」必须包含合作方对应的企业微信成员，否则推送时会返回 `invaliduser`。
+
+### 11.3 绑定外部联系人
+
+企业微信应用消息推送需要 `external_userid`（外部联系人 ID），KOC LOOP 不自动拉取，需要手动绑定：
+
+1. 在企业微信「客户联系 → 外部联系人」里找到合作方对应的客户；
+2. 复制其 external_userid（形如 `woxxxxxx`）；
+3. 在 KOC LOOP 后台「合作方」页面，找到对应合作方行，点「绑定」粘贴 ID → 「保存」；
+4. 点「测试发送」验证；如果返回 `应用消息发送失败`，多半是应用可见范围未包含该外部联系人，或 external_userid 复制错了。
+
+外部联系人 ID 不入库后不会自动同步企业微信端的变更；如果合作方的外部联系人在企业微信里被删除或转移，需要回来更新这一列。
+
+### 11.4 验证
+
+部署后到后台「合作方」页面，顶部「企业微信连通性」面板应显示三项就绪状态：群机器人、自建应用消息、临期阈值。逐项点「测试」：
+
+- 「测试群机器人」：群内收到 `[KOC LOOP 测试] 群机器人连通性正常，时间 ...` 即配置成功；
+- 合作方行的「测试发送」：对应外部联系人收到同样格式的测试消息即绑定正确。
+
+如果面板显示「未配置」但已填环境变量，先确认容器加载了新的 `.env.self-hosted`（重启服务），再回到该页面点「刷新状态」。
+
+### 11.5 任务发布到企微客户群（企业群发）
+
+平台支持把任务以「企业群发」方式推送到多个外部客户群。**企业微信不允许外部群添加群机器人 webhook**，官方路径是半自动群发：平台创建群发任务 → 各群的群主在企业微信客户端「群发助手」里点击发送 → 消息才送达客户群。依赖 11.2 的同一个自建应用，无需新增环境变量；数据库表 `wecom_group_chats` / `wecom_group_pushes` 由迁移 `mysql/0009_wecom_group_push.sql` 创建（`npm run db:migrate` 自动执行）。
+
+使用前需在企业微信管理后台完成三项配置：
+
+1. **客户联系 → 配置 → 可调用应用**：把该自建应用加入可调用列表，否则客户群相关接口会报权限错误；
+2. **应用可见范围**：必须包含所有目标群的群主（群主是群发任务的确认人）；
+3. **应用可信 IP**（应用详情页 → 开发者接口 → 企业可信 IP）：必须包含服务器出口 IP，否则接口返回 `60020 not allow to access from your ip`。
+
+使用流程：后台「任务中心」→ 任务行「发布到企微群」→ 弹窗内「同步群列表」（拉取客户联系下的正常状态客户群）→ 勾选目标群、确认文案 → 「创建群发」。创建成功后平台返回 msgid 并落库 `wecom_group_pushes`，各群主在企微客户端收到群发助手提醒，**点击发送后**消息（含 KOC 领取链接）才出现在群里。
+
+限制与注意：每个客户群每月最多接收「当月天数」条企业群发；群主使用的企业微信客户端需 ≥4.1.10 才能免选群直接发送；无效的 chat_id 会进入失败列表但不影响其他群。
--- a/部署指南.md
+++ b/部署指南.md
@@ -1,4 +1,6 @@
-# KOC LOOP 部署指南
+# KOC LOOP Sites 旧版部署指南
+
+> 此文档仅适用于历史 `codex/sites-release-controls` 分支。`main` 已切换为 Next.js + MySQL + Nginx 私有化部署，正式部署请使用 [KOC LOOP 私有化部署指南](KOC%20LOOP%20私有化部署指南.md)，不要按本文把 `main` 发布到 Sites。

 KOC LOOP 由两个独立站点组成：

--- a/koc-portal/app/batch-workbook-upload.ts
+++ b/koc-portal/app/batch-workbook-upload.ts
@@ -0,0 +1,341 @@
+import { strFromU8, unzipSync, zipSync } from "fflate";
+
+export const PARTNER_BATCH_UPLOAD_MAX_BYTES = 80_000_000;
+
+type WorkbookCell = {
+  reference: string;
+  row: number;
+  column: number;
+  attributes: string;
+  body: string;
+  value: string;
+};
+
+type ScreenshotColumns = {
+  headerRow: number;
+  columns: Set<number>;
+};
+
+function decodeXml(value: string) {
+  return value
+    .replace(/<[^>]+>/g, "")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&apos;/g, "'")
+    .replace(/&amp;/g, "&")
+    .replace(/&#(\d+);/g, (_, code) => String.fromCodePoint(Number(code)))
+    .replace(/&#x([0-9a-f]+);/gi, (_, code) =>
+      String.fromCodePoint(Number.parseInt(code, 16)),
+    );
+}
+
+function textNodes(xml: string) {
+  return [...xml.matchAll(/<t\b[^>]*>([\s\S]*?)<\/t>/g)]
+    .map((match) => decodeXml(match[1]))
+    .join("");
+}
+
+function columnIndex(reference: string) {
+  const letters = reference.match(/^[A-Z]+/i)?.[0]?.toUpperCase() ?? "";
+  let result = 0;
+  for (const letter of letters) result = result * 26 + letter.charCodeAt(0) - 64;
+  return Math.max(0, result - 1);
+}
+
+function normalizeHeader(value: string) {
+  return value.replace(/[\s_\-（）()]/g, "").toLocaleLowerCase("zh-CN");
+}
+
+function parseCells(worksheetXml: string, sharedStrings: string[]) {
+  const cells: WorkbookCell[] = [];
+  for (const match of worksheetXml.matchAll(/<c\b([^>]*)>([\s\S]*?)<\/c>/g)) {
+    const attributes = match[1];
+    const body = match[2];
+    const reference = attributes.match(/\br="([A-Z]+\d+)"/i)?.[1] ?? "";
+    if (!reference) continue;
+    const type = attributes.match(/\bt="([^"]+)"/)?.[1] ?? "";
+    const rawValue = body.match(/<v>([\s\S]*?)<\/v>/)?.[1] ?? "";
+    const value =
+      type === "s"
+        ? sharedStrings[Number(rawValue)] ?? ""
+        : type === "inlineStr"
+          ? textNodes(body)
+          : decodeXml(rawValue);
+    cells.push({
+      reference,
+      row: Number(reference.match(/\d+$/)?.[0] ?? 0),
+      column: columnIndex(reference),
+      attributes,
+      body,
+      value: value.trim(),
+    });
+  }
+  return cells;
+}
+
+function findScreenshotColumns(cells: WorkbookCell[]): ScreenshotColumns {
+  for (let row = 1; row <= 8; row += 1) {
+    const columns = new Set<number>();
+    for (const cell of cells) {
+      if (cell.row !== row) continue;
+      const header = normalizeHeader(cell.value);
+      if (
+        header === normalizeHeader("笔记截图") ||
+        header === normalizeHeader("发布截图") ||
+        header === normalizeHeader("数据分析截图") ||
+        header === normalizeHeader("数据分析截图（单篇笔记数据分析截图）") ||
+        header === normalizeHeader("创作者中心截图")
+      ) {
+        columns.add(cell.column);
+      }
+    }
+    if (columns.size >= 2) return { headerRow: row, columns };
+  }
+  throw new Error("表格结构不正确，请使用本领取页面导出的批量回填表");
+}
+
+function relationshipMap(xml: string) {
+  const relationships = new Map<string, string>();
+  for (const match of xml.matchAll(/<Relationship\b([^>]*)\/?\s*>/g)) {
+    const id = match[1].match(/\bId="([^"]+)"/)?.[1] ?? "";
+    const target = match[1].match(/\bTarget="([^"]+)"/)?.[1] ?? "";
+    if (id && target) relationships.set(id, decodeXml(target));
+  }
+  return relationships;
+}
+
+function normalizeZipPath(value: string) {
+  const result: string[] = [];
+  for (const part of value.split("/")) {
+    if (!part || part === ".") continue;
+    if (part === "..") result.pop();
+    else result.push(part);
+  }
+  return result.join("/");
+}
+
+function resolveZipPath(base: string, target: string) {
+  const slash = base.lastIndexOf("/");
+  const directory = slash >= 0 ? base.slice(0, slash + 1) : "";
+  return normalizeZipPath(`${directory}${target}`);
+}
+
+function wpsScreenshotMedia(
+  entries: Record<string, Uint8Array>,
+  cells: WorkbookCell[],
+  screenshotColumns: ScreenshotColumns,
+) {
+  const result = new Set<string>();
+  let expected = 0;
+  const screenshotIds = new Set<string>();
+  for (const cell of cells) {
+    if (
+      cell.row <= screenshotColumns.headerRow ||
+      !screenshotColumns.columns.has(cell.column)
+    ) {
+      continue;
+    }
+    const id = decodeXml(cell.body).match(/DISPIMG\("([^"]+)"/i)?.[1];
+    if (id) {
+      expected += 1;
+      screenshotIds.add(id);
+    }
+  }
+  if (screenshotIds.size === 0) return { result, expected, resolved: 0 };
+
+  const cellImagesXml = entries["xl/cellimages.xml"]
+    ? strFromU8(entries["xl/cellimages.xml"])
+    : "";
+  const relationships = relationshipMap(
+    entries["xl/_rels/cellimages.xml.rels"]
+      ? strFromU8(entries["xl/_rels/cellimages.xml.rels"])
+      : "",
+  );
+  let resolved = 0;
+  for (const match of cellImagesXml.matchAll(
+    /<(?:etc:)?cellImage\b[^>]*>([\s\S]*?)<\/(?:etc:)?cellImage>/g,
+  )) {
+    const id = match[1].match(/<(?:xdr:)?cNvPr\b[^>]*\bname="([^"]+)"/)?.[1];
+    const relationshipId = match[1].match(
+      /<(?:a:)?blip\b[^>]*\br:embed="([^"]+)"/,
+    )?.[1];
+    if (!id || !relationshipId || !screenshotIds.has(id)) continue;
+    const target = relationships.get(relationshipId);
+    if (!target) continue;
+    result.add(resolveZipPath("xl/cellimages.xml", target));
+    resolved += 1;
+  }
+  return { result, expected, resolved };
+}
+
+function drawingScreenshotMedia(
+  entries: Record<string, Uint8Array>,
+  screenshotColumns: ScreenshotColumns,
+) {
+  const result = new Set<string>();
+  let expected = 0;
+  let resolved = 0;
+  const worksheetXml = entries["xl/worksheets/sheet1.xml"]
+    ? strFromU8(entries["xl/worksheets/sheet1.xml"])
+    : "";
+  const sheetRelationships = relationshipMap(
+    entries["xl/worksheets/_rels/sheet1.xml.rels"]
+      ? strFromU8(entries["xl/worksheets/_rels/sheet1.xml.rels"])
+      : "",
+  );
+  const drawingId = worksheetXml.match(/<drawing\b[^>]*r:id="([^"]+)"/)?.[1];
+  const drawingTarget = drawingId ? sheetRelationships.get(drawingId) : undefined;
+  if (!drawingTarget) return { result, expected, resolved };
+  const drawingPath = resolveZipPath("xl/worksheets/sheet1.xml", drawingTarget);
+  const drawingXml = entries[drawingPath] ? strFromU8(entries[drawingPath]) : "";
+  const relationshipPath = `${drawingPath.slice(0, drawingPath.lastIndexOf("/") + 1)}_rels/${drawingPath.slice(drawingPath.lastIndexOf("/") + 1)}.rels`;
+  const drawingRelationships = relationshipMap(
+    entries[relationshipPath] ? strFromU8(entries[relationshipPath]) : "",
+  );
+  for (const anchor of drawingXml.matchAll(
+    /<xdr:(?:oneCellAnchor|twoCellAnchor)\b[^>]*>[\s\S]*?<xdr:from>([\s\S]*?)<\/xdr:from>[\s\S]*?<a:blip\b[^>]*r:embed="([^"]+)"[\s\S]*?<\/xdr:(?:oneCellAnchor|twoCellAnchor)>/g,
+  )) {
+    const column = Number(anchor[1].match(/<xdr:col>(\d+)<\/xdr:col>/)?.[1]);
+    const zeroBasedRow = Number(anchor[1].match(/<xdr:row>(\d+)<\/xdr:row>/)?.[1]);
+    if (
+      !Number.isInteger(column) ||
+      !Number.isInteger(zeroBasedRow) ||
+      zeroBasedRow + 1 <= screenshotColumns.headerRow ||
+      !screenshotColumns.columns.has(column)
+    ) {
+      continue;
+    }
+    expected += 1;
+    const target = drawingRelationships.get(anchor[2]);
+    if (!target) continue;
+    result.add(resolveZipPath(drawingPath, target));
+    resolved += 1;
+  }
+  return { result, expected, resolved };
+}
+
+function richValueScreenshotMedia(
+  entries: Record<string, Uint8Array>,
+  cells: WorkbookCell[],
+  screenshotColumns: ScreenshotColumns,
+) {
+  const result = new Set<string>();
+  let expected = 0;
+  let resolved = 0;
+  const metadataXml = entries["xl/metadata.xml"]
+    ? strFromU8(entries["xl/metadata.xml"])
+    : "";
+  const richValueXml = entries["xl/richData/rdrichvalue.xml"]
+    ? strFromU8(entries["xl/richData/rdrichvalue.xml"])
+    : "";
+  const richValueRelXml = entries["xl/richData/richValueRel.xml"]
+    ? strFromU8(entries["xl/richData/richValueRel.xml"])
+    : "";
+  const relationships = relationshipMap(
+    entries["xl/richData/_rels/richValueRel.xml.rels"]
+      ? strFromU8(entries["xl/richData/_rels/richValueRel.xml.rels"])
+      : "",
+  );
+  if (!metadataXml || !richValueXml || !richValueRelXml || !relationships.size) {
+    return { result, expected, resolved };
+  }
+  const valueMetadataXml =
+    metadataXml.match(/<valueMetadata\b[^>]*>([\s\S]*?)<\/valueMetadata>/)?.[1] ??
+    "";
+  const metadataToRichValue = [
+    ...valueMetadataXml.matchAll(
+      /<bk\b[^>]*>[\s\S]*?<rc\b[^>]*\bv="(\d+)"[^>]*\/>[\s\S]*?<\/bk>/g,
+    ),
+  ].map((match) => Number(match[1]));
+  const richValueToRelationship = [
+    ...richValueXml.matchAll(/<rv\b[^>]*>([\s\S]*?)<\/rv>/g),
+  ].map((match) => Number(match[1].match(/<v>(\d+)<\/v>/)?.[1] ?? -1));
+  const relationshipIds = [
+    ...richValueRelXml.matchAll(/<rel\b[^>]*\br:id="([^"]+)"[^>]*\/>/g),
+  ].map((match) => match[1]);
+
+  for (const cell of cells) {
+    if (
+      cell.row <= screenshotColumns.headerRow ||
+      !screenshotColumns.columns.has(cell.column)
+    ) {
+      continue;
+    }
+    const metadataIndex = Number(cell.attributes.match(/\bvm="(\d+)"/)?.[1] ?? 0);
+    if (!metadataIndex) continue;
+    expected += 1;
+    const richValueIndex = metadataToRichValue[metadataIndex - 1];
+    const relationshipIndex = richValueToRelationship[richValueIndex];
+    const relationshipId = relationshipIds[relationshipIndex];
+    const target = relationships.get(relationshipId);
+    if (!target) continue;
+    result.add(resolveZipPath("xl/richData/richValueRel.xml", target));
+    resolved += 1;
+  }
+  return { result, expected, resolved };
+}
+
+export type CompactedPartnerBatchWorkbook = {
+  bytes: Uint8Array;
+  removedMediaCount: number;
+  preservedScreenshotCount: number;
+};
+
+/**
+ * Oversized exports are usually caused by full-resolution source images. The
+ * upload only needs the two screenshot columns, so retain those image entries
+ * and omit source media from the temporary upload copy.
+ */
+export function compactPartnerBatchWorkbookForUpload(
+  input: Uint8Array,
+): CompactedPartnerBatchWorkbook {
+  const isMediaFile = (name: string) =>
+    name.startsWith("xl/media/") && !name.endsWith("/");
+  const structure = unzipSync(input, {
+    filter: (file) => !isMediaFile(file.name),
+  });
+  const worksheetXml = structure["xl/worksheets/sheet1.xml"]
+    ? strFromU8(structure["xl/worksheets/sheet1.xml"])
+    : "";
+  if (!worksheetXml) throw new Error("Excel 中没有可读取的批量回填工作表");
+  const sharedXml = structure["xl/sharedStrings.xml"]
+    ? strFromU8(structure["xl/sharedStrings.xml"])
+    : "";
+  const sharedStrings = [
+    ...sharedXml.matchAll(/<si\b[^>]*>([\s\S]*?)<\/si>/g),
+  ].map((match) => textNodes(match[1]));
+  const cells = parseCells(worksheetXml, sharedStrings);
+  const screenshotColumns = findScreenshotColumns(cells);
+  const formats = [
+    wpsScreenshotMedia(structure, cells, screenshotColumns),
+    drawingScreenshotMedia(structure, screenshotColumns),
+    richValueScreenshotMedia(structure, cells, screenshotColumns),
+  ];
+  const screenshotMedia = new Set<string>();
+  let expectedScreenshotCount = 0;
+  let resolvedScreenshotCount = 0;
+  for (const format of formats) {
+    expectedScreenshotCount += format.expected;
+    resolvedScreenshotCount += format.resolved;
+    for (const name of format.result) screenshotMedia.add(name);
+  }
+  if (resolvedScreenshotCount < expectedScreenshotCount) {
+    throw new Error("表格中的回填截图无法完整识别，请重新导出最新版回填表");
+  }
+
+  let mediaCount = 0;
+  const entries = unzipSync(input, {
+    filter: (file) => {
+      if (!isMediaFile(file.name)) return true;
+      mediaCount += 1;
+      return screenshotMedia.has(file.name);
+    },
+  });
+  const bytes = zipSync(entries, { level: 6 });
+  return {
+    bytes,
+    removedMediaCount: Math.max(0, mediaCount - screenshotMedia.size),
+    preservedScreenshotCount: screenshotMedia.size,
+  };
+}
--- a/koc-portal/app/globals.css
+++ b/koc-portal/app/globals.css
@@ -48,6 +48,104 @@ button:disabled {
  opacity: 0.5;
 }

+.platform-badge {
+  display: inline-flex !important;
+  width: auto !important;
+  height: 24px;
+  align-items: center;
+  flex: none;
+  gap: 5px;
+  margin: 0 !important;
+  padding: 2px 7px 2px 3px;
+  border: 1px solid #e1e7e4;
+  border-radius: 8px;
+  color: #52615c !important;
+  background: rgb(255 255 255 / 0.92);
+  font-size: 9px !important;
+  font-weight: 720;
+  line-height: 1 !important;
+  white-space: nowrap;
+}
+
+.platform-badge.compact {
+  height: 19px;
+  gap: 4px;
+  padding: 2px 5px 2px 2px;
+  border-radius: 6px;
+  font-size: 8px !important;
+}
+
+.platform-logo {
+  display: grid !important;
+  width: 18px !important;
+  height: 18px !important;
+  place-items: center;
+  flex: none;
+  overflow: hidden;
+  margin: 0 !important;
+  border-radius: 5px;
+  line-height: 1 !important;
+}
+
+.platform-badge.compact .platform-logo {
+  width: 14px !important;
+  height: 14px !important;
+  border-radius: 4px;
+}
+
+.platform-logo.xiaohongshu {
+  color: white !important;
+  background: #ff2442;
+}
+
+.platform-logo.xiaohongshu b {
+  color: inherit;
+  font-size: 5px;
+  font-weight: 900;
+  letter-spacing: -0.12em;
+  transform: translateX(-0.2px);
+}
+
+.platform-badge.compact .platform-logo.xiaohongshu b {
+  font-size: 4px;
+}
+
+.platform-logo.douyin {
+  background: #080b12;
+}
+
+.platform-logo.douyin svg {
+  width: 16px;
+  height: 16px;
+}
+
+.platform-badge.compact .platform-logo.douyin svg {
+  width: 13px;
+  height: 13px;
+}
+
+.platform-logo.douyin .douyin-cyan { fill: #25f4ee; transform: translate(-0.7px, 0.7px); }
+.platform-logo.douyin .douyin-red { fill: #fe2c55; transform: translate(0.7px, -0.3px); }
+.platform-logo.douyin .douyin-white { fill: #fff; }
+
+.platform-meta-line,
+.hero-platform-line {
+  display: inline-flex !important;
+  min-width: 0;
+  align-items: center;
+  flex-wrap: wrap;
+  gap: 6px;
+}
+
+.platform-meta-line > span,
+.hero-platform-line > * {
+  margin: 0 !important;
+}
+
+.hero-platform-line {
+  margin-bottom: 10px;
+}
+
 .portal-shell {
  width: min(100%, 1120px);
  min-height: 100vh;
@@ -679,6 +777,15 @@ footer {
  cursor: not-allowed;
 }

+.batch-workbook-input {
+  position: absolute;
+  width: 1px;
+  height: 1px;
+  overflow: hidden;
+  opacity: 0;
+  pointer-events: none;
+}
+
 .share-composer {
  display: grid;
  grid-template-columns: minmax(0, 1fr) minmax(190px, 0.45fr) auto;
@@ -843,6 +950,34 @@ footer {
  text-align: center;
 }

+.note-thumb.platform-video-thumb {
+  display: grid;
+  place-items: center;
+}
+
+.platform-badge.logo-only {
+  width: 34px !important;
+  height: 34px;
+  padding: 0;
+  border: 0;
+  background: transparent;
+}
+
+.platform-badge.logo-only .platform-logo {
+  width: 34px !important;
+  height: 34px !important;
+  border-radius: 9px;
+}
+
+.platform-badge.logo-only .platform-logo.douyin svg {
+  width: 29px;
+  height: 29px;
+}
+
+.platform-badge.logo-only .platform-logo.xiaohongshu b {
+  font-size: 8px;
+}
+
 .note-index {
  display: grid;
  width: 34px;
@@ -1135,12 +1270,32 @@ footer {
  white-space: pre-wrap;
 }

-.note-images {
+.note-images,
+.note-videos {
  margin-top: 30px;
  padding-top: 24px;
  border-top: 1px solid #edf0ee;
 }

+.note-video-grid {
+  display: grid;
+  gap: 14px;
+}
+
+.note-video-card {
+  overflow: hidden;
+  border: 1px solid #e4e9e6;
+  border-radius: 12px;
+  background: #102a22;
+}
+
+.note-video-card video {
+  display: block;
+  width: 100%;
+  max-height: 680px;
+  background: #0b1f19;
+}
+
 .note-images-heading {
  display: flex;
  align-items: flex-end;
@@ -1234,7 +1389,8 @@ footer {
  font-weight: 650;
 }

-.note-image-actions button {
+.note-image-actions button,
+.note-image-actions a {
  height: 28px;
  padding: 0 10px;
  border: 1px solid #cfe1d9;
@@ -1243,9 +1399,12 @@ footer {
  background: #f2f8f5;
  font-size: 8px;
  font-weight: 680;
+  line-height: 26px;
+  text-decoration: none;
 }

-.note-image-actions button:hover {
+.note-image-actions button:hover,
+.note-image-actions a:hover {
  border-color: #9fc9b8;
  background: #eaf5f0;
 }
@@ -1633,6 +1792,11 @@ footer {
  font-weight: 620;
 }

+.toast.success {
+  color: var(--green-deep);
+  font-weight: 720;
+}
+
 .loading-shell {
  display: flex;
  align-items: center;
@@ -1782,9 +1946,14 @@ footer {

  .section-actions {
    width: 100%;
+    flex-wrap: wrap;
    justify-content: space-between;
  }

+  .section-actions > span {
+    margin-right: auto;
+  }
+
  .share-composer {
    grid-template-columns: 1fr;
  }
--- a/koc-portal/app/page.tsx
+++ b/koc-portal/app/page.tsx
@@ -6,6 +6,10 @@ import {
  formatShanghaiDate as formatDate,
  parseStoredDate,
 } from "./date-utils";
+import {
+  compactPartnerBatchWorkbookForUpload,
+  PARTNER_BATCH_UPLOAD_MAX_BYTES,
+} from "./batch-workbook-upload";

 type Assignment = {
  id: string;
@@ -30,6 +34,11 @@ type Assignment = {
    width: number | null;
    height: number | null;
  }>;
+  videos: Array<{
+    index: number;
+    width: number | null;
+    height: number | null;
+  }>;
 };

 type DelegationSummary = {
@@ -63,6 +72,8 @@ type TaskPayload = {
    dueAt: string;
    status: string;
    type: "content_publish" | "screenshot_collect";
+    platform: "小红书" | "抖音";
+    contentFormat: "image_text" | "video";
  };
  claim: null | {
    id: string;
@@ -92,7 +103,12 @@ function resolveAdminOrigin() {
  return window.location.origin;
 }

-function partnerApi(path: "/api/partner" | "/api/partner-upload") {
+function partnerApi(
+  path:
+    | "/api/partner"
+    | "/api/partner-upload"
+    | "/api/partner-batch-workbook",
+) {
  return `${resolveAdminOrigin()}${path}`;
 }

@@ -111,6 +127,11 @@ function statusLabel(item: Assignment, taskType = "content_publish") {
 }

 const MAX_TASK_RESULT_SCREENSHOTS = 9;
+const PUBLISH_BACKFILL_SUCCESS = "这篇笔记的发布记录回填成功啦～";
+
+function toastClassName(message: string) {
+  return message === PUBLISH_BACKFILL_SUCCESS ? "toast success" : "toast";
+}

 function resultScreenshotKeys(value: string | null) {
  const text = String(value ?? "").trim();
@@ -155,6 +176,36 @@ function safeFileBase(item: Assignment) {
  );
 }

+function PlatformBadge({
+  platform,
+  compact = false,
+  logoOnly = false,
+}: {
+  platform: "小红书" | "抖音";
+  compact?: boolean;
+  logoOnly?: boolean;
+}) {
+  return (
+    <span
+      className={`platform-badge ${compact ? "compact" : ""} ${logoOnly ? "logo-only" : ""}`}
+      aria-label={logoOnly ? platform : undefined}
+    >
+      <span className={`platform-logo ${platform === "抖音" ? "douyin" : "xiaohongshu"}`} aria-hidden="true">
+        {platform === "抖音" ? (
+          <svg viewBox="0 0 24 24" focusable="false">
+            <path className="douyin-cyan" d="M14.2 3.2v9.5a4.7 4.7 0 1 1-3.4-4.5v2.7a2.1 2.1 0 1 0 .8 1.7V3.2h2.6Zm0 0c.4 2.4 2 3.8 4.3 4.1V10c-1.7-.1-3.2-.7-4.3-1.7V3.2Z" />
+            <path className="douyin-red" d="M15.2 2.5V12a4.7 4.7 0 1 1-3.4-4.5v2.7a2.1 2.1 0 1 0 .8 1.7V2.5h2.6Zm0 0c.4 2.4 2 3.8 4.3 4.1v2.7c-1.7-.1-3.2-.7-4.3-1.7V2.5Z" />
+            <path className="douyin-white" d="M14.7 2.9v9.5a4.7 4.7 0 1 1-3.4-4.5v2.7a2.1 2.1 0 1 0 .8 1.7V2.9h2.6Zm0 0c.4 2.4 2 3.8 4.3 4.1v2.7c-1.7-.1-3.2-.7-4.3-1.7V2.9Z" />
+          </svg>
+        ) : (
+          <b>小红书</b>
+        )}
+      </span>
+      {!logoOnly && <span>{platform}</span>}
+    </span>
+  );
+}
+
 function exactArrayBuffer(bytes: Uint8Array) {
  const copy = new Uint8Array(bytes.byteLength);
  copy.set(bytes);
@@ -277,6 +328,7 @@ export default function Home() {
  const [adminOrigin, setAdminOrigin] = useState("");
  const [downloadingImage, setDownloadingImage] = useState<number | null>(null);
  const [batchDownloading, setBatchDownloading] = useState(false);
+  const [batchWorkbookWorking, setBatchWorkbookWorking] = useState(false);
  const [creatorWorking, setCreatorWorking] = useState(false);
  const [creatorStage, setCreatorStage] = useState("");
  const [selectedForShare, setSelectedForShare] = useState<string[]>([]);
@@ -289,6 +341,7 @@ export default function Home() {
  } | null>(null);
  const [noteContentCollapsed, setNoteContentCollapsed] = useState(false);
  const submitCardRef = useRef<HTMLFormElement | null>(null);
+  const batchWorkbookInputRef = useRef<HTMLInputElement | null>(null);
  const publishScreenshotPreview = useFilePreview(screenshot);
  const creatorScreenshotPreview = useFilePreview(creatorScreenshot);
  const taskResultPreviews = useMemo(
@@ -504,6 +557,25 @@ export default function Home() {
    return `${adminOrigin}/api/partner-image?${params}`;
  };

+  const noteVideoUrl = (
+    item: Assignment,
+    videoIndex: number,
+    download = false,
+  ) => {
+    const params = new URLSearchParams({
+      distribution: item.id,
+      index: String(videoIndex),
+      kind: "video",
+    });
+    if (download) params.set("download", "1");
+    if (delegationToken) params.set("share", delegationToken);
+    else {
+      params.set("task", taskToken);
+      params.set("claim", claimToken);
+    }
+    return `${adminOrigin}/api/partner-image?${params}`;
+  };
+
  const taskResultImageUrl = (item: Assignment, imageIndex: number) => {
    const params = new URLSearchParams({
      distribution: item.id,
@@ -527,6 +599,11 @@ export default function Home() {
      index: "1",
      kind,
    });
+    const evidenceKey =
+      kind === "publish"
+        ? item.publish_screenshot_key
+        : item.creator_screenshot_key;
+    if (evidenceKey) params.set("v", evidenceKey);
    if (delegationToken) params.set("share", delegationToken);
    else {
      params.set("task", taskToken);
@@ -644,6 +721,112 @@ export default function Home() {
    }
  };

+  const batchWorkbookParams = () => {
+    const params = new URLSearchParams();
+    if (delegationToken) params.set("share", delegationToken);
+    else {
+      params.set("task", taskToken);
+      params.set("claim", claimToken);
+    }
+    return params;
+  };
+
+  const exportBatchWorkbook = async () => {
+    try {
+      setBatchWorkbookWorking(true);
+      const response = await fetch(
+        `${partnerApi("/api/partner-batch-workbook")}?${batchWorkbookParams()}`,
+        { cache: "no-store" },
+      );
+      if (!response.ok) {
+        const result = (await response.json()) as { error?: string };
+        throw new Error(result.error || "Excel导出失败");
+      }
+      const disposition = response.headers.get("Content-Disposition") || "";
+      const encodedName = disposition.match(/filename\*=UTF-8''([^;]+)/i)?.[1];
+      const fileName = encodedName
+        ? decodeURIComponent(encodedName)
+        : `${payload?.task.name || "领取笔记"}-批量回填.xlsx`;
+      downloadBlob(await response.blob(), fileName);
+      setToast("Excel已导出，填写后从本页面上传即可批量回填");
+    } catch (reason) {
+      setToast(reason instanceof Error ? reason.message : "Excel导出失败");
+    } finally {
+      setBatchWorkbookWorking(false);
+    }
+  };
+
+  const importBatchWorkbook = async (event: ChangeEvent<HTMLInputElement>) => {
+    const file = event.target.files?.[0];
+    event.target.value = "";
+    if (!file) return;
+    try {
+      setBatchWorkbookWorking(true);
+      let uploadBody: Blob = file;
+      let compacted = false;
+      if (file.size > PARTNER_BATCH_UPLOAD_MAX_BYTES) {
+        setToast("文件较大，正在保留回填截图并精简原图…");
+        await new Promise<void>((resolve) => requestAnimationFrame(() => resolve()));
+        const compactedWorkbook = compactPartnerBatchWorkbookForUpload(
+          new Uint8Array(await file.arrayBuffer()),
+        );
+        if (compactedWorkbook.bytes.byteLength > PARTNER_BATCH_UPLOAD_MAX_BYTES) {
+          throw new Error("精简后的回填表仍超过80MB，请重新导出最新版回填表");
+        }
+        uploadBody = new Blob([exactArrayBuffer(compactedWorkbook.bytes)], {
+          type: "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
+        });
+        compacted = true;
+      }
+      const form = new FormData();
+      form.set("file", uploadBody, file.name);
+      form.set("taskToken", taskToken);
+      form.set("claimToken", claimToken);
+      form.set("delegationToken", delegationToken);
+      const response = await fetch(partnerApi("/api/partner-batch-workbook"), {
+        method: "POST",
+        body: form,
+      });
+      const contentType = response.headers.get("Content-Type") || "";
+      const result = (contentType.includes("application/json")
+        ? await response.json()
+        : {
+            error:
+              response.status === 413
+                ? "回填表超过上传限制，请重新导出最新版回填表"
+                : "批量回填服务暂时不可用，请稍后重试",
+          }) as {
+        error?: string;
+        updatedRows?: number;
+        publishedCount?: number;
+        noteScreenshotCount?: number;
+        analysisScreenshotCount?: number;
+      };
+      if (!response.ok) throw new Error(result.error || "批量回填失败");
+      await loadTask(taskToken, claimToken, delegationToken);
+      const details = [
+        result.publishedCount
+          ? `${result.publishedCount}篇发布信息`
+          : "",
+        result.noteScreenshotCount
+          ? `${result.noteScreenshotCount}张笔记截图`
+          : "",
+        result.analysisScreenshotCount
+          ? `${result.analysisScreenshotCount}张数据分析截图`
+          : "",
+      ].filter(Boolean);
+      setToast(
+        details.length > 0
+          ? `${compacted ? "文件已自动精简，" : ""}已更新${details.join("、")}`
+          : `${compacted ? "文件已自动精简，" : ""}表格已读取，没有需要更新的数据`,
+      );
+    } catch (reason) {
+      setToast(reason instanceof Error ? reason.message : "批量回填失败");
+    } finally {
+      setBatchWorkbookWorking(false);
+    }
+  };
+
  const prepareImage = async (item: Assignment, imageIndex: number) => {
    const response = await fetch(noteImageUrl(item, imageIndex));
    if (!response.ok) throw new Error("图片读取失败");
@@ -840,7 +1023,7 @@ export default function Home() {
          });
        }, 120);
      }
-      setToast("这篇笔记已回填，不会与其他笔记错配");
+      setToast(PUBLISH_BACKFILL_SUCCESS);
    } catch (reason) {
      setToast(reason instanceof Error ? reason.message : "回填失败");
    } finally {
@@ -987,7 +1170,10 @@ export default function Home() {
                <span>截图任务 {activeBatch.assignments.findIndex((item) => item.id === selected.id) + 1}</span>
                <span>只需上传结果截图</span>
              </div>
-              <p className="document-label">小红书搜索关键词</p>
+              <div className="document-label-row">
+                <p className="document-label">小红书搜索关键词</p>
+                <PlatformBadge platform={payload.task.platform} compact />
+              </div>
              <div className="note-title-row">
                <h1>{selected.title}</h1>
                <button
@@ -1107,7 +1293,7 @@ export default function Home() {
            </form>
          </div>
          <ImageLightbox image={previewImage} onClose={() => setPreviewImage(null)} />
-          {toast && <div className="toast">{toast}</div>}
+          {toast && <div className={toastClassName(toast)}>{toast}</div>}
        </main>
      );
    }
@@ -1136,7 +1322,9 @@ export default function Home() {
                <span>笔记内容已收起</span>
                <strong>{selected.title}</strong>
                <small>
-                  {selected.images.length} 张配图 · 已完成首次回填
+                  {selected.videos.length > 0
+                    ? `${selected.videos.length} 个视频`
+                    : `${selected.images.length} 张配图`} · 已完成首次回填
                </small>
              </div>
              <button
@@ -1150,6 +1338,7 @@ export default function Home() {
            <div className="note-document-content">
            <div className="note-document-meta">
              <span>笔记 {activeBatch.assignments.findIndex((item) => item.id === selected.id) + 1}</span>
+              <PlatformBadge platform={payload.task.platform} compact />
              <span>飞书源行 {selected.source_row ?? "—"}</span>
            </div>
            {selected.publish_url && (
@@ -1237,6 +1426,38 @@ export default function Home() {
                </div>
              </section>
            )}
+            {selected.videos.length > 0 && (
+              <section className="note-videos">
+                <div className="note-images-heading">
+                  <div>
+                    <p className="document-label">发布视频</p>
+                    <span>点击播放，可使用播放器菜单保存原视频</span>
+                  </div>
+                  <b>{selected.videos.length} 个</b>
+                </div>
+                <div className="note-video-grid">
+                  {selected.videos.map((video, index) => (
+                    <div className="note-video-card" key={video.index}>
+                      <video
+                        controls
+                        preload="metadata"
+                        playsInline
+                        src={noteVideoUrl(selected, video.index)}
+                      />
+                      <div className="note-image-actions">
+                        <span>视频 {index + 1}</span>
+                        <a
+                          href={noteVideoUrl(selected, video.index, true)}
+                          download={`${safeFileBase(selected)}-视频-${index + 1}.mp4`}
+                        >
+                          下载视频
+                        </a>
+                      </div>
+                    </div>
+                  ))}
+                </div>
+              </section>
+            )}
            </div>
          </article>

@@ -1261,7 +1482,7 @@ export default function Home() {
                inputMode="url"
                value={publishUrl}
                onChange={(event) => setPublishUrl(event.target.value)}
-                placeholder="可粘贴小红书长链、短链或整段分享文案"
+                placeholder={`可粘贴${payload.task.platform}作品链接或整段分享文案`}
                required
              />
              <small className="field-hint">
@@ -1399,7 +1620,6 @@ export default function Home() {
                      value={creatorExposure}
                      onChange={(event) => setCreatorExposure(event.target.value)}
                      placeholder="填写截图中的曝光量"
-                      required
                    />
                  </label>
                  <label>
@@ -1413,7 +1633,6 @@ export default function Home() {
                      value={creatorViews}
                      onChange={(event) => setCreatorViews(event.target.value)}
                      placeholder="填写截图中的阅读量"
-                      required
                    />
                  </label>
                </div>
@@ -1439,7 +1658,7 @@ export default function Home() {
          </form>
        </div>
        <ImageLightbox image={previewImage} onClose={() => setPreviewImage(null)} />
-        {toast && <div className="toast">{toast}</div>}
+        {toast && <div className={toastClassName(toast)}>{toast}</div>}
      </main>
    );
  }
@@ -1478,8 +1697,10 @@ export default function Home() {
                  : "合作社转派发布包"}
            </p>
            <h1>{payload.task.name}</h1>
-            <p>
-              {payload.task.brand} · {formatDate(payload.task.dueAt)}前{isScreenshotTask ? "提交" : "发布"}
+            <p className="platform-meta-line">
+              <span>{payload.task.brand}</span>
+              <PlatformBadge platform={payload.task.platform} compact />
+              <span>{formatDate(payload.task.dueAt)}前{isScreenshotTask ? "提交" : "发布"}</span>
              {payload.delegation ? ` · ${payload.delegation.label}` : ""}
            </p>
          </div>
@@ -1500,7 +1721,7 @@ export default function Home() {
                {isClaimOwner
                  ? isScreenshotTask
                    ? "打开一份查看关键词和要求，完成后单独上传截图；也可以转派给底层KOC"
-                    : "打开一篇，查看内容并单独回填；也可以选择笔记转派给底层KOC"
+                    : "可逐篇回填，也可导出Excel填写后批量上传；还可以选择笔记转派给底层KOC"
                  : isScreenshotTask
                    ? "打开任务查看搜索关键词和要求，完成后逐份上传截图"
                    : "打开一篇查看完整内容，发布后逐篇回填"}
@@ -1508,6 +1729,31 @@ export default function Home() {
            </div>
            <div className="section-actions">
              <span>{activeBatch.assignments.length} {isScreenshotTask ? "份" : "篇"}</span>
+              {!isScreenshotTask && (
+                <>
+                  <button
+                    type="button"
+                    disabled={batchWorkbookWorking}
+                    onClick={() => void exportBatchWorkbook()}
+                  >
+                    {batchWorkbookWorking ? "处理中…" : "导出Excel"}
+                  </button>
+                  <button
+                    type="button"
+                    disabled={batchWorkbookWorking}
+                    onClick={() => batchWorkbookInputRef.current?.click()}
+                  >
+                    上传回填表
+                  </button>
+                  <input
+                    ref={batchWorkbookInputRef}
+                    className="batch-workbook-input"
+                    type="file"
+                    accept=".xlsx,application/vnd.openxmlformats-officedocument.spreadsheetml.sheet"
+                    onChange={(event) => void importBatchWorkbook(event)}
+                  />
+                </>
+              )}
              {isClaimOwner && (
                <button
                  type="button"
@@ -1609,6 +1855,10 @@ export default function Home() {
                      loading="lazy"
                    />
                  </>
+                ) : item.videos.length > 0 ? (
+                  <span className="note-thumb platform-video-thumb">
+                    <PlatformBadge platform={payload.task.platform} logoOnly />
+                  </span>
                ) : (
                  <span className="note-thumb empty">无图</span>
                )}
@@ -1617,7 +1867,11 @@ export default function Home() {
                  <p>
                    {isScreenshotTask
                      ? `搜索关键词 · ${statusLabel(item, payload.task.type)}`
-                      : `${item.images.length} 张配图 · 飞书源行 ${item.source_row ?? "—"} · ${statusLabel(item, payload.task.type)}`}
+                      : <>
+                          <span>{item.videos.length > 0 ? `${item.videos.length} 个视频` : `${item.images.length} 张配图`}</span>
+                          <PlatformBadge platform={payload.task.platform} compact />
+                          <span>飞书源行 {item.source_row ?? "—"} · {statusLabel(item, payload.task.type)}</span>
+                        </>}
                    {!isScreenshotTask && item.creator_screenshot_key ? " · D7截图已交" : ""}
                    {item.delegation_label ? ` · 已转派给 ${item.delegation_label}` : ""}
                  </p>
@@ -1711,7 +1965,7 @@ export default function Home() {
              ? "请保存当前分享链接；完成任务后通过此链接上传截图即可。"
              : "请保存当前分享链接；发布后仍需通过此链接回填第7天截图、曝光量和阅读量。"}
        </div>
-        {toast && <div className="toast">{toast}</div>}
+        {toast && <div className={toastClassName(toast)}>{toast}</div>}
      </main>
    );
  }
@@ -1724,7 +1978,10 @@ export default function Home() {
      </header>
      <section className="task-hero">
        <div className="hero-copy">
-          <p className="micro">正在招募 · {payload.task.brand}</p>
+          <div className="hero-platform-line">
+            <p className="micro">正在招募 · {payload.task.brand}</p>
+            <PlatformBadge platform={payload.task.platform} />
+          </div>
          <h1>{payload.task.name}</h1>
          <p>
            {payload.task.type === "screenshot_collect"
@@ -1855,7 +2112,7 @@ export default function Home() {
        <div><span>3</span><strong>{payload.task.type === "screenshot_collect" ? "上传截图" : "单篇回填"}</strong><p>{payload.task.type === "screenshot_collect" ? "每份任务与截图一一对应" : "昵称、链接、截图一一对应"}</p></div>
      </section>
      <footer>由 KOC LOOP 提供任务与数据安全保障</footer>
-      {toast && <div className="toast">{toast}</div>}
+      {toast && <div className={toastClassName(toast)}>{toast}</div>}
    </main>
  );
 }
--- a/koc-portal/tests/rendered-html.test.mjs
+++ b/koc-portal/tests/rendered-html.test.mjs
@@ -37,14 +37,31 @@ test("keeps claiming minimal and backfill one-to-one", async () => {
  assert.match(page, /发布链接/);
  assert.match(page, /识别发布账号/);
  assert.match(page, /inputMode="url"/);
-  assert.match(page, /长链、短链或整段分享文案/);
+  assert.match(page, /作品链接或整段分享文案/);
  assert.doesNotMatch(page, /type="url"/);
  assert.match(page, /发布截图/);
  assert.match(page, /发布配图/);
  assert.match(page, /复制标题/);
  assert.match(page, /复制文案/);
  assert.match(page, /下载原图/);
+  assert.match(page, /下载视频/);
+  assert.match(page, /download\s*=\s*false/);
+  assert.match(page, /params\.set\("download", "1"\)/);
+  assert.match(page, /视频-\$\{index \+ 1\}\.mp4/);
+  assert.match(page, /function PlatformBadge/);
+  assert.match(page, /platform-logo/);
+  assert.match(page, /logoOnly/);
+  assert.match(page, /platform-video-thumb/);
+  assert.match(styles, /\.platform-logo\.xiaohongshu/);
+  assert.match(styles, /\.platform-logo\.douyin/);
+  assert.match(styles, /\.platform-badge\.logo-only/);
  assert.match(page, /批量保存图片/);
+  assert.match(page, /导出Excel/);
+  assert.match(page, /上传回填表/);
+  assert.match(page, /\/api\/partner-batch-workbook/);
+  assert.match(page, /compactPartnerBatchWorkbookForUpload/);
+  assert.match(page, /Content-Type/);
+  assert.match(page, /response\.status === 413/);
  assert.match(page, /navigator\.share/);
  assert.match(page, /zipSync/);
  assert.match(page, /navigator\.clipboard\.writeText/);
@@ -53,7 +70,9 @@ test("keeps claiming minimal and backfill one-to-one", async () => {
  assert.match(page, /找回领取记录/);
  assert.match(page, /action:\s*"recover"/);
  assert.match(page, /同一任务多次领取会分批展示/);
-  assert.match(page, /这篇笔记已回填，不会与其他笔记错配/);
+  assert.match(page, /这篇笔记的发布记录回填成功啦～/);
+  assert.match(page, /toastClassName\(toast\)/);
+  assert.match(styles, /\.toast\.success/);
  assert.match(page, /笔记内容已收起/);
  assert.match(page, /展开笔记内容/);
  assert.match(page, /收起笔记内容/);
@@ -61,7 +80,7 @@ test("keeps claiming minimal and backfill one-to-one", async () => {
  assert.match(page, /scrollIntoView/);
  assert.match(page, /matchMedia\("\(max-width: 620px\)"\)/);
  assert.match(styles, /\.note-document\.mobile-collapsed/);
-  assert.doesNotMatch(page, /批量回填/);
+  assert.match(page, /批量回填/);
  assert.doesNotMatch(page, /复制标题和正文/);
  assert.match(page, /CONFIGURED_ADMIN_ORIGIN/);
  assert.match(page, /window\.location\.origin/);
@@ -75,6 +94,7 @@ test("keeps claiming minimal and backfill one-to-one", async () => {
  assert.match(page, /creatorViews/);
  assert.match(page, /截图仅用于运营核对，不再自动OCR/);
  assert.match(page, /evidenceImageUrl/);
+  assert.match(page, /params\.set\("v", evidenceKey\)/);
  assert.match(page, /evidenceImageUrl\(selected, "publish"\)/);
  assert.match(page, /evidenceImageUrl\(selected, "creator"\)/);
  assert.match(page, /ImageLightbox/);
@@ -83,6 +103,8 @@ test("keeps claiming minimal and backfill one-to-one", async () => {
  assert.match(page, /creatorScreenshotPreview/);
  assert.match(page, /曝光量/);
  assert.match(page, /阅读量/);
+  assert.match(page, /placeholder="填写截图中的曝光量"\s*\/>/);
+  assert.match(page, /placeholder="填写截图中的阅读量"\s*\/>/);
  assert.doesNotMatch(page, /recognizeCreatorMetrics/);
  assert.match(page, /note-index \$\{\(isScreenshotTask \? item\.result_submitted_at : item\.publish_url\) \? "done" : ""\}/);
  assert.match(packageJson, /"fflate":\s*"0\.7\.4"/);
--- a/lib/account-enrichment-service.ts
+++ b/lib/account-enrichment-service.ts
@@ -1,7 +1,7 @@
 import {
  resolveXhsPublicAccountDetails,
-  resolveXhsAccountProfileFromMcp,
-  resolveXhsProfileDetailsFromMcp,
+  resolveAccountProfileFromMcp,
+  resolveProfileDetailsFromMcp,
  type CollectionMcpConfig,
 } from "./mcp-collection-client";
 import { hashText } from "./mvp-db";
@@ -11,15 +11,20 @@ type DistributionAccountRow = {
  id: string;
  account_id: string | null;
  publish_url: string | null;
+  platform: string;
+  claimant_contact: string | null;
 };

 type BackfillRow = DistributionAccountRow & {
+  resolved_account_id: string | null;
  nickname: string | null;
-  platform: string | null;
  platform_uid: string | null;
  public_account_id: string | null;
  profile_url: string | null;
  followers: number | null;
+  gender: string | null;
+  bio: string | null;
+  tags: string | null;
 };

 function isVerifiedXhsProfileUrl(value: string | null) {
@@ -37,6 +42,23 @@ function isVerifiedXhsProfileUrl(value: string | null) {
  }
 }

+function isVerifiedDouyinProfileUrl(value: string | null) {
+  if (!value) return false;
+  try {
+    const url = new URL(value);
+    const secUid = decodeURIComponent(url.pathname.match(/^\/user\/([^/?#]+)/)?.[1] ?? "");
+    return (
+      url.protocol === "https:" &&
+      (url.hostname === "douyin.com" ||
+        url.hostname.endsWith(".douyin.com")) &&
+      /^[A-Za-z0-9_-]{20,220}$/.test(secUid) &&
+      !/^\d+$/.test(secUid)
+    );
+  } catch {
+    return false;
+  }
+}
+
 export async function enrichDistributionAccount(
  db: DatabaseClient,
  distributionId: string,
@@ -44,27 +66,42 @@ export async function enrichDistributionAccount(
  fallbackNickname: string,
  mcpConfig: CollectionMcpConfig,
 ) {
-  const profile = await resolveXhsAccountProfileFromMcp(
-    publishUrl,
-    fallbackNickname,
-    mcpConfig,
-  );
  const current = await db
    .prepare(
-      `SELECT id, account_id, publish_url
-       FROM distributions
-       WHERE id = ?`,
+      `SELECT d.id, d.account_id, d.publish_url, t.platform,
+              cl.claimant_name AS claimant_contact
+       FROM distributions d
+       JOIN tasks t ON t.id = d.task_id
+       LEFT JOIN claims cl ON cl.id = d.claim_id
+       WHERE d.id = ?`,
    )
    .bind(distributionId)
    .first<DistributionAccountRow>();
  if (!current || current.publish_url !== publishUrl) {
    return { updated: false, reason: "stale" as const };
  }
+  const platform = current.platform === "抖音" ? "抖音" : "小红书";
+  const profile = await resolveAccountProfileFromMcp(
+    publishUrl,
+    fallbackNickname,
+    platform,
+    mcpConfig,
+  );

  const canonicalAccountId = `account-${hashText(
-    `小红书:${profile.platformUid}`,
+    `${platform}:${profile.platformUid}`,
  )}`;
-  if (current.account_id === canonicalAccountId) {
+  const existingAccount = await db
+    .prepare(
+      `SELECT id FROM accounts
+       WHERE platform = ? AND platform_uid = ?
+       LIMIT 1`,
+    )
+    .bind(platform, profile.platformUid)
+    .first<{ id: string }>();
+  const targetAccountId = existingAccount?.id || canonicalAccountId;
+  const currentContact = (current.claimant_contact || "").trim();
+  if (current.account_id === targetAccountId) {
    await db
      .prepare(
        `UPDATE accounts SET
@@ -82,6 +119,12 @@ export async function enrichDistributionAccount(
            WHEN ? IS NOT NULL AND (? > 0 OR followers = 0) THEN ?
            ELSE followers
          END,
+          gender = CASE WHEN ? != '' THEN ? ELSE gender END,
+          bio = CASE WHEN ? != '' THEN ? ELSE bio END,
+          current_contact = CASE WHEN ? != '' THEN ? ELSE current_contact END,
+          post_count = (
+            SELECT COUNT(*) FROM distributions WHERE account_id = ?
+          ),
          last_seen_at = CURRENT_TIMESTAMP
         WHERE id = ?`,
      )
@@ -96,12 +139,19 @@ export async function enrichDistributionAccount(
        profile.followers,
        profile.followers,
        profile.followers,
-        canonicalAccountId,
+        profile.gender,
+        profile.gender,
+        profile.bio,
+        profile.bio,
+        currentContact,
+        currentContact,
+        targetAccountId,
+        targetAccountId,
      )
      .run();
    return {
      updated: true,
-      accountId: canonicalAccountId,
+      accountId: targetAccountId,
      profileUrl: profile.profileUrl,
    };
  }
@@ -111,8 +161,9 @@ export async function enrichDistributionAccount(
    db
      .prepare(
        `INSERT INTO accounts
-          (id, platform, platform_uid, public_account_id, nickname, profile_url, ip_location, followers, post_count)
-         VALUES (?, '小红书', ?, ?, ?, ?, ?, ?, 1)
+          (id, platform, platform_uid, public_account_id, nickname, profile_url,
+           ip_location, followers, gender, bio, current_contact, post_count)
+         VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, 1)
         ON CONFLICT(platform, platform_uid) DO UPDATE SET
           public_account_id = CASE
             WHEN excluded.public_account_id != ''
@@ -131,17 +182,29 @@ export async function enrichDistributionAccount(
               THEN excluded.followers
             ELSE accounts.followers
           END,
-           post_count = accounts.post_count + 1,
+           gender = CASE
+             WHEN excluded.gender != '' THEN excluded.gender ELSE accounts.gender END,
+           bio = CASE
+             WHEN excluded.bio != '' THEN excluded.bio ELSE accounts.bio END,
+           current_contact = CASE
+             WHEN excluded.current_contact != ''
+               THEN excluded.current_contact
+             ELSE accounts.current_contact
+           END,
           last_seen_at = CURRENT_TIMESTAMP`,
      )
      .bind(
-        canonicalAccountId,
+        targetAccountId,
+        platform,
        profile.platformUid,
        profile.redId,
        profile.nickname || fallbackNickname,
        profile.profileUrl,
        profile.ipLocation,
        profile.followers ?? 0,
+        profile.gender,
+        profile.bio,
+        currentContact,
      ),
    db
      .prepare(
@@ -150,10 +213,25 @@ export async function enrichDistributionAccount(
          updated_at = CURRENT_TIMESTAMP
         WHERE id = ? AND publish_url = ?`,
      )
-      .bind(canonicalAccountId, distributionId, publishUrl),
+      .bind(targetAccountId, distributionId, publishUrl),
+    db
+      .prepare(
+        `UPDATE accounts SET post_count = (
+           SELECT COUNT(*) FROM distributions WHERE account_id = ?
+         ) WHERE id = ?`,
+      )
+      .bind(targetAccountId, targetAccountId),
  ]);

-  if (provisionalAccountId) {
+  if (provisionalAccountId && provisionalAccountId !== targetAccountId) {
+    await db
+      .prepare(
+        `UPDATE accounts SET post_count = (
+           SELECT COUNT(*) FROM distributions WHERE account_id = ?
+         ) WHERE id = ?`,
+      )
+      .bind(provisionalAccountId, provisionalAccountId)
+      .run();
    await db
      .prepare(
        `DELETE FROM accounts
@@ -165,14 +243,14 @@ export async function enrichDistributionAccount(
      )
      .bind(
        provisionalAccountId,
-        canonicalAccountId,
+        targetAccountId,
        provisionalAccountId,
      )
      .run();
  }
  return {
    updated: true,
-    accountId: canonicalAccountId,
+    accountId: targetAccountId,
    profileUrl: profile.profileUrl,
  };
 }
@@ -188,15 +266,22 @@ export async function backfillAccountProfiles(
        d.id,
        d.account_id,
        d.publish_url,
+        a.id AS resolved_account_id,
        a.nickname,
-        a.platform,
+        COALESCE(a.platform, t.platform) AS platform,
        a.platform_uid,
        a.public_account_id,
        a.profile_url,
-        a.followers
+        a.followers,
+        a.gender,
+        a.bio,
+        a.tags,
+        cl.claimant_name AS claimant_contact
       FROM distributions d
+       JOIN tasks t ON t.id = d.task_id
       LEFT JOIN accounts a ON a.id = d.account_id
-       WHERE d.publish_url IS NOT NULL
+       LEFT JOIN claims cl ON cl.id = d.claim_id
+      WHERE d.publish_url IS NOT NULL
         AND d.publish_url != ''
       ORDER BY d.updated_at DESC
       LIMIT 100`,
@@ -205,8 +290,34 @@ export async function backfillAccountProfiles(
  let attempted = 0;
  let updated = 0;
  let failed = 0;
+  const backfilledAccounts = new Set<string>();

  for (const row of rows.results) {
+    if (
+      row.resolved_account_id &&
+      !backfilledAccounts.has(row.resolved_account_id)
+    ) {
+      backfilledAccounts.add(row.resolved_account_id);
+      const claimantContact = row.claimant_contact?.trim() || "";
+      await db
+        .prepare(
+          `UPDATE accounts
+           SET current_contact = CASE
+                 WHEN ? != '' THEN ? ELSE current_contact
+               END,
+               post_count = (
+                 SELECT COUNT(*) FROM distributions WHERE account_id = ?
+               )
+           WHERE id = ?`,
+        )
+        .bind(
+          claimantContact,
+          claimantContact,
+          row.resolved_account_id,
+          row.resolved_account_id,
+        )
+        .run();
+    }
    if (attempted >= Math.max(1, Math.min(25, limit))) break;
    const noteId = (() => {
      try {
@@ -227,19 +338,31 @@ export async function backfillAccountProfiles(
      row.platform === "小红书" &&
      !isDemoAccount &&
      isVerifiedXhsProfileUrl(row.profile_url) &&
-      (!row.public_account_id || Number(row.followers ?? 0) === 0)
+      (!row.public_account_id ||
+        Number(row.followers ?? 0) === 0 ||
+        !row.gender ||
+        !row.bio)
    ) {
      attempted += 1;
      attemptedThisRow = true;
-      const details = await resolveXhsProfileDetailsFromMcp(
+      const details = await resolveProfileDetailsFromMcp(
        row.profile_url ?? "",
+        "小红书",
        mcpConfig,
-      ).catch(() =>
-        resolveXhsPublicAccountDetails(row.profile_url ?? ""),
-      );
+      ).catch(async () => ({
+        ...(await resolveXhsPublicAccountDetails(row.profile_url ?? "")),
+        gender: "" as const,
+        bio: "",
+        recentNoteTitles: [] as string[],
+        providerTags: [] as string[],
+      }));
      if (
        row.account_id &&
-        (details.redId || details.followers !== null)
+        (details.redId ||
+          details.followers !== null ||
+          details.gender ||
+          details.bio ||
+          details.recentNoteTitles.length > 0)
      ) {
        await db
          .prepare(
@@ -256,6 +379,8 @@ export async function backfillAccountProfiles(
                   WHEN ? != '' AND ? != '待识别' THEN ?
                   ELSE ip_location
                 END,
+                 gender = CASE WHEN ? != '' THEN ? ELSE gender END,
+                 bio = CASE WHEN ? != '' THEN ? ELSE bio END,
                 last_seen_at = CURRENT_TIMESTAMP
             WHERE id = ?`,
          )
@@ -268,6 +393,10 @@ export async function backfillAccountProfiles(
            details.ipLocation ?? "",
            details.ipLocation ?? "",
            details.ipLocation ?? "",
+            details.gender,
+            details.gender,
+            details.bio,
+            details.bio,
            row.account_id,
          )
          .run();
@@ -287,12 +416,17 @@ export async function backfillAccountProfiles(
    }
    const needsEnrichment =
      !isDemoAccount &&
-      (!row.account_id ||
+      (!row.resolved_account_id ||
      (row.platform === "小红书" &&
        !isVerifiedXhsProfileUrl(row.profile_url)) ||
+      (row.platform === "抖音" &&
+        !isVerifiedDouyinProfileUrl(row.profile_url)) ||
      (row.platform === "小红书" && !row.public_account_id) ||
+      (row.platform === "抖音" && !row.public_account_id) ||
      (row.platform === "小红书" &&
        Number(row.followers ?? 0) === 0) ||
+      (row.platform === "抖音" &&
+        Number(row.followers ?? 0) === 0) ||
      row.platform_uid?.startsWith("pending-") ||
      Boolean(noteId && row.platform_uid === noteId));
    if (!needsEnrichment || !row.publish_url) {
--- a/lib/collection-service.ts
+++ b/lib/collection-service.ts
@@ -1,5 +1,5 @@
 import {
-  collectXhsMetricsFromMcp,
+  collectMetricsFromMcp,
  type CollectionMcpConfig,
 } from "./mcp-collection-client";
 import { uid } from "./mvp-db";
@@ -10,6 +10,7 @@ type DistributionForCollection = {
  task_id: string;
  publish_url: string | null;
  ocr_status: string;
+  platform: string;
 };

 type ScheduledTask = {
@@ -165,11 +166,16 @@ export async function collectDistributionMetrics(
  mcpConfig: CollectionMcpConfig,
 ) {
  const current = await db
-    .prepare("SELECT * FROM distributions WHERE id = ?")
+    .prepare(
+      `SELECT d.*, t.platform
+       FROM distributions d
+       JOIN tasks t ON t.id = d.task_id
+       WHERE d.id = ?`,
+    )
    .bind(distributionId)
    .first<DistributionForCollection>();
  if (!current) throw new Error("分发记录不存在");
-  if (!current.publish_url) throw new Error("笔记尚未回填发布链接");
+  if (!current.publish_url) throw new Error("作品尚未回填发布链接");

  const scheduledAt = `${scheduledDate}T09:00:00+08:00`;
  const runId = uid("run");
@@ -203,7 +209,12 @@ export async function collectDistributionMetrics(
    .bind(distributionId, scheduledDate)
    .first<{ id: string; status: string }>();
  if (!run) throw new Error("采集任务创建失败");
-  if (run.status === "success") return { skipped: true };
+  // Scheduled jobs should remain idempotent, but an operator clicking
+  // “立即采集” is explicitly asking for a fresh snapshot. Reusing the same
+  // daily run lets us correct stale or previously mis-mapped platform data.
+  if (run.status === "success" && source !== "manual") {
+    return { skipped: true };
+  }

  const collectingDescription =
    source === "automatic"
@@ -235,8 +246,12 @@ export async function collectDistributionMetrics(
  ]);

  try {
-    const { likes, comments, collects } =
-      await collectXhsMetricsFromMcp(current.publish_url, mcpConfig);
+    const { likes, comments, collects, shares } =
+      await collectMetricsFromMcp(
+        current.publish_url,
+        current.platform === "抖音" ? "抖音" : "小红书",
+        mcpConfig,
+      );
    const dayWeight = scheduleDay ?? 1;
    const successDescription =
      source === "automatic"
@@ -253,6 +268,7 @@ export async function collectDistributionMetrics(
               likes = ?,
               comments = ?,
               collects = ?,
+               shares = ?,
               status_description = ?,
               completed_at = CURRENT_TIMESTAMP
           WHERE id = ?`,
@@ -261,6 +277,7 @@ export async function collectDistributionMetrics(
          likes,
          comments,
          collects,
+          shares,
          successDescription,
          run.id,
        ),
@@ -270,6 +287,7 @@ export async function collectDistributionMetrics(
           SET latest_likes = ?,
               latest_comments = ?,
               latest_collects = ?,
+               latest_shares = ?,
               collection_status = 'success',
               collection_status_description = ?,
               collection_updated_at = CURRENT_TIMESTAMP,
@@ -285,12 +303,13 @@ export async function collectDistributionMetrics(
          likes,
          comments,
          collects,
+          shares,
          successDescription,
          scheduleDay,
          distributionId,
        ),
    ]);
-    return { skipped: false, likes, comments, collects };
+    return { skipped: false, likes, comments, collects, shares };
  } catch (error) {
    const message =
      error instanceof Error ? error.message : "公开数据采集失败";
--- a/lib/feishu-client.ts
+++ b/lib/feishu-client.ts
@@ -2,7 +2,7 @@ const FEISHU_API_ORIGIN = "https://open.feishu.cn";
 const MAX_SHEET_ROWS = 5_000;
 const MAX_SHEET_COLUMNS = 100;
 const MAX_CONTENT_ROWS = 1_000;
-const MAX_MEDIA_BYTES = 20_000_000;
+const DEFAULT_MAX_MEDIA_BYTES = 200_000_000;

 export type FeishuBindings = {
  FEISHU_APP_ID?: string;
@@ -16,11 +16,20 @@ export type FeishuSourceImage = {
  height: number | null;
 };

+export type FeishuSourceVideo = {
+  index: number;
+  fileToken: string;
+  name: string;
+  mimeType: string;
+  size: number | null;
+};
+
 export type FeishuSourceRow = {
  sourceRow: number;
  title: string;
  body: string;
  images: FeishuSourceImage[];
+  videos: FeishuSourceVideo[];
 };

 export type FeishuSource = {
@@ -106,12 +115,47 @@ function cellText(value: unknown): string {
      .filter(Boolean)
      .join("");
  }
-  if (!isRecord(value) || value.type === "embed-image") return "";
+  if (
+    !isRecord(value) ||
+    value.type === "embed-image" ||
+    value.type === "attachment"
+  ) return "";
  if (typeof value.text === "string") return value.text.trim();
  if (typeof value.value === "string") return value.value.trim();
  return "";
 }

+function extractVideos(value: unknown, output: FeishuSourceVideo[]) {
+  if (Array.isArray(value)) {
+    for (const item of value) extractVideos(item, output);
+    return;
+  }
+  if (!isRecord(value)) return;
+  const fileToken = bindingValue(value.fileToken ?? value.file_token);
+  const mimeType = bindingValue(value.mimeType ?? value.mime_type);
+  const name = bindingValue(value.text ?? value.name ?? value.file_name);
+  const isVideo =
+    value.type === "attachment" &&
+    (mimeType.startsWith("video/") || /\.(?:mp4|mov|m4v|webm)$/i.test(name));
+  if (isVideo && fileToken) {
+    output.push({
+      index: 0,
+      fileToken,
+      name: name || "视频",
+      mimeType: mimeType || "video/mp4",
+      size:
+        typeof value.size === "number" && Number.isFinite(value.size)
+          ? value.size
+          : null,
+    });
+  }
+  for (const child of Object.values(value)) {
+    if (child !== value.fileToken && child !== value.file_token) {
+      extractVideos(child, output);
+    }
+  }
+}
+
 function extractImages(value: unknown, output: FeishuSourceImage[]) {
  if (Array.isArray(value)) {
    for (const item of value) extractImages(item, output);
@@ -167,6 +211,7 @@ function findHeader(values: unknown[][]) {
        titleIndex: number;
        bodyIndex: number;
        tagsIndex: number;
+        videoIndex: number;
        score: number;
      }
    | undefined;
@@ -185,9 +230,13 @@ function findHeader(values: unknown[][]) {
    const tagsIndex = headers.findIndex((header) =>
      headerMatches(header, [/标签/, /话题/, /^tags?$/]),
    );
+    const videoIndex = headers.findIndex((header) =>
+      headerMatches(header, [/^视频\d*$/, /视频文件/, /视频素材/]),
+    );
    const score =
      (titleIndex >= 0 ? 5 : 0) +
      (bodyIndex >= 0 ? 5 : 0) +
+      (videoIndex >= 0 ? 2 : 0) +
      (idIndex >= 0 ? 1 : 0) +
      (tagsIndex >= 0 ? 1 : 0);
    if (!best || score > best.score) {
@@ -197,6 +246,7 @@ function findHeader(values: unknown[][]) {
        titleIndex,
        bodyIndex,
        tagsIndex,
+        videoIndex,
        score,
      };
    }
@@ -217,6 +267,7 @@ function parseRows(values: unknown[][]) {
  const usedSourceRows = new Set<number>();
  const rows: FeishuSourceRow[] = [];
  let maxImageCount = 0;
+  let maxVideoCount = 0;

  for (
    let rowIndex = header.rowIndex + 1;
@@ -253,7 +304,20 @@ function parseRows(values: unknown[][]) {
      })
      .map((image, imageIndex) => ({ ...image, index: imageIndex + 1 }));
    maxImageCount = Math.max(maxImageCount, images.length);
-    rows.push({ sourceRow, title, body, images });
+    const collectedVideos: FeishuSourceVideo[] = [];
+    if (header.videoIndex >= 0) {
+      extractVideos(row[header.videoIndex], collectedVideos);
+    }
+    const seenVideoTokens = new Set<string>();
+    const videos = collectedVideos
+      .filter((video) => {
+        if (seenVideoTokens.has(video.fileToken)) return false;
+        seenVideoTokens.add(video.fileToken);
+        return true;
+      })
+      .map((video, videoIndex) => ({ ...video, index: videoIndex + 1 }));
+    maxVideoCount = Math.max(maxVideoCount, videos.length);
+    rows.push({ sourceRow, title, body, images, videos });
  }

  if (rows.length === 0) {
@@ -266,6 +330,7 @@ function parseRows(values: unknown[][]) {
    header.tagsIndex >= 0 ? cellText(headerRow[header.tagsIndex]) : "",
    cellText(headerRow[header.bodyIndex]) || "正文",
    ...Array.from({ length: maxImageCount }, (_, index) => `图片${index + 1}`),
+    ...Array.from({ length: maxVideoCount }, (_, index) => `视频${index + 1}`),
  ].filter(Boolean);

  return {
@@ -515,34 +580,37 @@ export async function downloadFeishuMedia(
  fileToken: string,
  bindings: FeishuBindings,
  fetchImpl: FetchLike = fetch,
+  options: { maxBytes?: number; label?: string; timeoutMs?: number } = {},
 ) {
+  const maxBytes = Math.max(1, options.maxBytes ?? DEFAULT_MAX_MEDIA_BYTES);
+  const label = bindingValue(options.label) || "素材";
  const normalizedToken = bindingValue(fileToken);
  if (!/^[A-Za-z0-9_-]{8,240}$/.test(normalizedToken)) {
-    throw new FeishuSourceError("飞书图片标识无效", 400);
+    throw new FeishuSourceError(`飞书${label}标识无效`, 400);
  }
  const token = await accessToken(bindings, fetchImpl);
  const response = await fetchImpl(
    `${FEISHU_API_ORIGIN}/open-apis/drive/v1/medias/${encodeURIComponent(normalizedToken)}/download`,
    {
      headers: { Authorization: `Bearer ${token}` },
-      signal: AbortSignal.timeout(20_000),
+      signal: AbortSignal.timeout(options.timeoutMs ?? 120_000),
    },
  );
  const declaredSize = Number(response.headers.get("content-length"));
-  if (Number.isFinite(declaredSize) && declaredSize > MAX_MEDIA_BYTES) {
-    throw new FeishuSourceError("飞书图片超过20MB，无法同步", 413);
+  if (Number.isFinite(declaredSize) && declaredSize > maxBytes) {
+    throw new FeishuSourceError(`飞书${label}文件过大，无法同步`, 413);
  }
  if (!response.ok) {
    throw new FeishuSourceError(
      response.status === 403
-        ? "飞书应用没有这张图片的下载权限"
-        : `下载飞书图片失败（HTTP ${response.status}）`,
+        ? `飞书应用没有这个${label}的下载权限`
+        : `下载飞书${label}失败（HTTP ${response.status}）`,
      response.status === 403 ? 403 : 502,
    );
  }
  const bytes = await response.arrayBuffer();
-  if (bytes.byteLength > MAX_MEDIA_BYTES) {
-    throw new FeishuSourceError("飞书图片超过20MB，无法同步", 413);
+  if (bytes.byteLength > maxBytes) {
+    throw new FeishuSourceError(`飞书${label}文件过大，无法同步`, 413);
  }
  return {
    bytes,
--- a/lib/mcp-collection-client.ts
+++ b/lib/mcp-collection-client.ts
@@ -17,6 +17,7 @@ export type XhsPublicMetrics = {
  likes: number;
  comments: number;
  collects: number;
+  shares: number;
 };

 export type XhsAccountProfile = {
@@ -26,6 +27,10 @@ export type XhsAccountProfile = {
  redId: string;
  ipLocation: string;
  followers: number | null;
+  gender: "" | "男" | "女";
+  bio: string;
+  recentNoteTitles: string[];
+  providerTags: string[];
 };

 type JsonRpcEnvelope = {
@@ -341,16 +346,118 @@ function metricsFromToolResult(result: ToolResult): XhsPublicMetrics {
  }
  if (!data) throw new Error("采集结果缺少互动数据");

+  const count = (value: unknown, label: string) =>
+    value === null || value === undefined || value === ""
+      ? 0
+      : metricValue(value, label);
  return {
-    likes: metricValue(data.likes, "点赞数"),
-    comments: metricValue(data.comments, "评论数"),
-    collects: metricValue(
-      data.collects ?? data.favorites ?? data.favourites,
+    likes: count(
+      data.likes ??
+        data.liked_count ??
+        data.likedCount ??
+        data.like_count ??
+        data.likeCount ??
+        data.digg_count ??
+        data.diggCount,
+      "点赞数",
+    ),
+    comments: count(
+      data.comments ?? data.comment_count ?? data.commentCount,
+      "评论数",
+    ),
+    collects: count(
+      data.collects ??
+        data.collected_count ??
+        data.collectedCount ??
+        data.favorites ??
+        data.favourites ??
+        data.collect_count ??
+        data.collectCount,
      "收藏数",
    ),
+    shares: count(
+      data.shares ??
+        data.share_count ??
+        data.shareCount ??
+        data.forwards ??
+        data.forward_count ??
+        data.forwardCount,
+      "转发数",
+    ),
  };
 }

+function usableDouyinSecUid(value: unknown) {
+  const candidate = stringValue(value);
+  return candidate && !/^\d+$/.test(candidate) && /^[A-Za-z0-9_-]{20,220}$/.test(candidate)
+    ? candidate
+    : "";
+}
+
+function verifiedDouyinProfileUrl(value: unknown) {
+  const candidate = stringValue(value);
+  if (!candidate) return "";
+  try {
+    const parsed = new URL(candidate);
+    const secUid = decodeURIComponent(parsed.pathname.match(/^\/user\/([^/?#]+)/)?.[1] ?? "");
+    if (
+      parsed.protocol === "https:" &&
+      (parsed.hostname === "douyin.com" || parsed.hostname.endsWith(".douyin.com")) &&
+      usableDouyinSecUid(secUid)
+    ) {
+      return parsed.toString();
+    }
+  } catch {
+    // The public redirect fallback below can still recover the profile URL.
+  }
+  return "";
+}
+
+async function douyinProfileFromPublicRedirect(
+  publishUrl: string,
+  fetchImpl: typeof fetch,
+  timeoutMs: number,
+) {
+  let parsed: URL;
+  try {
+    parsed = new URL(publishUrl);
+  } catch {
+    return null;
+  }
+  if (
+    parsed.protocol !== "https:" ||
+    !(parsed.hostname === "douyin.com" || parsed.hostname.endsWith(".douyin.com"))
+  ) {
+    return null;
+  }
+  try {
+    const response = await fetchImpl(parsed.toString(), {
+      method: "GET",
+      redirect: "manual",
+      signal: AbortSignal.timeout(Math.min(timeoutMs, 15_000)),
+      headers: {
+        "user-agent":
+          "Mozilla/5.0 (iPhone; CPU iPhone OS 18_0 like Mac OS X) AppleWebKit/605.1.15 Mobile/15E148",
+      },
+    });
+    const location = response.headers.get("location");
+    if (!location) return null;
+    const redirectUrl = new URL(location, parsed);
+    const secUid = usableDouyinSecUid(
+      redirectUrl.searchParams.get("sec_uid") ??
+        redirectUrl.searchParams.get("sec_user_id"),
+    );
+    return secUid
+      ? {
+          platformUid: secUid,
+          profileUrl: `https://www.douyin.com/user/${encodeURIComponent(secUid)}`,
+        }
+      : null;
+  } catch {
+    return null;
+  }
+}
+
 function successfulToolData(result: ToolResult, fallbackMessage: string) {
  const root = asRecord(result.payload);
  const response = asRecord(root?.response) ?? root;
@@ -446,6 +553,64 @@ function findValueByKeys(
  return undefined;
 }

+function profileGender(value: unknown): "" | "男" | "女" {
+  if (value === 1) return "男";
+  if (value === 2) return "女";
+  const normalized = String(value ?? "").trim().toLocaleLowerCase("zh-CN");
+  if (["男", "男性", "male", "m", "1"].includes(normalized)) return "男";
+  if (["女", "女性", "female", "f", "2"].includes(normalized)) return "女";
+  return "";
+}
+
+function recentNoteTitlesFromPayload(value: unknown) {
+  const titles: string[] = [];
+  const visit = (current: unknown, depth = 0) => {
+    if (depth > 12 || titles.length >= 20) return;
+    if (Array.isArray(current)) {
+      current.forEach((item) => visit(item, depth + 1));
+      return;
+    }
+    const record = asRecord(current);
+    if (!record) return;
+    const title = stringValue(record.title ?? record.note_title ?? record.noteTitle);
+    if (
+      title &&
+      (record.note_id || record.noteId || record.url || record.cover) &&
+      !titles.includes(title)
+    ) {
+      titles.push(title);
+    }
+    Object.values(record).forEach((child) => visit(child, depth + 1));
+  };
+  visit(value);
+  return titles;
+}
+
+function providerTagsFromUser(value: unknown) {
+  const user = findRecord(value, (record) =>
+    Boolean(
+      record.gender !== undefined ||
+        record.desc !== undefined ||
+        record.signature !== undefined ||
+        record.fansCount !== undefined ||
+        record.fans_count !== undefined,
+    ),
+  );
+  const raw = user?.tags;
+  if (!Array.isArray(raw)) return [];
+  return [
+    ...new Set(
+      raw
+        .map((item) =>
+          typeof item === "string"
+            ? item.trim()
+            : stringValue(asRecord(item)?.name ?? asRecord(item)?.title),
+        )
+        .filter(Boolean),
+    ),
+  ].slice(0, 5);
+}
+
 const FOLLOWER_KEYS = new Set([
  "fans",
  "fans_count",
@@ -529,10 +694,13 @@ async function xhsNoteIdFromShortLink(
  ) {
    return { noteId: "", profile: null };
  }
-  if (
-    url.hostname !== "xhslink.cn" &&
-    !url.hostname.endsWith(".xhslink.cn")
-  ) {
+  const isShortLink =
+    url.hostname === "xhslink.cn" ||
+    url.hostname.endsWith(".xhslink.cn");
+  const isXhsPage =
+    url.hostname === "xiaohongshu.com" ||
+    url.hostname.endsWith(".xiaohongshu.com");
+  if (!isShortLink && !isXhsPage) {
    return { noteId: "", profile: null };
  }

@@ -608,6 +776,10 @@ function accountProfileFromPublicPage(
    redId,
    ipLocation: "待识别",
    followers: null,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
  };
 }

@@ -698,15 +870,45 @@ function profileDetailsFromToolResult(result: ToolResult) {
  const redId =
    findStringByKey(payload, "red_id") ||
    findStringByKey(payload, "redId") ||
+    findStringByKey(payload, "unique_id") ||
+    findStringByKey(payload, "uniqueId") ||
+    findStringByKey(payload, "short_id") ||
+    findStringByKey(payload, "shortId") ||
+    findStringByKey(payload, "douyin_id") ||
    findStringByKey(payload, "userId") ||
    findStringByKey(payload, "user_id");
  const ipLocation =
    findStringByKey(payload, "ip_location") ||
    findStringByKey(payload, "ipLocation");
-  if (followers === null && !nickname && !redId && !ipLocation) {
+  const gender = profileGender(findValueByKeys(payload, new Set(["gender", "sex"])));
+  const bio =
+    findStringByKey(payload, "desc") ||
+    findStringByKey(payload, "description") ||
+    findStringByKey(payload, "signature") ||
+    findStringByKey(payload, "bio");
+  const recentNoteTitles = recentNoteTitlesFromPayload(payload);
+  const providerTags = providerTagsFromUser(payload);
+  if (
+    followers === null &&
+    !nickname &&
+    !redId &&
+    !ipLocation &&
+    !gender &&
+    !bio &&
+    recentNoteTitles.length === 0
+  ) {
    throw new Error("账号主页采集结果缺少可用字段");
  }
-  return { nickname, followers, redId, ipLocation };
+  return {
+    nickname,
+    followers,
+    redId,
+    ipLocation,
+    gender,
+    bio,
+    recentNoteTitles,
+    providerTags,
+  };
 }

 function accountProfileFromToolResult(
@@ -718,14 +920,18 @@ function accountProfileFromToolResult(
    data,
    (record) =>
      Boolean(
-        stringValue(record.user_id ?? record.userid) &&
+        stringValue(record.user_id ?? record.userid ?? record.userId) &&
          (stringValue(record.profile_url) ||
            stringValue(record.nickname ?? record.name)),
      ),
  );
  if (!user) throw new Error("账号主页识别结果缺少作者信息");
-  const platformUid = stringValue(user.user_id ?? user.userid);
-  const candidateProfileUrl = stringValue(user.profile_url);
+  const platformUid = stringValue(
+    user.user_id ?? user.userid ?? user.userId,
+  );
+  const candidateProfileUrl = stringValue(
+    user.profile_url ?? user.profileUrl,
+  );
  let profileUrl = "";
  if (candidateProfileUrl) {
    try {
@@ -755,6 +961,10 @@ function accountProfileFromToolResult(
    redId: stringValue(user.red_id),
    ipLocation: findStringByKey(data, "ip_location") || "待识别",
    followers: followerCountFromPayload(data),
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
  };
 }

@@ -771,18 +981,7 @@ async function completeAccountProfile(
      "parse_xhs_user_summary",
      { url: profile.profileUrl, use_proxy: true },
    ],
-    [
-      "fetch_user_detail",
-      { link: profile.profileUrl, plant: "xhs" },
-    ],
  ] as const) {
-    if (
-      completed.followers !== null &&
-      completed.redId &&
-      completed.ipLocation !== "待识别"
-    ) {
-      break;
-    }
    try {
      const result = await callMcpTool(
        fetchImpl,
@@ -801,6 +1000,16 @@ async function completeAccountProfile(
          details.ipLocation && details.ipLocation !== "待识别"
            ? details.ipLocation
            : completed.ipLocation,
+        gender: details.gender || completed.gender,
+        bio: details.bio || completed.bio,
+        recentNoteTitles:
+          details.recentNoteTitles.length > 0
+            ? details.recentNoteTitles
+            : completed.recentNoteTitles,
+        providerTags:
+          details.providerTags.length > 0
+            ? details.providerTags
+            : completed.providerTags,
      };
    } catch (error) {
      if (error instanceof McpSessionLostError) throw error;
@@ -852,17 +1061,14 @@ async function resolveAccountInSession(
  const endpoint = buildMcpUrl(config);
  const timeoutMs = Math.max(5_000, config.timeoutMs ?? 30_000);
  let noteId = xhsNoteIdFromUrl(publishUrl);
-  let publicPageProfile: XhsAccountProfile | null = null;
-  if (!noteId) {
-    const shortLink = await xhsNoteIdFromShortLink(
-      publishUrl,
-      fallbackNickname,
-      fetchImpl,
-      timeoutMs,
-    );
-    noteId = shortLink.noteId;
-    publicPageProfile = shortLink.profile;
-  }
+  const linkPage = await xhsNoteIdFromShortLink(
+    publishUrl,
+    fallbackNickname,
+    fetchImpl,
+    timeoutMs,
+  );
+  noteId = noteId || linkPage.noteId;
+  const publicPageProfile = linkPage.profile;

  try {
    const sessionId = await createMcpSession(
@@ -870,47 +1076,39 @@ async function resolveAccountInSession(
      endpoint,
      timeoutMs,
    );
-    if (!noteId) {
-      const noteResult = await callMcpTool(
-        fetchImpl,
-        endpoint,
-        sessionId,
-        timeoutMs,
-        "fetch_content_detail",
-        {
-          link: publishUrl,
-          plant: "xhs",
-          include_comments: false,
-          auto_cookie: true,
-        },
-      );
-      const noteData = successfulToolData(
-        noteResult,
-        "无法识别小红书笔记",
-      );
-      noteId =
-        stringValue(noteData.noteId ?? noteData.note_id) ||
-        findStringByKey(noteData, "noteId") ||
-        findStringByKey(noteData, "note_id");
-    }
-    if (!noteId) throw new Error("无法从发布链接识别小红书笔记ID");
-
-    const authorResult = await callMcpTool(
+    const noteResult = await callMcpTool(
      fetchImpl,
      endpoint,
      sessionId,
      timeoutMs,
-      "collect_xhs_wen_note_detail",
+      "fetch_content_detail",
      {
-        note_id: noteId,
-        need_desc: false,
-        include_raw: false,
+        link: publishUrl,
+        plant: "xhs",
+        include_comments: false,
+        auto_cookie: true,
      },
    );
-    const profile = accountProfileFromToolResult(
-      authorResult,
-      fallbackNickname,
+    const noteData = successfulToolData(
+      noteResult,
+      "无法识别小红书笔记",
    );
+    noteId =
+      noteId ||
+      stringValue(noteData.noteId ?? noteData.note_id) ||
+      findStringByKey(noteData, "noteId") ||
+      findStringByKey(noteData, "note_id");
+    if (!noteId) throw new Error("无法从发布链接识别小红书笔记ID");
+    let profile: XhsAccountProfile;
+    try {
+      profile = accountProfileFromToolResult(
+        noteResult,
+        fallbackNickname,
+      );
+    } catch {
+      if (!publicPageProfile) throw new Error("笔记数据缺少公开作者主页");
+      profile = publicPageProfile;
+    }
    return completeAccountProfile(
      profile,
      fetchImpl,
@@ -937,6 +1135,7 @@ async function resolveAccountInSession(

 async function collectInSession(
  publishUrl: string,
+  platform: "小红书" | "抖音",
  config: CollectionMcpConfig,
  fetchImpl: typeof fetch,
 ) {
@@ -952,28 +1151,12 @@ async function collectInSession(
    "fetch_content_detail",
    {
      link: publishUrl,
-      plant: "xhs",
+      plant: platform === "抖音" ? "dy" : "xhs",
      include_comments: false,
      auto_cookie: true,
    },
  );
-  try {
-    return metricsFromToolResult(primary);
-  } catch {
-    const fallback = await callMcpTool(
-      fetchImpl,
-      endpoint,
-      sessionId,
-      timeoutMs,
-      "parse_xhs_note",
-      {
-        url: publishUrl,
-        include_comments: false,
-        auto_cookie: true,
-      },
-    );
-    return metricsFromToolResult(fallback);
-  }
+  return metricsFromToolResult(primary);
 }

 export function resolveCollectionMcpConfig(
@@ -1005,7 +1188,7 @@ export async function collectXhsMetricsFromMcp(
  let lastError: unknown;
  for (let attempt = 0; attempt < MAX_MCP_ATTEMPTS; attempt += 1) {
    try {
-      return await collectInSession(parsed.toString(), config, fetchImpl);
+      return await collectInSession(parsed.toString(), "小红书", config, fetchImpl);
    } catch (error) {
      lastError = error;
      if (!isRetryableTransportError(error)) throw error;
@@ -1022,6 +1205,211 @@ export async function collectXhsMetricsFromMcp(
    : new Error("MCP采集服务暂时不可用");
 }

+export async function collectMetricsFromMcp(
+  publishUrl: string,
+  platform: "小红书" | "抖音",
+  config: CollectionMcpConfig,
+  fetchImpl: typeof fetch = fetch,
+) {
+  if (platform === "小红书") {
+    return collectXhsMetricsFromMcp(publishUrl, config, fetchImpl);
+  }
+  let parsed: URL;
+  try {
+    parsed = new URL(publishUrl);
+  } catch {
+    throw new Error("发布链接无效");
+  }
+  if (!["http:", "https:"].includes(parsed.protocol)) {
+    throw new Error("发布链接无效");
+  }
+  let lastError: unknown;
+  for (let attempt = 0; attempt < MAX_MCP_ATTEMPTS; attempt += 1) {
+    try {
+      return await collectInSession(parsed.toString(), platform, config, fetchImpl);
+    } catch (error) {
+      lastError = error;
+      if (!isRetryableTransportError(error)) throw error;
+    }
+  }
+  throw lastError instanceof Error
+    ? lastError
+    : new Error("MCP采集服务暂时不可用");
+}
+
+function douyinProfileFromToolResult(
+  result: ToolResult,
+  fallbackNickname: string,
+): XhsAccountProfile {
+  const data = successfulToolData(result, "无法识别抖音作品");
+  const author = findRecord(data, (record) =>
+    Boolean(
+      stringValue(
+        record.sec_uid ?? record.secUid ?? record.uid ?? record.user_id ?? record.userId,
+      ) && stringValue(record.nickname ?? record.name ?? record.unique_id ?? record.uniqueId),
+    ),
+  );
+  if (!author) throw new Error("抖音作品数据缺少作者信息");
+  const verifiedSecUid = usableDouyinSecUid(author.sec_uid ?? author.secUid);
+  const fallbackUid = stringValue(author.uid ?? author.user_id ?? author.userId);
+  const platformUid = verifiedSecUid || fallbackUid;
+  const publicId = stringValue(
+    author.unique_id ?? author.uniqueId ?? author.short_id ?? author.shortId ?? author.douyin_id,
+  );
+  const candidateProfileUrl = verifiedDouyinProfileUrl(
+    author.profile_url ?? author.profileUrl,
+  );
+  const profileUrl = candidateProfileUrl ||
+    (verifiedSecUid
+      ? `https://www.douyin.com/user/${encodeURIComponent(verifiedSecUid)}`
+      : "");
+  return {
+    platformUid,
+    nickname: stringValue(author.nickname ?? author.name) || fallbackNickname.trim(),
+    profileUrl,
+    redId: publicId,
+    ipLocation:
+      stringValue(author.ip_location ?? author.ipLocation) ||
+      findStringByKey(data, "ip_location") ||
+      findStringByKey(data, "ipLocation") ||
+      "待识别",
+    followers: followerCountFromPayload(author) ?? followerCountFromPayload(data),
+    gender: profileGender(author.gender ?? author.sex),
+    bio: stringValue(author.desc ?? author.description ?? author.signature ?? author.bio),
+    recentNoteTitles: recentNoteTitlesFromPayload(data),
+    providerTags: providerTagsFromUser(data),
+  };
+}
+
+async function resolveDouyinAccountInSession(
+  publishUrl: string,
+  fallbackNickname: string,
+  config: CollectionMcpConfig,
+  fetchImpl: typeof fetch,
+) {
+  const endpoint = buildMcpUrl(config);
+  const timeoutMs = Math.max(5_000, config.timeoutMs ?? 30_000);
+  const sessionId = await createMcpSession(fetchImpl, endpoint, timeoutMs);
+  const result = await callMcpTool(
+    fetchImpl,
+    endpoint,
+    sessionId,
+    timeoutMs,
+    "fetch_content_detail",
+    {
+      link: publishUrl,
+      plant: "dy",
+      include_comments: false,
+      auto_cookie: true,
+    },
+  );
+  let profile = douyinProfileFromToolResult(result, fallbackNickname);
+  if (!verifiedDouyinProfileUrl(profile.profileUrl)) {
+    const resolved = await douyinProfileFromPublicRedirect(
+      publishUrl,
+      fetchImpl,
+      timeoutMs,
+    );
+    if (resolved) {
+      profile = {
+        ...profile,
+        platformUid: resolved.platformUid,
+        profileUrl: resolved.profileUrl,
+      };
+    }
+  }
+  if (!verifiedDouyinProfileUrl(profile.profileUrl)) return profile;
+  try {
+    const detailsResult = await callMcpTool(
+      fetchImpl,
+      endpoint,
+      sessionId,
+      timeoutMs,
+      "parse_dy_user_summary",
+      { url: profile.profileUrl },
+    );
+    const details = profileDetailsFromToolResult(detailsResult);
+    profile = {
+      ...profile,
+      nickname: details.nickname || profile.nickname,
+      redId: details.redId || profile.redId,
+      followers: details.followers ?? profile.followers,
+      ipLocation:
+        details.ipLocation && details.ipLocation !== "待识别"
+          ? details.ipLocation
+          : profile.ipLocation,
+      gender: details.gender || profile.gender,
+      bio: details.bio || profile.bio,
+      recentNoteTitles:
+        details.recentNoteTitles.length > 0
+          ? details.recentNoteTitles
+          : profile.recentNoteTitles,
+      providerTags:
+        details.providerTags.length > 0
+          ? details.providerTags
+          : profile.providerTags,
+    };
+  } catch (error) {
+    if (error instanceof McpSessionLostError) throw error;
+  }
+  return profile;
+}
+
+export async function resolveAccountProfileFromMcp(
+  publishUrl: string,
+  fallbackNickname: string,
+  platform: "小红书" | "抖音",
+  config: CollectionMcpConfig,
+  fetchImpl: typeof fetch = fetch,
+) {
+  if (platform === "小红书") {
+    return resolveXhsAccountProfileFromMcp(
+      publishUrl,
+      fallbackNickname,
+      config,
+      fetchImpl,
+    );
+  }
+  let lastError: unknown;
+  for (let attempt = 0; attempt < MAX_MCP_ATTEMPTS; attempt += 1) {
+    try {
+      return await resolveDouyinAccountInSession(
+        publishUrl,
+        fallbackNickname,
+        config,
+        fetchImpl,
+      );
+    } catch (error) {
+      lastError = error;
+      if (!isRetryableTransportError(error)) throw error;
+    }
+  }
+  throw lastError instanceof Error ? lastError : new Error("抖音账号识别失败");
+}
+
+export async function resolveProfileDetailsFromMcp(
+  profileUrl: string,
+  platform: "小红书" | "抖音",
+  config: CollectionMcpConfig,
+  fetchImpl: typeof fetch = fetch,
+) {
+  if (platform === "小红书") {
+    return resolveXhsProfileDetailsFromMcp(profileUrl, config, fetchImpl);
+  }
+  const endpoint = buildMcpUrl(config);
+  const timeoutMs = Math.max(5_000, config.timeoutMs ?? 30_000);
+  const sessionId = await createMcpSession(fetchImpl, endpoint, timeoutMs);
+  const result = await callMcpTool(
+    fetchImpl,
+    endpoint,
+    sessionId,
+    timeoutMs,
+    "parse_dy_user_summary",
+    { url: profileUrl },
+  );
+  return profileDetailsFromToolResult(result);
+}
+
 export async function resolveXhsAccountProfileFromMcp(
  publishUrl: string,
  fallbackNickname: string,
--- a/lib/mcp-operations.ts
+++ b/lib/mcp-operations.ts
@@ -12,7 +12,7 @@ import {
  type CollectionMcpBindings,
 } from "./mcp-collection-client";
 import { ensureSchema, getRawDb } from "./mvp-db";
-import { extractXhsPublishUrl } from "./publish-url";
+import { extractAnyPublishUrl } from "./publish-url";
 import { buildClaimUrl } from "./task-service";

 export type McpOperationBindings = CollectionMcpBindings & {
@@ -123,9 +123,9 @@ export async function taskGet(taskId: string, portalUrl: string) {
  const [notes, claims, runs] = await Promise.all([
    db
      .prepare(
-        `SELECT c.id AS content_id, c.source_row, c.title, c.body, c.image_assets, c.status AS content_status,
+        `SELECT c.id AS content_id, c.source_row, c.title, c.body, c.image_assets, c.video_assets, c.status AS content_status,
          d.id AS distribution_id, d.status AS distribution_status, d.publish_url, d.publish_time,
-          d.exposure, d.views, d.latest_likes, d.latest_comments, d.latest_collects,
+          d.exposure, d.views, d.latest_likes, d.latest_comments, d.latest_collects, d.latest_shares,
          d.collection_status, d.collection_status_description, d.collection_updated_at,
          a.id AS account_id, a.nickname AS account_nickname, a.profile_url, a.followers,
          p.name AS cooperation_source, cl.claimant_name
@@ -172,12 +172,14 @@ export async function taskGet(taskId: string, portalUrl: string) {
    notes: notes.results.map((row) => ({
      ...row,
      image_assets: parseJsonArray(row.image_assets),
+      video_assets: parseJsonArray(row.video_assets),
      total_interactions:
        row.latest_likes == null
          ? null
          : Number(row.latest_likes) +
            Number(row.latest_comments ?? 0) +
-            Number(row.latest_collects ?? 0),
+            Number(row.latest_collects ?? 0) +
+            Number(row.latest_shares ?? 0),
    })),
    claims: claims.results,
    collection_runs: runs.results,
@@ -220,7 +222,8 @@ export async function recoveryList(
  const [rows, count] = await Promise.all([
    db
      .prepare(
-        `SELECT d.*, t.name AS task_name, c.source_row, c.title,
+        `SELECT d.*, t.name AS task_name, t.platform AS task_platform,
+          t.content_format, c.source_row, c.title,
          a.nickname AS account_nickname, a.profile_url, p.name AS cooperation_source,
          cl.claimant_name ${base}
         ORDER BY COALESCE(d.publish_time, d.claimed_at) DESC LIMIT ${limit} OFFSET ${offset}`,
@@ -241,7 +244,7 @@ export async function recoveryList(
      total_interactions:
        row.latest_likes == null
          ? null
-          : Number(row.latest_likes) + Number(row.latest_comments ?? 0) + Number(row.latest_collects ?? 0),
+          : Number(row.latest_likes) + Number(row.latest_comments ?? 0) + Number(row.latest_collects ?? 0) + Number(row.latest_shares ?? 0),
    })),
  };
 }
@@ -358,9 +361,11 @@ function resourceWhere(input: ResourceFilters) {
  const conditions: string[] = [];
  const bindings: unknown[] = [];
  if (input.query?.trim()) {
-    conditions.push("(a.nickname LIKE ? ESCAPE '\\' OR a.public_account_id LIKE ? ESCAPE '\\')");
+    conditions.push(
+      "(a.nickname LIKE ? ESCAPE '\\' OR a.public_account_id LIKE ? ESCAPE '\\' OR a.current_contact LIKE ? ESCAPE '\\' OR a.tags LIKE ? ESCAPE '\\' OR a.bio LIKE ? ESCAPE '\\')",
+    );
    const pattern = like(input.query.trim());
-    bindings.push(pattern, pattern);
+    bindings.push(pattern, pattern, pattern, pattern, pattern);
  }
  if (input.ipLocation?.trim()) {
    conditions.push("a.ip_location LIKE ? ESCAPE '\\'");
@@ -369,8 +374,12 @@ function resourceWhere(input: ResourceFilters) {
  if (input.cooperationSource?.trim()) {
    conditions.push(
      `(a.cooperation_source LIKE ? ESCAPE '\\' OR
-        EXISTS (SELECT 1 FROM distributions dx JOIN partners px ON px.id = dx.partner_id
-        WHERE dx.account_id = a.id AND px.name LIKE ? ESCAPE '\\'))`,
+        EXISTS (SELECT 1 FROM distributions dx
+        JOIN partners px ON px.id = dx.partner_id
+        LEFT JOIN claims cx ON cx.id = dx.claim_id
+        WHERE dx.account_id = a.id
+          AND (cx.claimant_name IS NULL OR px.name != cx.claimant_name)
+          AND px.name LIKE ? ESCAPE '\\'))`,
    );
    const pattern = like(input.cooperationSource.trim());
    bindings.push(pattern, pattern);
@@ -389,7 +398,12 @@ export async function resourceSearch(input: ResourceFilters) {
  const db = getRawDb();
  const select = `SELECT a.*,
    (SELECT COUNT(*) FROM distributions d WHERE d.account_id = a.id) AS cooperation_count,
-    (SELECT GROUP_CONCAT(DISTINCT p.name) FROM distributions d JOIN partners p ON p.id = d.partner_id WHERE d.account_id = a.id) AS cooperation_sources`;
+    (SELECT GROUP_CONCAT(DISTINCT p.name)
+     FROM distributions d
+     JOIN partners p ON p.id = d.partner_id
+     LEFT JOIN claims c ON c.id = d.claim_id
+     WHERE d.account_id = a.id
+       AND (c.claimant_name IS NULL OR p.name != c.claimant_name)) AS cooperation_sources`;
  const [rows, count] = await Promise.all([
    db.prepare(`${select} FROM accounts a ${where} ORDER BY a.last_seen_at DESC LIMIT ${limit} OFFSET ${offset}`)
      .bind(...bindings).all<Record<string, unknown>>(),
@@ -422,7 +436,7 @@ export async function resourceGet(accountId: string) {
  if (!account) throw new Error("账号不存在");
  const history = await db.prepare(
    `SELECT d.id AS distribution_id, d.task_id, t.name AS task_name, c.title,
-      d.publish_url, d.publish_time, d.latest_likes, d.latest_comments, d.latest_collects,
+      d.publish_url, d.publish_time, d.latest_likes, d.latest_comments, d.latest_collects, d.latest_shares,
      d.exposure, d.views, p.name AS cooperation_source, cl.claimant_name
     FROM distributions d JOIN tasks t ON t.id = d.task_id
     JOIN contents c ON c.id = d.content_id JOIN partners p ON p.id = d.partner_id
@@ -438,7 +452,7 @@ export async function backfillResourceProfile(
 ) {
  await ensureSchema();
  const db = getRawDb();
-  const publishUrl = input.publishUrl ? extractXhsPublishUrl(input.publishUrl) : "";
+  const publishUrl = input.publishUrl ? extractAnyPublishUrl(input.publishUrl) : "";
  const row = input.distributionId
    ? await db.prepare(
        `SELECT d.id, d.publish_url, COALESCE(a.nickname, '') AS nickname
--- a/lib/mcp-tools.ts
+++ b/lib/mcp-tools.ts
@@ -52,7 +52,7 @@ const pagination = {
 };

 const resourceFilters = {
-  query: z.string().max(100).optional().describe("账号名称或小红书号/抖音号，支持模糊搜索"),
+  query: z.string().max(100).optional().describe("账号名称、账号号、简介或标签，支持模糊搜索"),
  ip_location: z.string().max(100).optional().describe("IP地区关键词，支持模糊搜索"),
  cooperation_source: z.string().max(100).optional().describe("历史合作来源关键词，支持模糊搜索"),
  platform: z.string().max(30).optional().describe("平台，例如小红书；不传表示全部"),
@@ -159,7 +159,7 @@ export function registerMcpOperationTools(server: McpServer, options: Options) {
    "collection_collect_now",
    {
      title: "立即采集指定笔记",
-      description: "对指定分发记录立即采集点赞、收藏和评论数据。",
+      description: "对指定分发记录立即采集互动数据；小红书为点赞/收藏/评论，抖音另含转发。",
      inputSchema: z.object({
        distribution_id: z.string().min(1).describe("分发记录ID，可从 task_get 或 recovery_list 获取"),
        schedule_day: z.number().int().min(1).max(7).optional().describe("标记为第几天采集，可不传"),
@@ -190,7 +190,7 @@ export function registerMcpOperationTools(server: McpServer, options: Options) {
    "resource_search",
    {
      title: "搜索 KOC 账号资源",
-      description: "按账号名称/账号号、IP地区、合作来源或平台搜索 KOC 资源。",
+      description: "按账号名称/账号号/标签、IP地区、合作来源或平台搜索 KOC 资源。",
      inputSchema: z.object({ ...resourceFilters, ...pagination }),
      annotations: { readOnlyHint: true, destructiveHint: false, idempotentHint: true },
    },
@@ -204,7 +204,7 @@ export function registerMcpOperationTools(server: McpServer, options: Options) {
    "resource_get",
    {
      title: "查看 KOC 账号详情",
-      description: "查看账号主页、账号号、粉丝数、IP地区以及全部合作记录。",
+      description: "查看账号主页、账号号、粉丝数、性别、简介、标签、IP地区以及全部合作记录。",
      inputSchema: z.object({ account_id: z.string().min(1).describe("KOC LOOP 账号ID") }),
      annotations: { readOnlyHint: true, destructiveHint: false, idempotentHint: true },
    },
@@ -218,10 +218,10 @@ export function registerMcpOperationTools(server: McpServer, options: Options) {
    "resource_backfill_profile",
    {
      title: "补全公开账号信息",
-      description: "根据已回填的发布链接补全账号主页、昵称、小红书号、IP地区和粉丝数。",
+      description: "根据已回填的小红书或抖音作品链接补全账号主页、昵称、账号号、IP地区和粉丝数。",
      inputSchema: z.object({
        distribution_id: z.string().optional().describe("分发记录ID，和发布链接二选一"),
-        publish_url: z.string().optional().describe("小红书发布链接或包含链接的分享文案"),
+        publish_url: z.string().optional().describe("小红书或抖音作品链接，也可传包含链接的分享文案"),
      }).refine((value) => Boolean(value.distribution_id || value.publish_url), "请提供分发记录ID或发布链接"),
      annotations: { readOnlyHint: false, destructiveHint: false, idempotentHint: true, openWorldHint: true },
    },
--- a/lib/mvp-db.ts
+++ b/lib/mvp-db.ts
@@ -94,6 +94,8 @@ export async function ensureSchema(database?: DatabaseClient) {
      due_at TEXT NOT NULL,
      status TEXT NOT NULL DEFAULT 'active',
      task_type TEXT NOT NULL DEFAULT 'content_publish',
+      platform TEXT NOT NULL DEFAULT '小红书',
+      content_format TEXT NOT NULL DEFAULT 'image_text',
      source_url TEXT NOT NULL DEFAULT '',
      source_sheet_id TEXT NOT NULL DEFAULT '',
      source_sheet_name TEXT NOT NULL DEFAULT '',
@@ -110,6 +112,7 @@ export async function ensureSchema(database?: DatabaseClient) {
      title TEXT NOT NULL,
      body TEXT NOT NULL DEFAULT '',
      image_assets TEXT NOT NULL DEFAULT '[]',
+      video_assets TEXT NOT NULL DEFAULT '[]',
      status TEXT NOT NULL DEFAULT 'available',
      source TEXT NOT NULL DEFAULT '飞书内容表',
      source_row INTEGER,
@@ -124,9 +127,13 @@ export async function ensureSchema(database?: DatabaseClient) {
      profile_url TEXT NOT NULL DEFAULT '',
      ip_location TEXT NOT NULL DEFAULT '待识别',
      followers INTEGER NOT NULL DEFAULT 0,
+      gender TEXT NOT NULL DEFAULT '',
+      bio TEXT NOT NULL DEFAULT '',
+      tags TEXT NOT NULL DEFAULT '',
      post_count INTEGER NOT NULL DEFAULT 0,
      avg_views INTEGER NOT NULL DEFAULT 0,
      cooperation_source TEXT NOT NULL DEFAULT '',
+      current_contact TEXT NOT NULL DEFAULT '',
      first_seen_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP,
      last_seen_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP
    )`,
@@ -185,6 +192,7 @@ export async function ensureSchema(database?: DatabaseClient) {
      latest_likes INTEGER,
      latest_comments INTEGER,
      latest_collects INTEGER,
+      latest_shares INTEGER,
      collection_status TEXT NOT NULL DEFAULT 'pending',
      collection_status_description TEXT,
      collection_updated_at TEXT,
@@ -202,6 +210,7 @@ export async function ensureSchema(database?: DatabaseClient) {
      likes INTEGER,
      comments INTEGER,
      collects INTEGER,
+      shares INTEGER,
      status_description TEXT,
      started_at TEXT,
      completed_at TEXT,
@@ -293,6 +302,14 @@ export async function ensureSchema(database?: DatabaseClient) {
    "public_account_id",
    "public_account_id TEXT NOT NULL DEFAULT ''",
  );
+  await ensureColumn("accounts", "gender", "gender TEXT NOT NULL DEFAULT ''");
+  await ensureColumn("accounts", "bio", "bio TEXT NOT NULL DEFAULT ''");
+  await ensureColumn("accounts", "tags", "tags TEXT NOT NULL DEFAULT ''");
+  await ensureColumn(
+    "accounts",
+    "current_contact",
+    "current_contact TEXT NOT NULL DEFAULT ''",
+  );
  await ensureColumn("distributions", "claim_id", "claim_id TEXT");
  await ensureColumn(
    "distributions",
@@ -349,6 +366,11 @@ export async function ensureSchema(database?: DatabaseClient) {
    "last_collection_day",
    "last_collection_day INTEGER",
  );
+  await ensureColumn(
+    "partners",
+    "wecom_external_user_id",
+    "wecom_external_user_id TEXT",
+  );

  const tasksWithoutShare = await db
    .prepare(
@@ -796,22 +818,44 @@ export async function getDashboardData() {
    await Promise.all([
      db.prepare("SELECT * FROM partners ORDER BY created_at DESC").all(),
      db.prepare("SELECT * FROM tasks ORDER BY created_at DESC").all(),
-      db.prepare("SELECT * FROM accounts ORDER BY last_seen_at DESC").all(),
+      db
+        .prepare(
+          `SELECT
+            a.*,
+            COALESCE(
+              (
+                SELECT d.publish_url
+                FROM distributions d
+                WHERE d.account_id = a.id
+                  AND TRIM(COALESCE(d.publish_url, '')) != ''
+                ORDER BY d.updated_at DESC, d.claimed_at DESC
+                LIMIT 1
+              ),
+              ''
+            ) AS latest_publish_url
+          FROM accounts a
+          ORDER BY a.last_seen_at DESC`,
+        )
+        .all(),
      db
        .prepare(
          `SELECT
            d.*,
            c.title AS content_title,
            p.name AS partner_name,
+            cl.claimant_name AS claimant_name,
            a.nickname AS account_nickname,
            a.platform AS account_platform,
            t.name AS task_name,
            t.brand AS task_brand,
            t.task_type AS task_type,
+            t.platform AS task_platform,
+            t.content_format AS content_format,
            t.due_at AS due_at
          FROM distributions d
          JOIN contents c ON c.id = d.content_id
          JOIN partners p ON p.id = d.partner_id
+          LEFT JOIN claims cl ON cl.id = d.claim_id
          JOIN tasks t ON t.id = d.task_id
          LEFT JOIN accounts a ON a.id = d.account_id
          ORDER BY d.updated_at DESC, d.claimed_at DESC`,
--- a/lib/partner-batch-workbook.ts
+++ b/lib/partner-batch-workbook.ts
@@ -0,0 +1,547 @@
+import path from "node:path";
+import { strFromU8, unzipSync } from "fflate";
+
+export const PARTNER_BATCH_HEADERS = [
+  "序号（不能改）",
+  "标题",
+  "笔记内容（正文+话题）",
+  "图片",
+  "发布链接",
+  "笔记截图",
+  "数据分析截图（单篇笔记数据分析截图）",
+  "_系统笔记ID",
+  "_原笔记截图",
+  "_原数据分析截图",
+] as const;
+
+export const PARTNER_BATCH_VISIBLE_COLUMN_COUNT = 7;
+export const PARTNER_BATCH_MAX_BYTES = 80_000_000;
+
+export type PartnerBatchWorkbookColumns = {
+  headers: string[];
+  columnWidths: number[];
+  sourceImageStartColumn: number;
+  sourceImageCount: number;
+  sourceVideoStartColumn: number;
+  sourceVideoCount: number;
+  publishUrlColumn: number;
+  publishScreenshotColumn: number;
+  creatorScreenshotColumn: number;
+  systemColumn: number;
+};
+
+function nonNegativeInteger(value: number) {
+  return Number.isFinite(value) ? Math.max(0, Math.floor(value)) : 0;
+}
+
+export function buildPartnerBatchWorkbookColumns(input: {
+  contentFormat: "image_text" | "video";
+  maxSourceImages: number;
+  maxSourceVideos: number;
+}): PartnerBatchWorkbookColumns {
+  const sourceImageCount =
+    input.contentFormat === "video"
+      ? 0
+      : Math.max(1, nonNegativeInteger(input.maxSourceImages));
+  const sourceVideoCount =
+    input.contentFormat === "video"
+      ? Math.max(1, nonNegativeInteger(input.maxSourceVideos))
+      : 0;
+  const sourceImageStartColumn = 3;
+  const sourceVideoStartColumn = sourceImageStartColumn + sourceImageCount;
+  const publishUrlColumn = sourceVideoStartColumn + sourceVideoCount;
+  const publishScreenshotColumn = publishUrlColumn + 1;
+  const creatorScreenshotColumn = publishScreenshotColumn + 1;
+  const systemColumn = creatorScreenshotColumn + 1;
+  return {
+    headers: [
+      "序号（不能改）",
+      "标题",
+      "笔记内容（正文+话题）",
+      ...Array.from(
+        { length: sourceImageCount },
+        (_, index) => `图片${index + 1}`,
+      ),
+      ...Array.from(
+        { length: sourceVideoCount },
+        (_, index) => `视频${index + 1}`,
+      ),
+      "发布链接",
+      "笔记截图",
+      "数据分析截图（单篇笔记数据分析截图）",
+      "_系统笔记ID",
+      "_原笔记截图",
+      "_原数据分析截图",
+    ],
+    columnWidths: [
+      14,
+      30,
+      62,
+      ...Array.from({ length: sourceImageCount }, () => 24),
+      ...Array.from({ length: sourceVideoCount }, () => 20),
+      45,
+      28,
+      32,
+      22,
+      22,
+      22,
+    ],
+    sourceImageStartColumn,
+    sourceImageCount,
+    sourceVideoStartColumn,
+    sourceVideoCount,
+    publishUrlColumn,
+    publishScreenshotColumn,
+    creatorScreenshotColumn,
+    systemColumn,
+  };
+}
+
+function firstForwardedValue(value: string | null) {
+  return value?.split(",")[0]?.trim() ?? "";
+}
+
+function httpOrigin(value: string) {
+  try {
+    const url = new URL(value);
+    return url.protocol === "http:" || url.protocol === "https:"
+      ? url.origin
+      : "";
+  } catch {
+    return "";
+  }
+}
+
+export function resolvePartnerWorkbookOrigin(
+  request: Request,
+  configuredOrigin = "",
+) {
+  const requestUrl = new URL(request.url);
+  const host =
+    firstForwardedValue(request.headers.get("x-forwarded-host")) ||
+    firstForwardedValue(request.headers.get("host"));
+  const forwardedProtocol = firstForwardedValue(
+    request.headers.get("x-forwarded-proto"),
+  ).toLowerCase();
+  const protocol = ["http", "https"].includes(forwardedProtocol)
+    ? forwardedProtocol
+    : requestUrl.protocol.replace(":", "");
+  const proxyOrigin = host ? httpOrigin(`${protocol}://${host}`) : "";
+  return (
+    httpOrigin(configuredOrigin) ||
+    proxyOrigin ||
+    httpOrigin(requestUrl.origin) ||
+    requestUrl.origin
+  );
+}
+
+export type PartnerBatchImage = {
+  bytes: Uint8Array;
+  contentType: string;
+  fileName: string;
+};
+
+export type PartnerBatchImportRow = {
+  spreadsheetRow: number;
+  sequence: string;
+  title: string;
+  publishUrl: string;
+  distributionId: string;
+  originalPublishScreenshotKey: string;
+  originalCreatorScreenshotKey: string;
+  publishScreenshot: PartnerBatchImage | null;
+  creatorScreenshot: PartnerBatchImage | null;
+};
+
+function decodeXml(value: string) {
+  return value
+    .replace(/<[^>]+>/g, "")
+    .replace(/&lt;/g, "<")
+    .replace(/&gt;/g, ">")
+    .replace(/&quot;/g, '"')
+    .replace(/&apos;/g, "'")
+    .replace(/&amp;/g, "&")
+    .replace(/&#(\d+);/g, (_, code) => String.fromCodePoint(Number(code)))
+    .replace(/&#x([0-9a-f]+);/gi, (_, code) =>
+      String.fromCodePoint(Number.parseInt(code, 16)),
+    );
+}
+
+function xmlAttribute(value: string) {
+  return decodeXml(value);
+}
+
+function textNodes(xml: string) {
+  return [...xml.matchAll(/<t\b[^>]*>([\s\S]*?)<\/t>/g)]
+    .map((match) => decodeXml(match[1]))
+    .join("");
+}
+
+function columnIndex(reference: string) {
+  const letters = reference.match(/^[A-Z]+/i)?.[0]?.toUpperCase() ?? "";
+  let result = 0;
+  for (const letter of letters) result = result * 26 + letter.charCodeAt(0) - 64;
+  return Math.max(0, result - 1);
+}
+
+function parseWorksheet(xml: string, sharedStrings: string[]) {
+  const rows: string[][] = [];
+  for (const rowMatch of xml.matchAll(/<row\b([^>]*)>([\s\S]*?)<\/row>/g)) {
+    const rowNumber = Number(
+      rowMatch[1].match(/\br="(\d+)"/)?.[1] ?? rows.length + 1,
+    );
+    const values: string[] = [];
+    for (const cellMatch of rowMatch[2].matchAll(
+      /<c\b([^>]*)>([\s\S]*?)<\/c>/g,
+    )) {
+      const attributes = cellMatch[1];
+      const body = cellMatch[2];
+      const reference = attributes.match(/\br="([A-Z]+\d+)"/i)?.[1] ?? "A1";
+      const type = attributes.match(/\bt="([^"]+)"/)?.[1] ?? "";
+      const rawValue = body.match(/<v>([\s\S]*?)<\/v>/)?.[1] ?? "";
+      let value = "";
+      if (type === "s") value = sharedStrings[Number(rawValue)] ?? "";
+      else if (type === "inlineStr") value = textNodes(body);
+      else if (type === "b") value = rawValue === "1" ? "TRUE" : "FALSE";
+      else value = decodeXml(rawValue);
+      values[columnIndex(reference)] = value.trim();
+    }
+    while (rows.length < rowNumber - 1) rows.push([]);
+    rows[rowNumber - 1] = values;
+  }
+  return rows;
+}
+
+function normalizeHeader(value: string) {
+  return value.replace(/[\s_\-（）()]/g, "").toLocaleLowerCase("zh-CN");
+}
+
+function isSourceImageHeader(value: string) {
+  return /^(?:图片|发布配图)\d*$/.test(normalizeHeader(value));
+}
+
+function headerAliases(header: (typeof PARTNER_BATCH_HEADERS)[number]) {
+  const aliases: Record<string, string[]> = {
+    "序号（不能改）": ["序号（不能改）", "序号"],
+    标题: ["标题"],
+    "笔记内容（正文+话题）": ["笔记内容（正文+话题）", "笔记内容"],
+    图片: ["图片", "发布配图"],
+    发布链接: ["发布链接"],
+    笔记截图: ["笔记截图", "发布截图"],
+    "数据分析截图（单篇笔记数据分析截图）": [
+      "数据分析截图（单篇笔记数据分析截图）",
+      "数据分析截图",
+      "创作者中心截图",
+    ],
+    _系统笔记ID: ["_系统笔记ID", "系统笔记ID"],
+    _原笔记截图: ["_原笔记截图"],
+    _原数据分析截图: ["_原数据分析截图"],
+  };
+  return aliases[header] ?? [header];
+}
+
+function findHeader(rows: string[][]) {
+  for (let rowIndex = 0; rowIndex < Math.min(rows.length, 8); rowIndex += 1) {
+    const mapping = new Map<(typeof PARTNER_BATCH_HEADERS)[number], number>();
+    rows[rowIndex].forEach((value, column) => {
+      for (const header of PARTNER_BATCH_HEADERS) {
+        if (header === "图片" && isSourceImageHeader(value)) {
+          if (!mapping.has(header)) mapping.set(header, column);
+          break;
+        }
+        if (
+          headerAliases(header).some(
+            (alias) => normalizeHeader(alias) === normalizeHeader(value),
+          )
+        ) {
+          mapping.set(header, column);
+          break;
+        }
+      }
+    });
+    if (
+      mapping.has("序号（不能改）") &&
+      mapping.has("标题") &&
+      mapping.has("发布链接") &&
+      mapping.has("笔记截图") &&
+      mapping.has("数据分析截图（单篇笔记数据分析截图）") &&
+      mapping.has("_系统笔记ID")
+    ) {
+      return { rowIndex, mapping };
+    }
+  }
+  return null;
+}
+
+function relationshipMap(xml: string) {
+  const relationships = new Map<string, string>();
+  for (const match of xml.matchAll(/<Relationship\b([^>]*)\/?\s*>/g)) {
+    const attributes = match[1];
+    const id = attributes.match(/\bId="([^"]+)"/)?.[1] ?? "";
+    const target = attributes.match(/\bTarget="([^"]+)"/)?.[1] ?? "";
+    if (id && target) relationships.set(id, xmlAttribute(target));
+  }
+  return relationships;
+}
+
+function contentType(bytes: Uint8Array, fileName: string) {
+  if (bytes[0] === 0x89 && bytes[1] === 0x50) return "image/png";
+  if (bytes[0] === 0xff && bytes[1] === 0xd8) return "image/jpeg";
+  if (bytes[0] === 0x47 && bytes[1] === 0x49 && bytes[2] === 0x46) {
+    return "image/gif";
+  }
+  if (String.fromCharCode(...bytes.slice(8, 12)) === "WEBP") return "image/webp";
+  const extension = path.extname(fileName).toLowerCase();
+  return extension === ".png"
+    ? "image/png"
+    : extension === ".gif"
+      ? "image/gif"
+      : extension === ".webp"
+        ? "image/webp"
+        : "image/jpeg";
+}
+
+function resolveZipPath(base: string, target: string) {
+  return path.posix.normalize(path.posix.join(path.posix.dirname(base), target));
+}
+
+function parseDrawingImages(entries: Record<string, Uint8Array>) {
+  const images = new Map<string, PartnerBatchImage>();
+  const sheetRelationshipsXml = entries["xl/worksheets/_rels/sheet1.xml.rels"]
+    ? strFromU8(entries["xl/worksheets/_rels/sheet1.xml.rels"])
+    : "";
+  const sheetRelationships = relationshipMap(sheetRelationshipsXml);
+  const sheetXml = entries["xl/worksheets/sheet1.xml"]
+    ? strFromU8(entries["xl/worksheets/sheet1.xml"])
+    : "";
+  const drawingId = sheetXml.match(/<drawing\b[^>]*r:id="([^"]+)"/)?.[1] ?? "";
+  const drawingTarget = sheetRelationships.get(drawingId);
+  if (!drawingTarget) return images;
+  const drawingPath = resolveZipPath("xl/worksheets/sheet1.xml", drawingTarget);
+  const drawingXml = entries[drawingPath] ? strFromU8(entries[drawingPath]) : "";
+  const drawingRelationshipsPath = path.posix.join(
+    path.posix.dirname(drawingPath),
+    "_rels",
+    `${path.posix.basename(drawingPath)}.rels`,
+  );
+  const drawingRelationships = relationshipMap(
+    entries[drawingRelationshipsPath]
+      ? strFromU8(entries[drawingRelationshipsPath])
+      : "",
+  );
+  for (const anchor of drawingXml.matchAll(
+    /<xdr:(?:oneCellAnchor|twoCellAnchor)\b[^>]*>[\s\S]*?<xdr:from>([\s\S]*?)<\/xdr:from>[\s\S]*?<a:blip\b[^>]*r:embed="([^"]+)"[\s\S]*?<\/xdr:(?:oneCellAnchor|twoCellAnchor)>/g,
+  )) {
+    const column = Number(anchor[1].match(/<xdr:col>(\d+)<\/xdr:col>/)?.[1]);
+    const row = Number(anchor[1].match(/<xdr:row>(\d+)<\/xdr:row>/)?.[1]);
+    const mediaTarget = drawingRelationships.get(anchor[2]);
+    if (!Number.isInteger(column) || !Number.isInteger(row) || !mediaTarget) continue;
+    const mediaPath = resolveZipPath(drawingPath, mediaTarget);
+    const bytes = entries[mediaPath];
+    if (!bytes) continue;
+    images.set(`${row + 1}:${column}`, {
+      bytes,
+      contentType: contentType(bytes, mediaPath),
+      fileName: path.posix.basename(mediaPath),
+    });
+  }
+  return images;
+}
+
+function parseRichValueImages(entries: Record<string, Uint8Array>) {
+  const images = new Map<string, PartnerBatchImage>();
+  const worksheetXml = entries["xl/worksheets/sheet1.xml"]
+    ? strFromU8(entries["xl/worksheets/sheet1.xml"])
+    : "";
+  const metadataXml = entries["xl/metadata.xml"]
+    ? strFromU8(entries["xl/metadata.xml"])
+    : "";
+  const richValueXml = entries["xl/richData/rdrichvalue.xml"]
+    ? strFromU8(entries["xl/richData/rdrichvalue.xml"])
+    : "";
+  const richValueRelXml = entries["xl/richData/richValueRel.xml"]
+    ? strFromU8(entries["xl/richData/richValueRel.xml"])
+    : "";
+  const richValueRelRelationships = relationshipMap(
+    entries["xl/richData/_rels/richValueRel.xml.rels"]
+      ? strFromU8(entries["xl/richData/_rels/richValueRel.xml.rels"])
+      : "",
+  );
+  if (
+    !worksheetXml ||
+    !metadataXml ||
+    !richValueXml ||
+    !richValueRelXml ||
+    !richValueRelRelationships.size
+  ) {
+    return images;
+  }
+
+  const valueMetadataXml =
+    metadataXml.match(/<valueMetadata\b[^>]*>([\s\S]*?)<\/valueMetadata>/)?.[1] ??
+    "";
+  const metadataToRichValue = [
+    ...valueMetadataXml.matchAll(/<bk\b[^>]*>[\s\S]*?<rc\b[^>]*\bv="(\d+)"[^>]*\/>[\s\S]*?<\/bk>/g),
+  ].map((match) => Number(match[1]));
+  const richValueToRelationship = [
+    ...richValueXml.matchAll(/<rv\b[^>]*>([\s\S]*?)<\/rv>/g),
+  ].map((match) => Number(match[1].match(/<v>(\d+)<\/v>/)?.[1] ?? -1));
+  const relationshipIds = [
+    ...richValueRelXml.matchAll(/<rel\b[^>]*\br:id="([^"]+)"[^>]*\/>/g),
+  ].map((match) => match[1]);
+
+  for (const cell of worksheetXml.matchAll(
+    /<c\b([^>]*)>[\s\S]*?<\/c>/g,
+  )) {
+    const reference = cell[1].match(/\br="([A-Z]+\d+)"/i)?.[1] ?? "";
+    const metadataIndex = Number(cell[1].match(/\bvm="(\d+)"/)?.[1] ?? 0);
+    const row = Number(reference.match(/\d+$/)?.[0] ?? 0);
+    const column = columnIndex(reference);
+    if (!reference || !metadataIndex || !row) continue;
+    const richValueIndex = metadataToRichValue[metadataIndex - 1];
+    const relationshipIndex = richValueToRelationship[richValueIndex];
+    const relationshipId = relationshipIds[relationshipIndex];
+    const mediaTarget = richValueRelRelationships.get(relationshipId);
+    if (!mediaTarget) continue;
+    const mediaPath = resolveZipPath(
+      "xl/richData/richValueRel.xml",
+      mediaTarget,
+    );
+    const bytes = entries[mediaPath];
+    if (!bytes) continue;
+    images.set(`${row}:${column}`, {
+      bytes,
+      contentType: contentType(bytes, mediaPath),
+      fileName: path.posix.basename(mediaPath),
+    });
+  }
+  return images;
+}
+
+function parseWpsCellImages(entries: Record<string, Uint8Array>) {
+  const images = new Map<string, PartnerBatchImage>();
+  const worksheetXml = entries["xl/worksheets/sheet1.xml"]
+    ? strFromU8(entries["xl/worksheets/sheet1.xml"])
+    : "";
+  const cellImagesXml = entries["xl/cellimages.xml"]
+    ? strFromU8(entries["xl/cellimages.xml"])
+    : "";
+  const relationships = relationshipMap(
+    entries["xl/_rels/cellimages.xml.rels"]
+      ? strFromU8(entries["xl/_rels/cellimages.xml.rels"])
+      : "",
+  );
+  if (!worksheetXml || !cellImagesXml || !relationships.size) return images;
+
+  const imageIdToRelationship = new Map<string, string>();
+  for (const match of cellImagesXml.matchAll(
+    /<(?:etc:)?cellImage\b[^>]*>([\s\S]*?)<\/(?:etc:)?cellImage>/g,
+  )) {
+    const imageId = match[1].match(
+      /<(?:xdr:)?cNvPr\b[^>]*\bname="([^"]+)"/,
+    )?.[1];
+    const relationshipId = match[1].match(
+      /<(?:a:)?blip\b[^>]*\br:embed="([^"]+)"/,
+    )?.[1];
+    if (imageId && relationshipId) {
+      imageIdToRelationship.set(imageId, relationshipId);
+    }
+  }
+
+  for (const cell of worksheetXml.matchAll(/<c\b([^>]*)>([\s\S]*?)<\/c>/g)) {
+    const reference = cell[1].match(/\br="([A-Z]+\d+)"/i)?.[1] ?? "";
+    const imageId = decodeXml(cell[2]).match(/DISPIMG\("([^"]+)"/i)?.[1];
+    if (!reference || !imageId) continue;
+    const relationshipId = imageIdToRelationship.get(imageId);
+    const mediaTarget = relationshipId
+      ? relationships.get(relationshipId)
+      : undefined;
+    if (!mediaTarget) continue;
+    const mediaPath = resolveZipPath("xl/cellimages.xml", mediaTarget);
+    const bytes = entries[mediaPath];
+    if (!bytes) continue;
+    const row = Number(reference.match(/\d+$/)?.[0] ?? 0);
+    if (!row) continue;
+    images.set(`${row}:${columnIndex(reference)}`, {
+      bytes,
+      contentType: contentType(bytes, mediaPath),
+      fileName: path.posix.basename(mediaPath),
+    });
+  }
+  return images;
+}
+
+function parseImages(entries: Record<string, Uint8Array>) {
+  const images = parseDrawingImages(entries);
+  for (const [cell, image] of parseRichValueImages(entries)) {
+    images.set(cell, image);
+  }
+  for (const [cell, image] of parseWpsCellImages(entries)) {
+    images.set(cell, image);
+  }
+  return images;
+}
+
+function valueAt(
+  row: string[],
+  mapping: Map<(typeof PARTNER_BATCH_HEADERS)[number], number>,
+  header: (typeof PARTNER_BATCH_HEADERS)[number],
+) {
+  const column = mapping.get(header);
+  return column === undefined ? "" : String(row[column] ?? "").trim();
+}
+
+export function parsePartnerBatchWorkbook(input: ArrayBuffer | Uint8Array) {
+  const bytes = input instanceof Uint8Array ? input : new Uint8Array(input);
+  if (bytes.byteLength > PARTNER_BATCH_MAX_BYTES) {
+    throw new Error("批量回填表不能超过80MB");
+  }
+  const entries = unzipSync(bytes);
+  const worksheetBytes = entries["xl/worksheets/sheet1.xml"];
+  if (!worksheetBytes) throw new Error("Excel 中没有可读取的批量回填工作表");
+  const sharedXml = entries["xl/sharedStrings.xml"]
+    ? strFromU8(entries["xl/sharedStrings.xml"])
+    : "";
+  const sharedStrings = [
+    ...sharedXml.matchAll(/<si\b[^>]*>([\s\S]*?)<\/si>/g),
+  ].map((match) => textNodes(match[1]));
+  const rows = parseWorksheet(strFromU8(worksheetBytes), sharedStrings);
+  const header = findHeader(rows);
+  if (!header) {
+    throw new Error("表格结构不正确，请使用本领取页面导出的批量回填表");
+  }
+  const images = parseImages(entries);
+  const publishScreenshotColumn = header.mapping.get("笔记截图")!;
+  const creatorScreenshotColumn = header.mapping.get(
+    "数据分析截图（单篇笔记数据分析截图）",
+  )!;
+  const result: PartnerBatchImportRow[] = [];
+  for (let index = header.rowIndex + 1; index < rows.length; index += 1) {
+    const row = rows[index];
+    const distributionId = valueAt(row, header.mapping, "_系统笔记ID");
+    if (!distributionId && !row.some((value) => String(value ?? "").trim())) continue;
+    result.push({
+      spreadsheetRow: index + 1,
+      sequence: valueAt(row, header.mapping, "序号（不能改）"),
+      title: valueAt(row, header.mapping, "标题"),
+      publishUrl: valueAt(row, header.mapping, "发布链接"),
+      distributionId,
+      originalPublishScreenshotKey: valueAt(
+        row,
+        header.mapping,
+        "_原笔记截图",
+      ),
+      originalCreatorScreenshotKey: valueAt(
+        row,
+        header.mapping,
+        "_原数据分析截图",
+      ),
+      publishScreenshot:
+        images.get(`${index + 1}:${publishScreenshotColumn}`) ?? null,
+      creatorScreenshot:
+        images.get(`${index + 1}:${creatorScreenshotColumn}`) ?? null,
+    });
+  }
+  if (result.length === 0) throw new Error("表格中没有可回填的笔记");
+  return result;
+}
--- a/lib/partner-utils.ts
+++ b/lib/partner-utils.ts
@@ -1,22 +1,31 @@
 import { hashText } from "./mvp-db";
 import {
+  extractPublishUrl,
  extractXhsPublishUrl,
+  platformFromPublishUrl,
  safeHttpUrl,
+  type SupportedPlatform,
 } from "./publish-url";

 export { extractXhsPublishUrl } from "./publish-url";

-export function accountFromPublishLink(input: string) {
-  const url = safeHttpUrl(extractXhsPublishUrl(input));
+export function accountFromPublishLink(
+  input: string,
+  expectedPlatform?: SupportedPlatform,
+) {
+  const extracted = expectedPlatform
+    ? extractPublishUrl(input, expectedPlatform)
+    : extractXhsPublishUrl(input) || extractPublishUrl(input, "抖音");
+  const url = safeHttpUrl(extracted);
  if (!url) return null;
-  const platform =
-    url.hostname.includes("xiaohongshu") || url.hostname.includes("xhslink")
-      ? "小红书"
-      : "其他平台";
+  const platform = platformFromPublishUrl(url.toString());
+  if (!platform || (expectedPlatform && platform !== expectedPlatform)) return null;
  const noteId =
-    url.pathname.match(
-      /\/(?:discovery\/item|explore)\/([A-Za-z0-9_-]{8,80})/,
-    )?.[1] ?? "";
+    platform === "抖音"
+      ? url.pathname.match(/\/(?:video|note)\/([A-Za-z0-9_-]{8,80})/)?.[1] ?? ""
+      : url.pathname.match(
+          /\/(?:discovery\/item|explore)\/([A-Za-z0-9_-]{8,80})/,
+        )?.[1] ?? "";
  const platformUid = `pending-${hashText(
    noteId || `${url.origin}${url.pathname}`,
  )}`;
--- a/lib/publish-url.ts
+++ b/lib/publish-url.ts
@@ -25,3 +25,44 @@ export function extractXhsPublishUrl(input: string) {
  }
  return "";
 }
+
+export type SupportedPlatform = "小红书" | "抖音";
+
+function platformMatches(url: URL, platform: SupportedPlatform) {
+  const hostname = url.hostname.toLowerCase();
+  if (platform === "抖音") {
+    return hostname === "douyin.com" || hostname.endsWith(".douyin.com");
+  }
+  return (
+    hostname === "xiaohongshu.com" ||
+    hostname.endsWith(".xiaohongshu.com") ||
+    hostname === "xhslink.cn" ||
+    hostname.endsWith(".xhslink.cn")
+  );
+}
+
+export function extractPublishUrl(
+  input: string,
+  platform: SupportedPlatform = "小红书",
+) {
+  const candidates =
+    input.match(/https?:\/\/[^\s<>"'，。！？；：、【】（）]+/gi) ?? [];
+  for (const candidate of candidates) {
+    const cleaned = candidate.replace(/[.,!?;:~)\]}]+$/g, "");
+    const url = safeHttpUrl(cleaned);
+    if (url && platformMatches(url, platform)) return url.toString();
+  }
+  return "";
+}
+
+export function extractAnyPublishUrl(input: string) {
+  return extractPublishUrl(input, "小红书") || extractPublishUrl(input, "抖音");
+}
+
+export function platformFromPublishUrl(input: string): SupportedPlatform | "" {
+  const url = safeHttpUrl(input);
+  if (!url) return "";
+  if (platformMatches(url, "小红书")) return "小红书";
+  if (platformMatches(url, "抖音")) return "抖音";
+  return "";
+}
--- a/lib/recovery-workbook.ts
+++ b/lib/recovery-workbook.ts
@@ -13,6 +13,9 @@ export type RecoveryWorkbookRow = {
  images: Array<{
    column: number;
    image: RecoveryWorkbookImage;
+    offsetX?: number;
+    maxWidth?: number;
+    maxHeight?: number;
  }>;
  hyperlinks?: Array<{
    column: number;
@@ -25,6 +28,7 @@ type WorkbookOptions = {
  headers: string[];
  columnWidths: number[];
  rows: RecoveryWorkbookRow[];
+  hiddenColumns?: number[];
 };

 const XML_HEADER = '<?xml version="1.0" encoding="UTF-8" standalone="yes"?>';
@@ -130,9 +134,17 @@ function imageDimensions(image: RecoveryWorkbookImage) {
  return { width: 4, height: 3 };
 }

-function imageDisplaySize(image: RecoveryWorkbookImage) {
+function imageDisplaySize(
+  image: RecoveryWorkbookImage,
+  maxWidth = 160,
+  maxHeight = 150,
+) {
  const dimensions = imageDimensions(image);
-  const scale = Math.min(160 / dimensions.width, 150 / dimensions.height, 1);
+  const scale = Math.min(
+    maxWidth / dimensions.width,
+    maxHeight / dimensions.height,
+    1,
+  );
  return {
    width: Math.max(28, Math.round(dimensions.width * scale)),
    height: Math.max(28, Math.round(dimensions.height * scale)),
@@ -152,6 +164,14 @@ export function buildRecoveryWorkbook(options: WorkbookOptions) {
  const imageEntries = options.rows.flatMap((row, rowIndex) =>
    row.images.map((item) => ({ ...item, row: rowIndex + 1 })),
  );
+  const imageCells = new Set<string>();
+  imageEntries.forEach((entry) => {
+    const key = `${entry.row}:${entry.column}`;
+    if (imageCells.has(key)) {
+      throw new Error("Excel 单元格内只能嵌入一张图片，请为每张图片分配独立列");
+    }
+    imageCells.add(key);
+  });
  const hyperlinkEntries = options.rows.flatMap((row, rowIndex) =>
    (row.hyperlinks ?? [])
      .map((item) => ({
@@ -178,7 +198,7 @@ export function buildRecoveryWorkbook(options: WorkbookOptions) {
          const reference = `${columnName(columnIndex)}${number}`;
          const value = row.cells[columnIndex] ?? "";
          if (imageColumns.has(columnIndex)) {
-            return inlineCell(reference, value || "见图", 4);
+            return inlineCell(reference, value, 4);
          }
          return typeof value === "number"
            ? numberCell(reference, value, 3)
@@ -196,17 +216,22 @@ export function buildRecoveryWorkbook(options: WorkbookOptions) {
  const columns = options.headers
    .map((_, index) => {
      const width = Math.min(70, Math.max(8, options.columnWidths[index] ?? 14));
-      return `<col min="${index + 1}" max="${index + 1}" width="${width}" customWidth="1"/>`;
+      const hidden = options.hiddenColumns?.includes(index) ? ' hidden="1"' : "";
+      return `<col min="${index + 1}" max="${index + 1}" width="${width}" customWidth="1"${hidden}/>`;
    })
    .join("");

  const drawingXml = imageEntries.length
    ? `${XML_HEADER}<xdr:wsDr xmlns:xdr="http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing" xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships">${imageEntries
        .map((entry, index) => {
-          const size = imageDisplaySize(entry.image);
-          const width = size.width * 9525;
-          const height = size.height * 9525;
-          return `<xdr:oneCellAnchor><xdr:from><xdr:col>${entry.column}</xdr:col><xdr:colOff>57150</xdr:colOff><xdr:row>${entry.row}</xdr:row><xdr:rowOff>57150</xdr:rowOff></xdr:from><xdr:ext cx="${width}" cy="${height}"/><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="${index + 1}" name="${cleanXmlText(entry.image.description || `图片${index + 1}`)}"/><xdr:cNvPicPr><a:picLocks noChangeAspect="1"/></xdr:cNvPicPr></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId${index + 1}"/><a:stretch><a:fillRect/></a:stretch></xdr:blipFill><xdr:spPr><a:prstGeom prst="rect"><a:avLst/></a:prstGeom></xdr:spPr></xdr:pic><xdr:clientData/></xdr:oneCellAnchor>`;
+          const size = imageDisplaySize(
+            entry.image,
+            entry.maxWidth ?? 160,
+            entry.maxHeight ?? 150,
+          );
+          const offsetX = entry.offsetX ?? 6;
+          const offsetY = 6;
+          return `<xdr:twoCellAnchor editAs="twoCell"><xdr:from><xdr:col>${entry.column}</xdr:col><xdr:colOff>${offsetX * 9525}</xdr:colOff><xdr:row>${entry.row}</xdr:row><xdr:rowOff>${offsetY * 9525}</xdr:rowOff></xdr:from><xdr:to><xdr:col>${entry.column}</xdr:col><xdr:colOff>${(offsetX + size.width) * 9525}</xdr:colOff><xdr:row>${entry.row}</xdr:row><xdr:rowOff>${(offsetY + size.height) * 9525}</xdr:rowOff></xdr:to><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="${index + 1}" name="${cleanXmlText(entry.image.description || `图片${index + 1}`)}"/><xdr:cNvPicPr><a:picLocks noChangeAspect="1"/></xdr:cNvPicPr></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId${index + 1}"/><a:stretch><a:fillRect/></a:stretch></xdr:blipFill><xdr:spPr><a:prstGeom prst="rect"><a:avLst/></a:prstGeom></xdr:spPr></xdr:pic><xdr:clientData/></xdr:twoCellAnchor>`;
        })
        .join("")}</xdr:wsDr>`
    : "";
@@ -227,7 +252,10 @@ export function buildRecoveryWorkbook(options: WorkbookOptions) {
  const imageContentTypes = [...imageFormats.entries()]
    .map(([extension, contentType]) => `<Default Extension="${extension}" ContentType="${contentType}"/>`)
    .join("");
-  const contentTypes = `${XML_HEADER}<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"><Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/><Default Extension="xml" ContentType="application/xml"/>${imageContentTypes}<Override PartName="/xl/workbook.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet.main+xml"/><Override PartName="/xl/worksheets/sheet1.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.worksheet+xml"/><Override PartName="/xl/styles.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.styles+xml"/>${imageEntries.length ? '<Override PartName="/xl/drawings/drawing1.xml" ContentType="application/vnd.openxmlformats-officedocument.drawing+xml"/>' : ""}<Override PartName="/docProps/core.xml" ContentType="application/vnd.openxmlformats-package.core-properties+xml"/><Override PartName="/docProps/app.xml" ContentType="application/vnd.openxmlformats-officedocument.extended-properties+xml"/></Types>`;
+  const drawingContentType = imageEntries.length
+    ? '<Override PartName="/xl/drawings/drawing1.xml" ContentType="application/vnd.openxmlformats-officedocument.drawing+xml"/>'
+    : "";
+  const contentTypes = `${XML_HEADER}<Types xmlns="http://schemas.openxmlformats.org/package/2006/content-types"><Default Extension="rels" ContentType="application/vnd.openxmlformats-package.relationships+xml"/><Default Extension="xml" ContentType="application/xml"/>${imageContentTypes}<Override PartName="/xl/workbook.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.sheet.main+xml"/><Override PartName="/xl/worksheets/sheet1.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.worksheet+xml"/><Override PartName="/xl/styles.xml" ContentType="application/vnd.openxmlformats-officedocument.spreadsheetml.styles+xml"/>${drawingContentType}<Override PartName="/docProps/core.xml" ContentType="application/vnd.openxmlformats-package.core-properties+xml"/><Override PartName="/docProps/app.xml" ContentType="application/vnd.openxmlformats-officedocument.extended-properties+xml"/></Types>`;
  const hyperlinkRelationshipOffset = imageEntries.length ? 2 : 1;
  const hyperlinksXml = hyperlinkEntries.length
    ? `<hyperlinks>${hyperlinkEntries
@@ -266,7 +294,9 @@ export function buildRecoveryWorkbook(options: WorkbookOptions) {
  }
  if (imageEntries.length) {
    files["xl/drawings/drawing1.xml"] = strToU8(drawingXml);
-    files["xl/drawings/_rels/drawing1.xml.rels"] = strToU8(drawingRelationships);
+    files["xl/drawings/_rels/drawing1.xml.rels"] = strToU8(
+      drawingRelationships,
+    );
    imageEntries.forEach((entry, index) => {
      const format = imageFormat(entry.image.contentType, entry.image.bytes);
      files[`xl/media/image${index + 1}.${format.extension}`] = entry.image.bytes;
--- a/lib/resource-import.ts
+++ b/lib/resource-import.ts
@@ -1,7 +1,7 @@
 import { strFromU8, unzipSync } from "fflate";

-export const RESOURCE_IMPORT_MAX_ROWS = 100;
-export const RESOURCE_IMPORT_MAX_BYTES = 5 * 1024 * 1024;
+export const RESOURCE_IMPORT_MAX_ROWS = 10_000;
+export const RESOURCE_IMPORT_MAX_BYTES = 20 * 1024 * 1024;

 export type ResourceImportRow = {
  rowNumber: number;
@@ -12,6 +12,9 @@ export type ResourceImportRow = {
  ipLocation: string;
  followers: number;
  followersResolved: boolean;
+  gender: "" | "男" | "女";
+  bio: string;
+  tags: string[];
  cooperationSource: string;
  errors: string[];
 };
@@ -22,6 +25,9 @@ const HEADER_ALIASES = {
  publicAccountId: ["账号ID", "账号号", "小红书号", "抖音号"],
  ipLocation: ["IP属地", "IP地址", "IP地区", "IP所在地"],
  followers: ["粉丝数", "粉丝", "粉丝量"],
+  gender: ["性别"],
+  bio: ["简介", "账号简介", "个人简介"],
+  tags: ["标签", "账号标签"],
  cooperationSource: ["合作来源", "资源来源", "历史合作来源"],
 } as const;

@@ -60,9 +66,11 @@ function parseWorksheet(xml: string, sharedStrings: string[]) {
    const rowAttributes = rowMatch[1];
    const rowNumber = Number(rowAttributes.match(/\br="(\d+)"/)?.[1] ?? rows.length + 1);
    const values: string[] = [];
-    for (const cellMatch of rowMatch[2].matchAll(/<c\b([^>]*)>([\s\S]*?)<\/c>/g)) {
+    for (const cellMatch of rowMatch[2].matchAll(
+      /<c\b([^>]*?)(?:\/>|>([\s\S]*?)<\/c>)/g,
+    )) {
      const attributes = cellMatch[1];
-      const body = cellMatch[2];
+      const body = cellMatch[2] ?? "";
      const reference = attributes.match(/\br="([A-Z]+\d+)"/i)?.[1] ?? "A1";
      const type = attributes.match(/\bt="([^"]+)"/)?.[1] ?? "";
      const rawValue = body.match(/<v>([\s\S]*?)<\/v>/)?.[1] ?? "";
@@ -183,6 +191,15 @@ export function platformFromProfileUrl(profileUrl: string) {
    ) {
      return "小红书";
    }
+    if (
+      ((url.hostname === "douyin.com" || url.hostname.endsWith(".douyin.com")) &&
+        /^\/user\/[^/]+/i.test(url.pathname)) ||
+      ((url.hostname === "iesdouyin.com" ||
+        url.hostname.endsWith(".iesdouyin.com")) &&
+        /^\/share\/user\/[^/]+/i.test(url.pathname))
+    ) {
+      return "抖音";
+    }
  } catch {
    // URL validation is reported by normalizeRows.
  }
@@ -212,10 +229,42 @@ export function parseResourceFollowers(value: string) {
  };
 }

+export function normalizeResourceGender(value: string) {
+  const normalized = value.trim().toLocaleLowerCase("zh-CN");
+  if (!normalized || ["未知", "未填写", "待识别", "unknown"].includes(normalized)) {
+    return { value: "" as const, valid: true };
+  }
+  if (["男", "男性", "male", "m"].includes(normalized)) {
+    return { value: "男" as const, valid: true };
+  }
+  if (["女", "女性", "female", "f"].includes(normalized)) {
+    return { value: "女" as const, valid: true };
+  }
+  return { value: "" as const, valid: false };
+}
+
+export function normalizeResourceTags(value: string | string[]) {
+  const source = Array.isArray(value) ? value.join(",") : value;
+  return [
+    ...new Set(
+      source
+        .split(/[,，、;；|]/)
+        .map((item) => item.trim().replace(/^#+/, ""))
+        .filter(Boolean),
+    ),
+  ];
+}
+
 export function resourceImportMissingFields(
  row: Pick<
    ResourceImportRow,
-    "nickname" | "publicAccountId" | "ipLocation" | "followersResolved"
+    | "nickname"
+    | "publicAccountId"
+    | "ipLocation"
+    | "followersResolved"
+    | "gender"
+    | "bio"
+    | "tags"
  >,
 ) {
  const missing: string[] = [];
@@ -225,6 +274,8 @@ export function resourceImportMissingFields(
    missing.push("ipLocation");
  }
  if (!row.followersResolved) missing.push("followers");
+  if (!row.gender) missing.push("gender");
+  if (!row.bio.trim()) missing.push("bio");
  return missing;
 }

@@ -242,22 +293,39 @@ function normalizeRows(rows: string[][]) {
    const platform = platformFromProfileUrl(profileUrl);
    const rawFollowers = valueAt(source, header.mapping, "followers");
    const parsedFollowers = parseResourceFollowers(rawFollowers);
+    const parsedGender = normalizeResourceGender(
+      valueAt(source, header.mapping, "gender"),
+    );
+    const tags = normalizeResourceTags(valueAt(source, header.mapping, "tags"));
+    const ipLocation = valueAt(source, header.mapping, "ipLocation");
    const errors: string[] = [];
    if (!rawProfileUrl) errors.push("账号主页不能为空");
    else if (!profileUrl) errors.push("账号主页链接格式不正确");
-    else if (!platform) errors.push("当前自动解析仅支持小红书账号主页");
+    else if (!platform) errors.push("当前自动解析仅支持小红书或抖音账号主页");
    if (!parsedFollowers.valid) {
      errors.push("粉丝数格式不正确，请填写数字或如 1.3万、10+");
    }
+    if (!parsedGender.valid) {
+      errors.push("性别格式不正确，请填写男、女或留空");
+    }
+    if (tags.length > 5) {
+      errors.push("单个账号最多填写 5 个标签，请用逗号分隔");
+    }
+    if (/^\d+$/.test(ipLocation)) {
+      errors.push("IP属地格式不正确，请填写省份、地区或国家名称");
+    }
    result.push({
      rowNumber: index + 1,
      platform,
      nickname: valueAt(source, header.mapping, "nickname"),
      publicAccountId: valueAt(source, header.mapping, "publicAccountId"),
      profileUrl,
-      ipLocation: valueAt(source, header.mapping, "ipLocation"),
+      ipLocation,
      followers: parsedFollowers.value,
      followersResolved: parsedFollowers.resolved,
+      gender: parsedGender.value,
+      bio: valueAt(source, header.mapping, "bio"),
+      tags: tags.slice(0, 5),
      cooperationSource: valueAt(source, header.mapping, "cooperationSource"),
      errors,
    });
@@ -269,6 +337,59 @@ function normalizeRows(rows: string[][]) {
  return result;
 }

+export type ManualResourceInput = {
+  profileUrl: string;
+  nickname?: string;
+  publicAccountId?: string;
+  ipLocation?: string;
+  followers?: string;
+  gender?: string;
+  bio?: string;
+  tags?: string | string[];
+  cooperationSource?: string;
+};
+
+export function buildManualRow(input: ManualResourceInput): ResourceImportRow {
+  const rawProfileUrl = (input.profileUrl ?? "").trim();
+  const profileUrl = normalizeProfileUrl(rawProfileUrl);
+  const platform = platformFromProfileUrl(profileUrl);
+  const parsedFollowers = parseResourceFollowers(input.followers ?? "");
+  const parsedGender = normalizeResourceGender(input.gender ?? "");
+  const tags = normalizeResourceTags(input.tags ?? "");
+  const ipLocation = (input.ipLocation ?? "").trim();
+  const errors: string[] = [];
+  if (!rawProfileUrl) errors.push("账号主页不能为空");
+  else if (!profileUrl) errors.push("账号主页链接格式不正确");
+  else if (!platform) errors.push("当前自动解析仅支持小红书或抖音账号主页");
+  if (!parsedFollowers.valid) {
+    errors.push("粉丝数格式不正确，请填写数字或如 1.3万、10+");
+  }
+  if (!parsedGender.valid) {
+    errors.push("性别格式不正确，请填写男、女或留空");
+  }
+  if (tags.length > 5) {
+    errors.push("单个账号最多填写 5 个标签，请用逗号分隔");
+  }
+  if (ipLocation && /^\d+$/.test(ipLocation)) {
+    errors.push("IP属地格式不正确，请填写省份、地区或国家名称");
+  }
+  return {
+    rowNumber: 1,
+    platform,
+    nickname: (input.nickname ?? "").trim(),
+    publicAccountId: (input.publicAccountId ?? "").trim(),
+    profileUrl,
+    ipLocation,
+    followers: parsedFollowers.value,
+    followersResolved: parsedFollowers.resolved,
+    gender: parsedGender.value,
+    bio: (input.bio ?? "").trim(),
+    tags: tags.slice(0, 5),
+    cooperationSource: (input.cooperationSource ?? "").trim(),
+    errors,
+  };
+}
+
 export function parseResourceImportFile(fileName: string, bytes: Uint8Array) {
  const extension = fileName.toLocaleLowerCase().split(".").pop();
  const workbooks =
--- a/lib/resource-write.ts
+++ b/lib/resource-write.ts
@@ -0,0 +1,366 @@
+import { ensureSchema, getRawDb } from "./mvp-db";
+import {
+  resolveCollectionMcpConfig,
+  resolveProfileDetailsFromMcp,
+  resolveXhsPublicAccountDetails,
+  type CollectionMcpBindings,
+} from "./mcp-collection-client";
+import { getRuntimeEnv } from "./runtime-env";
+import {
+  mergeCooperationSources,
+  normalizeProfileUrl,
+  resourceImportMissingFields,
+  resourcePlatformUid,
+  type ResourceImportRow,
+} from "./resource-import";
+
+export type AccountRow = {
+  id: string;
+  platform: string;
+  platform_uid: string;
+  public_account_id: string;
+  nickname: string;
+  profile_url: string;
+  ip_location: string;
+  followers: number;
+  gender: string;
+  bio: string;
+  tags: string;
+  cooperation_source: string;
+};
+
+export type AnalyzedRow = ResourceImportRow & {
+  action: "create" | "update" | "error";
+  accountId: string;
+  platformUid: string;
+  cooperationSource: string;
+};
+
+export const RESOURCE_IMPORT_SYNC_ENRICH_ROWS = 100;
+const RESOURCE_IMPORT_DB_BATCH_SIZE = 100;
+
+export function identityKey(platform: string, value: string) {
+  return `${platform.trim().toLocaleLowerCase("zh-CN")}|${value
+    .trim()
+    .toLocaleLowerCase("zh-CN")}`;
+}
+
+export async function loadAccounts() {
+  await ensureSchema();
+  return getRawDb()
+    .prepare(
+      `SELECT id, platform, platform_uid, public_account_id, nickname,
+              profile_url, ip_location, followers, gender, bio, tags,
+              cooperation_source
+       FROM accounts`,
+    )
+    .all<AccountRow>();
+}
+
+async function mapConcurrent<T, R>(
+  items: T[],
+  limit: number,
+  worker: (item: T) => Promise<R>,
+) {
+  const results = new Array<R>(items.length);
+  let cursor = 0;
+  await Promise.all(
+    Array.from({ length: Math.min(limit, items.length) }, async () => {
+      while (cursor < items.length) {
+        const index = cursor;
+        cursor += 1;
+        results[index] = await worker(items[index]);
+      }
+    }),
+  );
+  return results;
+}
+
+export function mergeExistingFields(
+  rows: ResourceImportRow[],
+  accounts: AccountRow[],
+) {
+  const existingByProfile = new Map<string, AccountRow>();
+  for (const account of accounts) {
+    const profileUrl = normalizeProfileUrl(account.profile_url || "");
+    if (profileUrl) {
+      existingByProfile.set(identityKey(account.platform, profileUrl), account);
+    }
+  }
+  return rows.map((row) => {
+    if (
+      row.errors.length > 0 ||
+      !row.profileUrl ||
+      !["小红书", "抖音"].includes(row.platform)
+    ) {
+      return row;
+    }
+    const existing = existingByProfile.get(identityKey(row.platform, row.profileUrl));
+    const existingNickname =
+      existing?.nickname && existing.nickname !== "待识别账号"
+        ? existing.nickname
+        : "";
+    const existingIpLocation =
+      existing?.ip_location && existing.ip_location !== "待识别"
+        ? existing.ip_location
+        : "";
+    const existingGender: ResourceImportRow["gender"] =
+      existing?.gender === "男" || existing?.gender === "女"
+        ? existing.gender
+        : "";
+    return {
+      ...row,
+      nickname: row.nickname || existingNickname,
+      publicAccountId: row.publicAccountId || existing?.public_account_id || "",
+      ipLocation: row.ipLocation || existingIpLocation,
+      followers: row.followersResolved
+        ? row.followers
+        : Number(existing?.followers || 0),
+      followersResolved:
+        row.followersResolved || Number(existing?.followers || 0) > 0,
+      gender: row.gender || existingGender,
+      bio: row.bio || existing?.bio || "",
+      tags: row.tags.length > 0
+        ? row.tags
+        : (existing?.tags || "")
+            .split(/[,，、;；|]/)
+            .map((item) => item.trim())
+            .filter(Boolean)
+            .slice(0, 5),
+    };
+  });
+}
+
+export async function enrichRows(
+  rows: ResourceImportRow[],
+  accounts: AccountRow[],
+) {
+  const mcpConfig = resolveCollectionMcpConfig(
+    getRuntimeEnv() as unknown as CollectionMcpBindings,
+  );
+  const baselineRows = mergeExistingFields(rows, accounts);
+  return mapConcurrent(baselineRows, 4, async (baseline) => {
+    const row = baseline;
+    if (
+      row.errors.length > 0 ||
+      !row.profileUrl ||
+      !["小红书", "抖音"].includes(row.platform)
+    ) {
+      return row;
+    }
+    if (resourceImportMissingFields(baseline).length === 0) {
+      return baseline;
+    }
+
+    let details: {
+      nickname: string | null;
+      redId: string | null;
+      followers: number | null;
+      ipLocation: string | null;
+      gender: "" | "男" | "女";
+      bio: string;
+      recentNoteTitles: string[];
+      providerTags: string[];
+    } = await resolveProfileDetailsFromMcp(
+      row.profileUrl,
+      row.platform === "抖音" ? "抖音" : "小红书",
+      mcpConfig,
+    ).catch(() => ({
+      nickname: null,
+      redId: null,
+      followers: null,
+      ipLocation: null,
+      gender: "" as const,
+      bio: "",
+      recentNoteTitles: [],
+      providerTags: [],
+    }));
+    const mcpResult = {
+      nickname: baseline.nickname || details.nickname?.trim() || "",
+      publicAccountId: baseline.publicAccountId || details.redId?.trim() || "",
+      ipLocation: baseline.ipLocation || details.ipLocation?.trim() || "",
+      followersResolved: baseline.followersResolved || details.followers !== null,
+      gender: baseline.gender || details.gender,
+      bio: baseline.bio || details.bio,
+      tags: baseline.tags,
+    };
+    if (
+      row.platform === "小红书" &&
+      resourceImportMissingFields(mcpResult).length > 0
+    ) {
+      const publicDetails = await resolveXhsPublicAccountDetails(row.profileUrl).catch(
+        () => ({ nickname: null, redId: null, followers: null, ipLocation: null }),
+      );
+      details = {
+        ...details,
+        nickname: details.nickname || publicDetails.nickname,
+        redId: details.redId || publicDetails.redId,
+        followers: details.followers ?? publicDetails.followers,
+        ipLocation: details.ipLocation || publicDetails.ipLocation,
+      };
+    }
+    return {
+      ...baseline,
+      nickname: baseline.nickname || details.nickname?.trim() || "待识别账号",
+      publicAccountId: baseline.publicAccountId || details.redId?.trim() || "",
+      ipLocation: baseline.ipLocation || details.ipLocation?.trim() || "待识别",
+      followers: baseline.followersResolved
+        ? baseline.followers
+        : (details.followers ?? 0),
+      followersResolved:
+        baseline.followersResolved || details.followers !== null,
+      gender: baseline.gender || details.gender,
+      bio: baseline.bio || details.bio,
+      tags: baseline.tags,
+    };
+  });
+}
+
+export function analyzeRows(
+  rows: ResourceImportRow[],
+  accountRows: AccountRow[],
+) {
+  const profileMap = new Map<string, AccountRow>();
+  const publicIdMap = new Map<string, AccountRow>();
+  const platformUidMap = new Map<string, AccountRow>();
+  for (const account of accountRows) {
+    const profileUrl = normalizeProfileUrl(account.profile_url || "");
+    if (profileUrl) profileMap.set(identityKey(account.platform, profileUrl), account);
+    if (account.public_account_id) {
+      publicIdMap.set(identityKey(account.platform, account.public_account_id), account);
+    }
+    platformUidMap.set(identityKey(account.platform, account.platform_uid), account);
+  }
+
+  return rows.map<AnalyzedRow>((row) => {
+    const platformUid = resourcePlatformUid(row);
+    const profileMatch = row.profileUrl
+      ? profileMap.get(identityKey(row.platform, row.profileUrl))
+      : undefined;
+    const publicIdMatch = row.publicAccountId
+      ? publicIdMap.get(identityKey(row.platform, row.publicAccountId))
+      : undefined;
+    const uidMatch = platformUidMap.get(identityKey(row.platform, platformUid));
+    const matches = [profileMatch, publicIdMatch, uidMatch].filter(
+      (account): account is AccountRow => Boolean(account),
+    );
+    const matchedIds = [...new Set(matches.map((account) => account.id))];
+    const errors = [...row.errors];
+    if (matchedIds.length > 1) {
+      errors.push("账号主页和账号ID匹配到不同的现有账号，请先核对");
+    }
+    const existing = matchedIds.length === 1 ? matches[0] : undefined;
+    const accountId = existing?.id ?? `account-${crypto.randomUUID().slice(0, 12)}`;
+    const analyzed: AnalyzedRow = {
+      ...row,
+      errors,
+      action: errors.length > 0 ? "error" : existing ? "update" : "create",
+      accountId,
+      platformUid: existing?.platform_uid ?? platformUid,
+      cooperationSource: mergeCooperationSources(
+        existing?.cooperation_source ?? "",
+        row.cooperationSource,
+      ),
+    };
+    if (analyzed.action !== "error") {
+      const virtual: AccountRow = {
+        id: accountId,
+        platform: row.platform,
+        platform_uid: analyzed.platformUid,
+        public_account_id: row.publicAccountId || existing?.public_account_id || "",
+        nickname: row.nickname,
+        profile_url: row.profileUrl || existing?.profile_url || "",
+        ip_location: row.ipLocation || existing?.ip_location || "待识别",
+        followers: row.followers || existing?.followers || 0,
+        gender: row.gender || existing?.gender || "",
+        bio: row.bio || existing?.bio || "",
+        tags: (row.tags.length > 0
+          ? row.tags
+          : (existing?.tags || "").split(/[,，、;；|]/).filter(Boolean)
+        ).slice(0, 5).join(","),
+        cooperation_source: analyzed.cooperationSource,
+      };
+      if (row.profileUrl) profileMap.set(identityKey(row.platform, row.profileUrl), virtual);
+      if (row.publicAccountId) {
+        publicIdMap.set(identityKey(row.platform, row.publicAccountId), virtual);
+      }
+      platformUidMap.set(identityKey(row.platform, analyzed.platformUid), virtual);
+    }
+    return analyzed;
+  });
+}
+
+function statementForAnalyzedRow(
+  db: ReturnType<typeof getRawDb>,
+  row: AnalyzedRow,
+) {
+  return row.action === "update"
+    ? db
+        .prepare(
+          `UPDATE accounts SET
+             nickname = ?,
+             public_account_id = CASE WHEN ? != '' THEN ? ELSE public_account_id END,
+             profile_url = CASE WHEN ? != '' THEN ? ELSE profile_url END,
+             ip_location = CASE
+               WHEN ? != '' AND ? != '待识别' THEN ? ELSE ip_location END,
+             followers = CASE WHEN ? = 1 THEN ? ELSE followers END,
+             gender = CASE WHEN ? != '' THEN ? ELSE gender END,
+             bio = CASE WHEN ? != '' THEN ? ELSE bio END,
+             tags = CASE WHEN ? != '' THEN ? ELSE tags END,
+             cooperation_source = ?,
+             last_seen_at = CURRENT_TIMESTAMP
+           WHERE id = ?`,
+        )
+        .bind(
+          row.nickname || "待识别账号",
+          row.publicAccountId,
+          row.publicAccountId,
+          row.profileUrl,
+          row.profileUrl,
+          row.ipLocation,
+          row.ipLocation,
+          row.ipLocation,
+          row.followersResolved ? 1 : 0,
+          row.followers,
+          row.gender,
+          row.gender,
+          row.bio,
+          row.bio,
+          row.tags.join(","),
+          row.tags.join(","),
+          row.cooperationSource,
+          row.accountId,
+        )
+    : db
+        .prepare(
+          `INSERT INTO accounts
+            (id, platform, platform_uid, public_account_id, nickname,
+             profile_url, ip_location, followers, post_count, avg_views,
+             gender, bio, tags, cooperation_source)
+           VALUES (?, ?, ?, ?, ?, ?, ?, ?, 0, 0, ?, ?, ?, ?)`,
+        )
+        .bind(
+          row.accountId,
+          row.platform,
+          row.platformUid,
+          row.publicAccountId,
+          row.nickname || "待识别账号",
+          row.profileUrl,
+          row.ipLocation || "待识别",
+          row.followers,
+          row.gender,
+          row.bio,
+          row.tags.join(","),
+          row.cooperationSource,
+        );
+}
+
+export async function writeAnalyzedRows(rows: AnalyzedRow[]) {
+  const db = getRawDb();
+  const statements = rows
+    .filter((row) => row.action !== "error")
+    .map((row) => statementForAnalyzedRow(db, row));
+  for (let index = 0; index < statements.length; index += RESOURCE_IMPORT_DB_BATCH_SIZE) {
+    await db.batch(statements.slice(index, index + RESOURCE_IMPORT_DB_BATCH_SIZE));
+  }
+}
--- a/lib/runtime-env.ts
+++ b/lib/runtime-env.ts
@@ -20,6 +20,12 @@ export type RuntimeEnv = {
  AI_TOOL_CENTER_MCP_KEY?: string;
  COLLECTION_MCP_URL?: string;
  COLLECTION_MCP_KEY?: string;
+  WECOM_CORP_ID?: string;
+  WECOM_AGENT_ID?: string;
+  WECOM_SECRET?: string;
+  WECOM_ROBOT_WEBHOOK?: string;
+  WECOM_NOTIFY_DUE_DAYS?: string;
+  WECOM_NOTIFY_ENABLED?: string;
  SEED_DEMO_DATA?: string;
  ENABLE_SCHEDULER?: string;
 };
--- a/lib/scheduler.ts
+++ b/lib/scheduler.ts
@@ -8,6 +8,11 @@ import {
 } from "./mcp-collection-client";
 import { ensureSchema, getRawDb } from "./mvp-db";
 import { getRuntimeEnv, isEnabled } from "./runtime-env";
+import {
+  resolveWecomConfig,
+  type WecomBindings,
+} from "./wecom-client";
+import { runDueSoonWecomNotifications } from "./wecom-notifier-service";

 declare global {
  var __kocLoopScheduler: ScheduledTask | undefined;
@@ -17,14 +22,31 @@ async function runDailyJob() {
  await withDatabaseLock("koc-loop-daily-collection", 0, async () => {
    await ensureSchema();
    const db = getRawDb();
+    const env = getRuntimeEnv();
    const config = resolveCollectionMcpConfig(
-      getRuntimeEnv() as unknown as CollectionMcpBindings,
+      env as unknown as CollectionMcpBindings,
    );
    const collections = await runScheduledCollections(db, Date.now(), config);
    const accounts = await backfillAccountProfiles(db, config, 10);
+    let wecom: Awaited<ReturnType<typeof runDueSoonWecomNotifications>> | null =
+      null;
+    if (isEnabled(env.WECOM_NOTIFY_ENABLED, true)) {
+      const wecomConfig = resolveWecomConfig(env as unknown as WecomBindings);
+      if (
+        wecomConfig.robotWebhook ||
+        (wecomConfig.corpId && wecomConfig.agentId && wecomConfig.secret)
+      ) {
+        try {
+          wecom = await runDueSoonWecomNotifications(db, wecomConfig);
+        } catch (error) {
+          console.error("[KOC LOOP] wecom notify failed", error);
+        }
+      }
+    }
    console.info("[KOC LOOP] daily scheduler completed", {
      collections,
      accounts,
+      wecom,
    });
  });
 }
--- a/lib/task-service.ts
+++ b/lib/task-service.ts
@@ -10,6 +10,8 @@ export type CreateDistributionTaskInput = {
  name: string;
  brand: string;
  dueAt: string;
+  platform?: "小红书" | "抖音";
+  contentFormat?: "image_text" | "video";
 };

 export type CreateScreenshotTaskInput = {
@@ -33,6 +35,8 @@ export type DistributionTaskCreation = {
  sheetId: string;
  sheetName: string;
  sourceUrl: string;
+  platform: "小红书" | "抖音";
+  contentFormat: "image_text" | "video";
 };

 export type ScreenshotTaskCreation = {
@@ -55,6 +59,8 @@ type TaskRow = {
  source_url: string;
  source_sheet_id: string;
  source_sheet_name: string;
+  platform: "小红书" | "抖音";
+  content_format: "image_text" | "video";
 };

 function normalizedValue(value: string) {
@@ -99,17 +105,27 @@ async function findExistingTask(
  return db
    .prepare(
      `SELECT id, share_token, name, brand, due_at, quantity,
-              source_url, source_sheet_id, source_sheet_name
+              source_url, source_sheet_id, source_sheet_name,
+              platform, content_format
       FROM tasks
       WHERE name = ?
         AND brand = ?
         AND due_at = ?
         AND source_url = ?
+         AND platform = ?
+         AND content_format = ?
         AND status IN ('active', 'importing')
       ORDER BY created_at DESC
       LIMIT 1`,
    )
-    .bind(input.name, input.brand, input.dueAt, sourceUrl)
+    .bind(
+      input.name,
+      input.brand,
+      input.dueAt,
+      sourceUrl,
+      input.platform ?? "小红书",
+      input.contentFormat ?? "image_text",
+    )
    .first<TaskRow>();
 }

@@ -124,9 +140,10 @@ async function insertTaskFromSource(
    .prepare(
      `INSERT INTO tasks
        (id, name, brand, quantity, claimed_quantity, due_at, status,
+         platform, content_format,
         source_url, source_sheet_id, source_sheet_name, source_synced_at,
         share_token)
-       VALUES (?, ?, ?, ?, 0, ?, 'importing', ?, ?, ?, ?, ?)`,
+       VALUES (?, ?, ?, ?, 0, ?, 'importing', ?, ?, ?, ?, ?, ?, ?)`,
    )
    .bind(
      taskId,
@@ -134,6 +151,8 @@ async function insertTaskFromSource(
      input.brand,
      source.rows.length,
      input.dueAt,
+      input.platform ?? "小红书",
+      input.contentFormat ?? "image_text",
      source.url,
      source.sheetId,
      source.sheetName,
@@ -149,11 +168,15 @@ async function insertTaskFromSource(
        ...image,
        key: `content-assets/${taskId}/${contentId}/${image.index}`,
      }));
+      const videoAssets = row.videos.map((video) => ({
+        ...video,
+        key: `content-videos/${taskId}/${contentId}/${video.index}`,
+      }));
      return db
        .prepare(
          `INSERT INTO contents
-            (id, task_id, title, body, image_assets, status, source, source_row)
-           VALUES (?, ?, ?, ?, ?, 'available', ?, ?)`,
+            (id, task_id, title, body, image_assets, video_assets, status, source, source_row)
+           VALUES (?, ?, ?, ?, ?, ?, 'available', ?, ?)`,
        )
        .bind(
          contentId,
@@ -161,6 +184,7 @@ async function insertTaskFromSource(
          row.title,
          row.body,
          JSON.stringify(imageAssets),
+          JSON.stringify(videoAssets),
          `飞书 · ${source.sheetName}`,
          row.sourceRow,
        );
@@ -189,11 +213,13 @@ export async function createDistributionTask(
  options: { deduplicate?: boolean } = {},
 ): Promise<DistributionTaskCreation> {
  await ensureSchema();
-  const input = {
+  const input: Required<CreateDistributionTaskInput> = {
    feishuUrl: normalizedFeishuUrl(rawInput.feishuUrl),
    name: normalizedValue(rawInput.name),
    brand: normalizedValue(rawInput.brand),
    dueAt: normalizedDueDate(rawInput.dueAt),
+    platform: rawInput.platform === "抖音" ? "抖音" : "小红书",
+    contentFormat: rawInput.contentFormat === "video" ? "video" : "image_text",
  };
  if (!input.feishuUrl || !input.name || !input.brand) {
    throw new Error("请补全飞书链接、任务名称和品牌/项目");
@@ -213,11 +239,19 @@ export async function createDistributionTask(
        sheetId: existing.source_sheet_id,
        sheetName: existing.source_sheet_name,
        sourceUrl: existing.source_url,
+        platform: existing.platform,
+        contentFormat: existing.content_format,
      };
    }
  }

  const source = await readFeishuSource(input.feishuUrl, bindings);
+  if (
+    input.contentFormat === "video" &&
+    source.rows.some((row) => row.videos.length === 0)
+  ) {
+    throw new Error("视频任务中存在未识别到视频的内容行，请检查飞书“视频”列");
+  }
  const inserted = await insertTaskFromSource(source, input);
  return {
    created: true,
@@ -230,6 +264,8 @@ export async function createDistributionTask(
    sheetId: source.sheetId,
    sheetName: source.sheetName,
    sourceUrl: source.url,
+    platform: input.platform,
+    contentFormat: input.contentFormat,
  };
 }

@@ -287,8 +323,8 @@ export async function createScreenshotTask(
      db
        .prepare(
          `INSERT INTO contents
-            (id, task_id, title, body, image_assets, status, source, source_row)
-           VALUES (?, ?, ?, ?, ?, 'available', '截图回收任务', ?)`,
+            (id, task_id, title, body, image_assets, video_assets, status, source, source_row)
+           VALUES (?, ?, ?, ?, ?, '[]', 'available', '截图回收任务', ?)`,
        )
        .bind(
          uid("content"),
--- a/lib/video-file.ts
+++ b/lib/video-file.ts
@@ -0,0 +1,45 @@
+const MP4_BRANDS = new Set([
+  "avc1",
+  "dash",
+  "isom",
+  "M4A ",
+  "M4B ",
+  "M4P ",
+  "M4V ",
+  "mp41",
+  "mp42",
+  "MSNV",
+]);
+
+function fourCharacters(bytes: Uint8Array, offset: number) {
+  return String.fromCharCode(...bytes.slice(offset, offset + 4));
+}
+
+function isMp4Brand(brand: string) {
+  return (
+    MP4_BRANDS.has(brand) ||
+    /^iso[2-9]$/.test(brand) ||
+    /^3g[2p]$/.test(brand.slice(0, 3))
+  );
+}
+
+/** Rejects HTML/JSON/error payloads and non-MP4 containers before download. */
+export function hasMp4FileSignature(input: ArrayBuffer | Uint8Array) {
+  const bytes = input instanceof Uint8Array ? input : new Uint8Array(input);
+  if (bytes.byteLength < 12 || fourCharacters(bytes, 4) !== "ftyp") {
+    return false;
+  }
+  const declaredSize = new DataView(
+    bytes.buffer,
+    bytes.byteOffset,
+    bytes.byteLength,
+  ).getUint32(0);
+  const boxEnd = Math.min(
+    bytes.byteLength,
+    declaredSize >= 12 ? declaredSize : bytes.byteLength,
+  );
+  for (let offset = 8; offset + 4 <= boxEnd; offset += 4) {
+    if (isMp4Brand(fourCharacters(bytes, offset))) return true;
+  }
+  return false;
+}
--- a/lib/wecom-client.ts
+++ b/lib/wecom-client.ts
@@ -0,0 +1,460 @@
+const WECOM_API_ORIGIN = "https://qyapi.weixin.qq.com";
+const DEFAULT_DUE_DAYS = 3;
+
+export type WecomBindings = {
+  WECOM_CORP_ID?: string;
+  WECOM_AGENT_ID?: string;
+  WECOM_SECRET?: string;
+  WECOM_ROBOT_WEBHOOK?: string;
+  WECOM_NOTIFY_DUE_DAYS?: string;
+};
+
+export type WecomConfig = {
+  corpId: string;
+  agentId: string;
+  secret: string;
+  robotWebhook: string;
+  dueDays: number;
+};
+
+type FetchLike = typeof fetch;
+
+type WecomEnvelope = {
+  errcode?: number;
+  errmsg?: string;
+  access_token?: string;
+  expires_in?: number;
+  invaliduser?: string;
+  follow_user?: Array<{ userid: string }>;
+  external_userid?: string[];
+  external_contact?: {
+    external_userid?: string;
+    name?: string;
+    avatar?: string;
+    corp_fullname?: string;
+  };
+  next_cursor?: string;
+  group_chat_list?: Array<{ chat_id?: string; status?: number }>;
+  group_chat?: {
+    name?: string;
+    owner?: string;
+    member_count?: number;
+  };
+  fail_list?: string[];
+  msgid?: string;
+};
+
+export type WecomExternalContact = {
+  externalUserId: string;
+  name: string;
+  avatar: string;
+  corpName: string;
+  ownerUserId: string;
+};
+
+type CachedAccessToken = {
+  corpId: string;
+  secret: string;
+  token: string;
+  expiresAt: number;
+};
+
+let cachedAccessToken: CachedAccessToken | null = null;
+
+export class WecomClientError extends Error {
+  status: number;
+
+  constructor(message: string, status = 502) {
+    super(message);
+    this.name = "WecomClientError";
+    this.status = status;
+  }
+}
+
+function bindingValue(value: unknown) {
+  return String(value ?? "").trim();
+}
+
+function safeMessage(value: unknown) {
+  return String(value ?? "").trim().slice(0, 240);
+}
+
+export function resolveWecomConfig(bindings: WecomBindings): WecomConfig {
+  return {
+    corpId: bindingValue(bindings.WECOM_CORP_ID),
+    agentId: bindingValue(bindings.WECOM_AGENT_ID),
+    secret: bindingValue(bindings.WECOM_SECRET),
+    robotWebhook: bindingValue(bindings.WECOM_ROBOT_WEBHOOK),
+    dueDays: parseDueDays(bindings.WECOM_NOTIFY_DUE_DAYS),
+  };
+}
+
+function parseDueDays(value: string | undefined) {
+  const parsed = Number(bindingValue(value));
+  if (!Number.isFinite(parsed) || parsed < 1) return DEFAULT_DUE_DAYS;
+  return Math.min(30, Math.floor(parsed));
+}
+
+function hasAppCredentials(config: WecomConfig) {
+  return Boolean(config.corpId && config.agentId && config.secret);
+}
+
+async function readEnvelope(
+  response: Response,
+  fallbackMessage: string,
+): Promise<WecomEnvelope> {
+  const text = await response.text();
+  try {
+    return JSON.parse(text) as WecomEnvelope;
+  } catch {
+    throw new WecomClientError(
+      `${fallbackMessage}（企业微信返回了非 JSON 响应）`,
+      502,
+    );
+  }
+}
+
+function ensureOk(
+  payload: WecomEnvelope,
+  fallbackMessage: string,
+) {
+  const code = Number(payload.errcode ?? 0);
+  if (code === 0) return;
+  const message = safeMessage(payload.errmsg) || fallbackMessage;
+  if (code === 40014 || code === 42001) {
+    throw new WecomClientError(`企业微信 access_token 无效：${message}`, 401);
+  }
+  throw new WecomClientError(`${fallbackMessage}（${message}）`, 502);
+}
+
+async function fetchAccessToken(
+  config: WecomConfig,
+  fetchImpl: FetchLike,
+) {
+  if (
+    cachedAccessToken?.corpId === config.corpId &&
+    cachedAccessToken?.secret === config.secret &&
+    cachedAccessToken.expiresAt > Date.now() + 60_000
+  ) {
+    return cachedAccessToken.token;
+  }
+  const url = new URL(`${WECOM_API_ORIGIN}/cgi-bin/gettoken`);
+  url.searchParams.set("corpid", config.corpId);
+  url.searchParams.set("corpsecret", config.secret);
+  const response = await fetchImpl(url.toString(), {
+    method: "GET",
+    signal: AbortSignal.timeout(12_000),
+  });
+  const payload = await readEnvelope(response, "获取企业微信 access_token 失败");
+  if (!response.ok) {
+    throw new WecomClientError(
+      `获取企业微信 access_token 失败（HTTP ${response.status}）`,
+      502,
+    );
+  }
+  ensureOk(payload, "获取企业微信 access_token 失败");
+  const token = bindingValue(payload.access_token);
+  if (!token) {
+    throw new WecomClientError("企业微信未返回有效 access_token", 502);
+  }
+  cachedAccessToken = {
+    corpId: config.corpId,
+    secret: config.secret,
+    token,
+    expiresAt:
+      Date.now() + Math.max(300, Number(payload.expires_in) || 7_200) * 1_000,
+  };
+  return token;
+}
+
+export async function sendWecomRobotMessage(
+  content: string,
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<void> {
+  if (!config.robotWebhook) {
+    console.warn("[KOC LOOP] wecom robot webhook not configured, skipping");
+    return;
+  }
+  const response = await fetchImpl(config.robotWebhook, {
+    method: "POST",
+    headers: { "Content-Type": "application/json; charset=utf-8" },
+    body: JSON.stringify({
+      msgtype: "text",
+      text: { content },
+    }),
+    signal: AbortSignal.timeout(10_000),
+  });
+  const payload = await readEnvelope(response, "企业微信群机器人推送失败");
+  if (!response.ok) {
+    throw new WecomClientError(
+      `企业微信群机器人推送失败（HTTP ${response.status}）`,
+      502,
+    );
+  }
+  ensureOk(payload, "企业微信群机器人推送失败");
+}
+
+export async function sendWecomAppMessage(
+  externalUserIds: string[],
+  content: string,
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<{ sent: number; failed: number; skipped: boolean }> {
+  const normalized = externalUserIds
+    .map((id) => bindingValue(id))
+    .filter((id) => id.length > 0);
+  if (normalized.length === 0) {
+    return { sent: 0, failed: 0, skipped: true };
+  }
+  if (!hasAppCredentials(config)) {
+    return { sent: 0, failed: 0, skipped: true };
+  }
+  const token = await fetchAccessToken(config, fetchImpl);
+  const url = `${WECOM_API_ORIGIN}/cgi-bin/message/send?access_token=${encodeURIComponent(token)}`;
+  const response = await fetchImpl(url, {
+    method: "POST",
+    headers: { "Content-Type": "application/json; charset=utf-8" },
+    body: JSON.stringify({
+      touser: normalized.join("|"),
+      msgtype: "text",
+      agentid: Number(config.agentId),
+      text: { content },
+    }),
+    signal: AbortSignal.timeout(12_000),
+  });
+  const payload = await readEnvelope(response, "企业微信应用消息推送失败");
+  if (!response.ok) {
+    throw new WecomClientError(
+      `企业微信应用消息推送失败（HTTP ${response.status}）`,
+      502,
+    );
+  }
+  ensureOk(payload, "企业微信应用消息推送失败");
+  const invalid = bindingValue(payload.invaliduser).split("|").filter(Boolean);
+  return {
+    sent: Math.max(0, normalized.length - invalid.length),
+    failed: invalid.length,
+    skipped: false,
+  };
+}
+
+export function clearWecomAccessTokenCacheForTests() {
+  cachedAccessToken = null;
+}
+
+export type WecomGroupChat = {
+  chatId: string;
+  name: string;
+  ownerUserId: string;
+  memberCount: number;
+  status: number;
+};
+
+export async function listCustomerGroupChats(
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<WecomGroupChat[]> {
+  if (!hasAppCredentials(config)) return [];
+  const token = await fetchAccessToken(config, fetchImpl);
+
+  const chatIds: Array<{ chatId: string; status: number }> = [];
+  let cursor = "";
+  do {
+    const listUrl = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/groupchat/list?access_token=${encodeURIComponent(token)}`;
+    const listResp = await fetchImpl(listUrl, {
+      method: "POST",
+      headers: { "Content-Type": "application/json; charset=utf-8" },
+      body: JSON.stringify({ limit: 100, cursor }),
+      signal: AbortSignal.timeout(12_000),
+    });
+    const listPayload = await readEnvelope(listResp, "获取企业微信客户群列表失败");
+    if (!listResp.ok) {
+      throw new WecomClientError(
+        `获取企业微信客户群列表失败（HTTP ${listResp.status}）`,
+        502,
+      );
+    }
+    ensureOk(listPayload, "获取企业微信客户群列表失败");
+    for (const chat of listPayload.group_chat_list ?? []) {
+      const chatId = bindingValue(chat.chat_id);
+      if (chatId) chatIds.push({ chatId, status: Number(chat.status ?? 0) });
+    }
+    cursor = bindingValue(listPayload.next_cursor);
+  } while (cursor);
+
+  const groups: WecomGroupChat[] = [];
+  for (const { chatId, status } of chatIds) {
+    if (status !== 0) continue;
+    const getUrl = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/groupchat/get?access_token=${encodeURIComponent(token)}`;
+    const getResp = await fetchImpl(getUrl, {
+      method: "POST",
+      headers: { "Content-Type": "application/json; charset=utf-8" },
+      body: JSON.stringify({ chat_id: chatId }),
+      signal: AbortSignal.timeout(12_000),
+    });
+    const getPayload = await readEnvelope(getResp, "获取企业微信客户群详情失败");
+    if (!getResp.ok) continue;
+    if (Number(getPayload.errcode ?? 0) !== 0) continue;
+    const detail = getPayload.group_chat;
+    groups.push({
+      chatId,
+      name: bindingValue(detail?.name),
+      ownerUserId: bindingValue(detail?.owner),
+      memberCount: Number(detail?.member_count ?? 0),
+      status,
+    });
+  }
+  return groups;
+}
+
+export type WecomGroupMsgLink = {
+  title: string;
+  desc?: string;
+  url: string;
+  picurl?: string;
+};
+
+export async function createGroupMsgTemplate(
+  config: WecomConfig,
+  params: {
+    sender: string;
+    chatIdList: string[];
+    text?: string;
+    link?: WecomGroupMsgLink;
+  },
+  fetchImpl: FetchLike = fetch,
+): Promise<{ msgid: string; failList: string[] }> {
+  if (!hasAppCredentials(config)) {
+    throw new WecomClientError(
+      "企业微信自建应用未配置（WECOM_CORP_ID / WECOM_AGENT_ID / WECOM_SECRET）",
+      400,
+    );
+  }
+  const chatIdList = params.chatIdList.map(bindingValue).filter(Boolean);
+  if (chatIdList.length === 0) {
+    throw new WecomClientError("客户群列表为空", 400);
+  }
+  const sender = bindingValue(params.sender);
+  if (!sender) {
+    throw new WecomClientError("发送成员（群主 userid）不能为空", 400);
+  }
+  const text = String(params.text ?? "").trim();
+  const link = params.link;
+  if (!text && !link) {
+    throw new WecomClientError("文本与图文附件不能同时为空", 400);
+  }
+
+  const token = await fetchAccessToken(config, fetchImpl);
+  const url = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/add_msg_template?access_token=${encodeURIComponent(token)}`;
+  const response = await fetchImpl(url, {
+    method: "POST",
+    headers: { "Content-Type": "application/json; charset=utf-8" },
+    body: JSON.stringify({
+      chat_type: "group",
+      chat_id_list: chatIdList,
+      sender,
+      ...(text ? { text: { content: text } } : {}),
+      ...(link
+        ? {
+            attachments: [
+              {
+                msgtype: "link",
+                link: {
+                  title: link.title,
+                  desc: link.desc ?? "",
+                  url: link.url,
+                  picurl: link.picurl ?? "",
+                },
+              },
+            ],
+          }
+        : {}),
+    }),
+    signal: AbortSignal.timeout(12_000),
+  });
+  const payload = await readEnvelope(response, "创建企业群发失败");
+  if (!response.ok) {
+    throw new WecomClientError(
+      `创建企业群发失败（HTTP ${response.status}）`,
+      502,
+    );
+  }
+  ensureOk(payload, "创建企业群发失败");
+  return {
+    msgid: bindingValue(payload.msgid),
+    failList: (payload.fail_list ?? []).filter(Boolean),
+  };
+}
+
+export async function listExternalContacts(
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<WecomExternalContact[]> {
+  if (!hasAppCredentials(config)) return [];
+  const token = await fetchAccessToken(config, fetchImpl);
+
+  const followUrl = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/get_follow_user_list?access_token=${encodeURIComponent(token)}`;
+  const followResp = await fetchImpl(followUrl, {
+    method: "GET",
+    signal: AbortSignal.timeout(12_000),
+  });
+  const followPayload = await readEnvelope(
+    followResp,
+    "获取企业微信跟进人列表失败",
+  );
+  if (!followResp.ok) {
+    throw new WecomClientError(
+      `获取企业微信跟进人列表失败（HTTP ${followResp.status}）`,
+      502,
+    );
+  }
+  ensureOk(followPayload, "获取企业微信跟进人列表失败");
+  const ownerUserIds = (followPayload.follow_user ?? [])
+    .map((user) => bindingValue(user.userid))
+    .filter(Boolean);
+
+  const ownerToExternalIds: Array<{ externalUserId: string; ownerUserId: string }> = [];
+  for (const ownerUserId of ownerUserIds) {
+    const listUrl = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/list?access_token=${encodeURIComponent(token)}&userid=${encodeURIComponent(ownerUserId)}`;
+    const listResp = await fetchImpl(listUrl, {
+      method: "GET",
+      signal: AbortSignal.timeout(12_000),
+    });
+    const listPayload = await readEnvelope(
+      listResp,
+      "获取企业微信外部联系人列表失败",
+    );
+    if (!listResp.ok) continue;
+    if (Number(listPayload.errcode ?? 0) !== 0) continue;
+    const externalIds = (listPayload.external_userid ?? []).filter(Boolean);
+    for (const externalUserId of externalIds) {
+      ownerToExternalIds.push({ externalUserId, ownerUserId });
+    }
+  }
+
+  const contacts: WecomExternalContact[] = [];
+  for (const { externalUserId, ownerUserId } of ownerToExternalIds) {
+    const getUrl = `${WECOM_API_ORIGIN}/cgi-bin/externalcontact/get?access_token=${encodeURIComponent(token)}&external_userid=${encodeURIComponent(externalUserId)}`;
+    const getResp = await fetchImpl(getUrl, {
+      method: "GET",
+      signal: AbortSignal.timeout(12_000),
+    });
+    const getPayload = await readEnvelope(
+      getResp,
+      "获取企业微信外部联系人详情失败",
+    );
+    if (!getResp.ok) continue;
+    if (Number(getPayload.errcode ?? 0) !== 0) continue;
+    const info = getPayload.external_contact;
+    if (!info) continue;
+    contacts.push({
+      externalUserId,
+      name: bindingValue(info.name),
+      avatar: bindingValue(info.avatar),
+      corpName: bindingValue(info.corp_fullname) || bindingValue(info.name),
+      ownerUserId,
+    });
+  }
+  return contacts;
+}
--- a/lib/wecom-group-push-service.ts
+++ b/lib/wecom-group-push-service.ts
@@ -0,0 +1,326 @@
+import type { DatabaseClient } from "./database";
+import { getRuntimeEnv } from "./runtime-env";
+import {
+  createGroupMsgTemplate,
+  listCustomerGroupChats,
+  WecomClientError,
+  type WecomConfig,
+} from "./wecom-client";
+
+type FetchLike = typeof fetch;
+
+export const WECOM_GROUP_BATCH_LIMIT = 2000;
+const TEXT_MAX_BYTES = 4000;
+const LINK_TITLE_MAX_BYTES = 128;
+const LINK_DESC_MAX_BYTES = 512;
+const LINK_URL_MAX_BYTES = 2048;
+
+export type TaskPushRow = {
+  id: string;
+  name: string;
+  brand: string;
+  quantity: number;
+  due_at: string;
+  task_type: string;
+  platform: string;
+  content_format: string;
+  share_token: string | null;
+};
+
+export type TaskPushMessage = {
+  text: string;
+  link: { title: string; desc: string; url: string } | null;
+};
+
+export type GroupChatRow = {
+  chat_id: string;
+  name: string;
+  owner_user_id: string;
+  member_count: number;
+  status: number;
+};
+
+export type GroupPushResult = {
+  sender: string;
+  msgid: string;
+  chatCount: number;
+  failList: string[];
+};
+
+export type GroupPushRecord = {
+  id: string;
+  taskId: string;
+  msgid: string;
+  sender: string;
+  chatIds: string[];
+  failList: string[];
+  createdAt: string;
+};
+
+function bindingValue(value: unknown) {
+  return String(value ?? "").trim();
+}
+
+export function trimToBytes(value: string, maxBytes: number) {
+  let output = "";
+  let used = 0;
+  for (const char of Array.from(value)) {
+    const size = Buffer.byteLength(char, "utf8");
+    if (used + size > maxBytes) break;
+    output += char;
+    used += size;
+  }
+  return output;
+}
+
+export function buildTaskGroupPushMessage(
+  task: TaskPushRow,
+  portalUrl: string,
+): TaskPushMessage {
+  const portal = portalUrl.trim().replace(/\/$/, "");
+  const shareToken = bindingValue(task.share_token);
+  const claimUrl = portal && shareToken ? `${portal}/?task=${shareToken}` : "";
+  const formatLabel =
+    task.task_type === "screenshot_collect"
+      ? "截图回收"
+      : task.content_format === "video"
+        ? "视频"
+        : "图文";
+  const lines = [
+    `【新任务】${task.name}`,
+    `品牌：${task.brand}｜数量：${task.quantity} 份｜截止：${task.due_at}`,
+    `平台：${task.platform}｜形式：${formatLabel}`,
+    claimUrl ? `领取链接：${claimUrl}` : "领取链接生成失败，请联系管理员",
+  ];
+  return {
+    text: trimToBytes(lines.join("\n"), TEXT_MAX_BYTES),
+    link: claimUrl
+      ? {
+          title: trimToBytes(task.name, LINK_TITLE_MAX_BYTES),
+          desc: trimToBytes(
+            `品牌 ${task.brand} · ${task.quantity} 份 · 截止 ${task.due_at}`,
+            LINK_DESC_MAX_BYTES,
+          ),
+          url: claimUrl.slice(0, LINK_URL_MAX_BYTES),
+        }
+      : null,
+  };
+}
+
+export function groupChatsByOwner(chats: GroupChatRow[]) {
+  const grouped = new Map<string, string[]>();
+  for (const chat of chats) {
+    const owner = bindingValue(chat.owner_user_id);
+    if (!owner) continue;
+    grouped.set(owner, [...(grouped.get(owner) ?? []), chat.chat_id]);
+  }
+  return [...grouped.entries()].map(([ownerUserId, chatIds]) => ({
+    ownerUserId,
+    chatIds,
+  }));
+}
+
+export async function syncGroupChats(
+  db: DatabaseClient,
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<GroupChatRow[]> {
+  if (!(config.corpId && config.agentId && config.secret)) {
+    throw new WecomClientError(
+      "企业微信自建应用未配置（WECOM_CORP_ID / WECOM_AGENT_ID / WECOM_SECRET）",
+      400,
+    );
+  }
+  const groups = await listCustomerGroupChats(config, fetchImpl);
+  const syncedAt = new Date().toISOString();
+  await db.prepare("DELETE FROM wecom_group_chats").run();
+  const inserts = groups.map((group) =>
+    db
+      .prepare(
+        `INSERT INTO wecom_group_chats
+          (chat_id, name, owner_user_id, member_count, status, synced_at)
+         VALUES (?, ?, ?, ?, ?, ?)`,
+      )
+      .bind(
+        group.chatId,
+        group.name,
+        group.ownerUserId,
+        group.memberCount,
+        group.status,
+        syncedAt,
+      ),
+  );
+  if (inserts.length > 0) await db.batch(inserts);
+  return groups.map((group) => ({
+    chat_id: group.chatId,
+    name: group.name,
+    owner_user_id: group.ownerUserId,
+    member_count: group.memberCount,
+    status: group.status,
+  }));
+}
+
+export async function listGroupChatRows(db: DatabaseClient) {
+  const result = await db
+    .prepare(
+      `SELECT chat_id, name, owner_user_id, member_count, status
+       FROM wecom_group_chats
+       ORDER BY member_count DESC`,
+    )
+    .all<GroupChatRow>();
+  return result.results;
+}
+
+export async function listGroupPushes(
+  db: DatabaseClient,
+  taskId?: string,
+): Promise<GroupPushRecord[]> {
+  const filter = taskId ? "WHERE task_id = ?" : "";
+  const statement = db
+    .prepare(
+      `SELECT id, task_id, msgid, sender, chat_ids, fail_list, created_at
+       FROM wecom_group_pushes
+       ${filter}
+       ORDER BY created_at DESC
+       LIMIT 50`,
+    )
+    .bind(...(taskId ? [taskId] : []));
+  const result = await statement.all<{
+    id: string;
+    task_id: string;
+    msgid: string;
+    sender: string;
+    chat_ids: string;
+    fail_list: string;
+    created_at: string;
+  }>();
+  return result.results.map((row) => ({
+    id: row.id,
+    taskId: row.task_id,
+    msgid: row.msgid,
+    sender: row.sender,
+    chatIds: safeParseArray(row.chat_ids),
+    failList: safeParseArray(row.fail_list),
+    createdAt: row.created_at,
+  }));
+}
+
+export async function pushTaskToGroupChats(
+  db: DatabaseClient,
+  params: { taskId: string; chatIds: unknown; text?: unknown },
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+): Promise<{ results: GroupPushResult[]; linkUrl: string }> {
+  const taskId = bindingValue(params.taskId);
+  if (!taskId) throw new WecomClientError("缺少 taskId", 400);
+
+  const requested = [
+    ...new Set(
+      (Array.isArray(params.chatIds) ? params.chatIds : [])
+        .map((id) => bindingValue(id))
+        .filter(Boolean),
+    ),
+  ];
+  if (requested.length === 0) {
+    throw new WecomClientError("请选择要发送的客户群", 400);
+  }
+
+  const task = await db
+    .prepare(
+      `SELECT id, name, brand, quantity, due_at, task_type, platform,
+              content_format, share_token
+       FROM tasks WHERE id = ?`,
+    )
+    .bind(taskId)
+    .first<TaskPushRow>();
+  if (!task) throw new WecomClientError("任务不存在", 404);
+
+  const chatRows: GroupChatRow[] = [];
+  for (let index = 0; index < requested.length; index += 100) {
+    const chunk = requested.slice(index, index + 100);
+    const placeholders = chunk.map(() => "?").join(", ");
+    const result = await db
+      .prepare(
+        `SELECT chat_id, name, owner_user_id, member_count, status
+         FROM wecom_group_chats WHERE chat_id IN (${placeholders})`,
+      )
+      .bind(...chunk)
+      .all<GroupChatRow>();
+    chatRows.push(...result.results);
+  }
+  const byId = new Map(chatRows.map((row) => [row.chat_id, row]));
+  const missing = requested.filter((id) => !byId.has(id));
+  if (missing.length > 0) {
+    throw new WecomClientError(
+      `以下客户群未同步到平台，请先同步群列表：${missing.join("、")}`,
+      400,
+    );
+  }
+  const noOwner = requested
+    .map((id) => byId.get(id))
+    .filter((row) => row && !bindingValue(row.owner_user_id));
+  if (noOwner.length > 0) {
+    throw new WecomClientError(
+      `群「${noOwner.map((row) => row?.name || row?.chat_id).join("、")}」缺少群主信息，请重新同步群列表`,
+      400,
+    );
+  }
+
+  const portalUrl = bindingValue(getRuntimeEnv().KOC_PORTAL_URL);
+  const fallback = buildTaskGroupPushMessage(task, portalUrl);
+  const text = bindingValue(params.text) || fallback.text;
+  const link = fallback.link;
+  if (!text && !link) {
+    throw new WecomClientError("文本与图文附件不能同时为空", 400);
+  }
+
+  const results: GroupPushResult[] = [];
+  const groups = requested.map((id) => byId.get(id)!);
+  for (const { ownerUserId, chatIds } of groupChatsByOwner(groups)) {
+    for (
+      let index = 0;
+      index < chatIds.length;
+      index += WECOM_GROUP_BATCH_LIMIT
+    ) {
+      const batch = chatIds.slice(index, index + WECOM_GROUP_BATCH_LIMIT);
+      const sent = await createGroupMsgTemplate(
+        config,
+        { sender: ownerUserId, chatIdList: batch, text, link: link ?? undefined },
+        fetchImpl,
+      );
+      await db
+        .prepare(
+          `INSERT INTO wecom_group_pushes
+            (id, task_id, msgid, sender, chat_ids, text_content, link_url, fail_list)
+           VALUES (?, ?, ?, ?, ?, ?, ?, ?)`,
+        )
+        .bind(
+          `wgp-${crypto.randomUUID().slice(0, 12)}`,
+          taskId,
+          sent.msgid,
+          ownerUserId,
+          JSON.stringify(batch),
+          text,
+          link?.url ?? "",
+          JSON.stringify(sent.failList),
+        )
+        .run();
+      results.push({
+        sender: ownerUserId,
+        msgid: sent.msgid,
+        chatCount: batch.length,
+        failList: sent.failList,
+      });
+    }
+  }
+  return { results, linkUrl: link?.url ?? "" };
+}
+
+function safeParseArray(value: string) {
+  try {
+    const parsed = JSON.parse(value);
+    return Array.isArray(parsed) ? parsed.map((item) => String(item)) : [];
+  } catch {
+    return [];
+  }
+}
--- a/lib/wecom-notifier-service.ts
+++ b/lib/wecom-notifier-service.ts
@@ -0,0 +1,171 @@
+import type { DatabaseClient } from "./database";
+import { shanghaiDateFromTimestamp } from "./collection-service";
+import {
+  sendWecomAppMessage,
+  sendWecomRobotMessage,
+  type WecomConfig,
+} from "./wecom-client";
+import { getRuntimeEnv } from "./runtime-env";
+
+type FetchLike = typeof fetch;
+
+export type WecomNotifySummary = {
+  dueSoonAttempted: number;
+  dueSoonSent: number;
+  dueSoonFailed: number;
+  dueSoonSkipped: number;
+  digestSent: boolean;
+};
+
+type DueSoonRow = {
+  distribution_id: string;
+  partner_id: string;
+  partner_name: string;
+  wecom_external_user_id: string | null;
+  task_name: string;
+  due_at: string;
+  content_title: string;
+};
+
+export function computeDueCutoff(today: string, dueDays: number) {
+  const match = /^(\d{4})-(\d{2})-(\d{2})$/.exec(today);
+  if (!match) return today;
+  const [, y, m, d] = match;
+  const date = new Date(
+    Date.UTC(Number(y), Number(m) - 1, Number(d)) + dueDays * 24 * 60 * 60 * 1_000,
+  );
+  return date.toISOString().slice(0, 10);
+}
+
+export async function runDueSoonWecomNotifications(
+  db: DatabaseClient,
+  config: WecomConfig,
+  fetchImpl: FetchLike = fetch,
+  now: number = Date.now(),
+): Promise<WecomNotifySummary> {
+  const today = shanghaiDateFromTimestamp(now);
+  const cutoff = computeDueCutoff(today, config.dueDays);
+  const portalUrl = bindingValue(getRuntimeEnv().KOC_PORTAL_URL);
+
+  const result = await db
+    .prepare(
+      `SELECT
+        d.id AS distribution_id,
+        d.partner_id,
+        p.name AS partner_name,
+        p.wecom_external_user_id,
+        t.name AS task_name,
+        t.due_at,
+        c.title AS content_title
+       FROM distributions d
+       JOIN tasks t ON t.id = d.task_id
+       JOIN partners p ON p.id = d.partner_id
+       JOIN contents c ON c.id = d.content_id
+       WHERE (d.publish_url IS NULL OR d.publish_url = '')
+         AND t.due_at IS NOT NULL AND t.due_at != ''
+         AND t.due_at <= ?
+       ORDER BY t.due_at ASC, p.name ASC`,
+    )
+    .bind(cutoff)
+    .all<DueSoonRow>();
+
+  const grouped = new Map<
+    string,
+    {
+      partnerName: string;
+      externalUserId: string | null;
+      rows: DueSoonRow[];
+    }
+  >();
+  for (const row of result.results) {
+    const entry = grouped.get(row.partner_id) ?? {
+      partnerName: row.partner_name,
+      externalUserId: row.wecom_external_user_id,
+      rows: [],
+    };
+    entry.rows.push(row);
+    if (!entry.externalUserId && row.wecom_external_user_id) {
+      entry.externalUserId = row.wecom_external_user_id;
+    }
+    grouped.set(row.partner_id, entry);
+  }
+
+  let dueSoonAttempted = 0;
+  let dueSoonSent = 0;
+  let dueSoonFailed = 0;
+  let dueSoonSkipped = 0;
+  const digestTasks: string[] = [];
+
+  for (const [, entry] of grouped) {
+    dueSoonAttempted += 1;
+    const external = entry.externalUserId
+      ? [entry.externalUserId]
+      : [];
+    const lines = entry.rows.slice(0, 5).map((row) => {
+      return `· 《${truncate(row.task_name, 24)}》— ${truncate(row.content_title, 24)}（截止 ${row.due_at}）`;
+    });
+    const overflow =
+      entry.rows.length > 5 ? `\n…还有 ${entry.rows.length - 5} 条` : "";
+    const link = portalUrl ? `\n领取链接：${portalUrl}` : "";
+    const content =
+      `${entry.partnerName}，你有 ${entry.rows.length} 条内容待发布：\n${lines.join("\n")}${overflow}${link}`;
+    const appResult = await sendWecomAppMessage(
+      external,
+      content,
+      config,
+      fetchImpl,
+    ).catch((error: unknown) => {
+      console.warn(
+        "[KOC LOOP] wecom app message failed",
+        { partner: entry.partnerName, error: safeError(error) },
+      );
+      return null;
+    });
+    if (appResult === null) {
+      dueSoonFailed += 1;
+    } else if (appResult.skipped) {
+      dueSoonSkipped += 1;
+    } else {
+      dueSoonSent += 1;
+    }
+    const earliestDue = entry.rows[0]?.due_at ?? "";
+    digestTasks.push(
+      `· ${entry.partnerName}（${entry.rows.length} 条，最近截止 ${earliestDue}）`,
+    );
+  }
+
+  let digestSent = false;
+  if (config.robotWebhook && digestTasks.length > 0) {
+    const digest =
+      `今日待发布催办（${today}，截止 ≤ ${cutoff}）：\n${digestTasks.join("\n")}`;
+    try {
+      await sendWecomRobotMessage(digest, config, fetchImpl);
+      digestSent = true;
+    } catch (error) {
+      console.warn(
+        "[KOC LOOP] wecom robot digest failed",
+        { error: safeError(error) },
+      );
+    }
+  }
+
+  return {
+    dueSoonAttempted,
+    dueSoonSent,
+    dueSoonFailed,
+    dueSoonSkipped,
+    digestSent,
+  };
+}
+
+function bindingValue(value: unknown) {
+  return String(value ?? "").trim();
+}
+
+function truncate(value: string, max: number) {
+  return value.length > max ? `${value.slice(0, max)}…` : value;
+}
+
+function safeError(error: unknown) {
+  return error instanceof Error ? error.message : String(error);
+}
--- a/lib/workbook-image.ts
+++ b/lib/workbook-image.ts
@@ -0,0 +1,83 @@
+import sharp from "sharp";
+
+export type WorkbookSourceImage = {
+  bytes: Uint8Array;
+  contentType: string;
+  width?: number | null;
+  height?: number | null;
+  description: string;
+};
+
+export type WorkbookImageNormalizationOptions = {
+  maxDimension?: number;
+  outputFormat?: "png" | "jpeg";
+  jpegQuality?: number;
+};
+
+const NORMALIZABLE_IMAGE = /^image\/(?:jpe?g|png|webp|gif|tiff?|avif|heic|heif)$/i;
+
+function hasImageSignature(bytes: Uint8Array) {
+  if (bytes.length < 4) return false;
+  if (bytes[0] === 0xff && bytes[1] === 0xd8) return true;
+  if (bytes[0] === 0x89 && bytes[1] === 0x50 && bytes[2] === 0x4e && bytes[3] === 0x47) {
+    return true;
+  }
+  if (bytes[0] === 0x47 && bytes[1] === 0x49 && bytes[2] === 0x46) return true;
+  if (
+    bytes.length >= 12 &&
+    String.fromCharCode(...bytes.slice(0, 4)) === "RIFF" &&
+    String.fromCharCode(...bytes.slice(8, 12)) === "WEBP"
+  ) {
+    return true;
+  }
+  return (
+    (bytes[0] === 0x49 && bytes[1] === 0x49 && bytes[2] === 0x2a && bytes[3] === 0x00) ||
+    (bytes[0] === 0x4d && bytes[1] === 0x4d && bytes[2] === 0x00 && bytes[3] === 0x2a)
+  );
+}
+
+/**
+ * Excel viewers disagree on whether JPEG EXIF orientation should be applied.
+ * Bake that orientation into the pixels before the image enters the workbook.
+ * Callers may also resize and encode large source images as JPEG to keep the
+ * generated workbook within the upload limit.
+ */
+export async function normalizeWorkbookImage(
+  image: WorkbookSourceImage,
+  options: WorkbookImageNormalizationOptions = {},
+): Promise<WorkbookSourceImage> {
+  if (!NORMALIZABLE_IMAGE.test(image.contentType) && !hasImageSignature(image.bytes)) {
+    return image;
+  }
+  try {
+    let normalized = sharp(image.bytes, { animated: false }).rotate();
+    if (options.maxDimension && options.maxDimension > 0) {
+      normalized = normalized.resize({
+        width: Math.floor(options.maxDimension),
+        height: Math.floor(options.maxDimension),
+        fit: "inside",
+        withoutEnlargement: true,
+      });
+    }
+    normalized =
+      options.outputFormat === "jpeg"
+        ? normalized
+            .flatten({ background: "#ffffff" })
+            .jpeg({
+              quality: Math.min(95, Math.max(50, options.jpegQuality ?? 82)),
+              mozjpeg: true,
+            })
+        : normalized.png({ compressionLevel: 6 });
+    const { data, info } = await normalized.toBuffer({ resolveWithObject: true });
+    return {
+      ...image,
+      bytes: new Uint8Array(data),
+      contentType:
+        options.outputFormat === "jpeg" ? "image/jpeg" : "image/png",
+      width: info.width,
+      height: info.height,
+    };
+  } catch {
+    return image;
+  }
+}
--- a/mysql/0001_init.sql
+++ b/mysql/0001_init.sql
@@ -50,6 +50,9 @@ CREATE TABLE IF NOT EXISTS accounts (
  profile_url TEXT NOT NULL DEFAULT (''),
  ip_location VARCHAR(255) NOT NULL DEFAULT '待识别',
  followers INT NOT NULL DEFAULT 0,
+  gender VARCHAR(16) NOT NULL DEFAULT '',
+  bio TEXT NOT NULL DEFAULT (''),
+  tags VARCHAR(500) NOT NULL DEFAULT '',
  post_count INT NOT NULL DEFAULT 0,
  avg_views INT NOT NULL DEFAULT 0,
  first_seen_at DATETIME(3) NOT NULL DEFAULT CURRENT_TIMESTAMP(3),
--- a/mysql/0005_platform_video_douyin.sql
+++ b/mysql/0005_platform_video_douyin.sql
@@ -0,0 +1,14 @@
+ALTER TABLE tasks
+  ADD COLUMN platform VARCHAR(32) NOT NULL DEFAULT '小红书' AFTER task_type;
+-- statement-breakpoint
+ALTER TABLE tasks
+  ADD COLUMN content_format VARCHAR(32) NOT NULL DEFAULT 'image_text' AFTER platform;
+-- statement-breakpoint
+ALTER TABLE contents
+  ADD COLUMN video_assets LONGTEXT NOT NULL DEFAULT ('[]') AFTER image_assets;
+-- statement-breakpoint
+ALTER TABLE distributions
+  ADD COLUMN latest_shares INT NULL AFTER latest_collects;
+-- statement-breakpoint
+ALTER TABLE collection_runs
+  ADD COLUMN shares INT NULL AFTER collects;
--- a/mysql/0006_account_profile_tags.sql
+++ b/mysql/0006_account_profile_tags.sql
@@ -0,0 +1,8 @@
+ALTER TABLE accounts
+  ADD COLUMN gender VARCHAR(16) NOT NULL DEFAULT '' AFTER followers;
+-- statement-breakpoint
+ALTER TABLE accounts
+  ADD COLUMN bio TEXT NOT NULL DEFAULT ('') AFTER gender;
+-- statement-breakpoint
+ALTER TABLE accounts
+  ADD COLUMN tags VARCHAR(500) NOT NULL DEFAULT '' AFTER bio;
--- a/mysql/0007_account_current_contact.sql
+++ b/mysql/0007_account_current_contact.sql
@@ -0,0 +1,2 @@
+ALTER TABLE accounts
+  ADD COLUMN current_contact VARCHAR(255) NOT NULL DEFAULT '' AFTER cooperation_source;
--- a/mysql/0008_partner_wecom_external.sql
+++ b/mysql/0008_partner_wecom_external.sql
@@ -0,0 +1,2 @@
+ALTER TABLE partners
+  ADD COLUMN wecom_external_user_id VARCHAR(128) AFTER completed_total;
--- a/mysql/0009_wecom_group_push.sql
+++ b/mysql/0009_wecom_group_push.sql
@@ -0,0 +1,23 @@
+CREATE TABLE IF NOT EXISTS `wecom_group_chats` (
+  `chat_id` VARCHAR(128) NOT NULL,
+  `name` VARCHAR(512) NOT NULL DEFAULT '',
+  `owner_user_id` VARCHAR(128) NOT NULL,
+  `member_count` INT NOT NULL DEFAULT 0,
+  `status` INT NOT NULL DEFAULT 0,
+  `synced_at` DATETIME(3) NOT NULL DEFAULT CURRENT_TIMESTAMP(3),
+  PRIMARY KEY (`chat_id`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci;
+-- statement-breakpoint
+CREATE TABLE IF NOT EXISTS `wecom_group_pushes` (
+  `id` VARCHAR(64) NOT NULL,
+  `task_id` VARCHAR(64) NOT NULL,
+  `msgid` VARCHAR(128) NOT NULL DEFAULT '',
+  `sender` VARCHAR(128) NOT NULL,
+  `chat_ids` TEXT NOT NULL,
+  `text_content` TEXT NOT NULL,
+  `link_url` TEXT NOT NULL,
+  `fail_list` TEXT NOT NULL,
+  `created_at` DATETIME(3) NOT NULL DEFAULT CURRENT_TIMESTAMP(3),
+  PRIMARY KEY (`id`),
+  KEY `wecom_group_pushes_task_idx` (`task_id`)
+) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_ai_ci;
--- a/next.config.ts
+++ b/next.config.ts
@@ -2,7 +2,8 @@ import type { NextConfig } from "next";

 const nextConfig: NextConfig = {
  output: "standalone",
-  serverExternalPackages: ["mysql2"],
+  serverExternalPackages: ["mysql2", "sharp"],
+  allowedDevOrigins: ["192.168.30.90", "localhost", "127.0.0.1"],
 };

 export default nextConfig;
--- a/package-lock.json
+++ b/package-lock.json
@@ -16,6 +16,7 @@
        "node-cron": "^4.2.1",
        "react": "19.2.6",
        "react-dom": "19.2.6",
+        "sharp": "^0.35.3",
        "zod": "^4.4.3"
      },
      "devDependencies": {
@@ -1420,7 +1421,6 @@
      "resolved": "https://registry.npmjs.org/@img/colour/-/colour-1.1.0.tgz",
      "integrity": "sha512-Td76q7j57o/tLVdgS746cYARfSyxk8iEfRxewL9h4OMzYhbW4TAcppl0mT4eyqXddh6L/jwoM75mo7ixa/pCeQ==",
      "license": "MIT",
-      "optional": true,
      "engines": {
        "node": ">=18"
      }
@@ -3836,7 +3836,6 @@
      "version": "2.1.2",
      "resolved": "https://registry.npmjs.org/detect-libc/-/detect-libc-2.1.2.tgz",
      "integrity": "sha512-Btj2BOOO83o3WyH59e8MgXsxEQVcarkUOpEYrubB0urwnN10yQ364rsiByU11nZlqWYZm05i/of7io4mzihBtQ==",
-      "devOptional": true,
      "license": "Apache-2.0",
      "engines": {
        "node": ">=8"
@@ -7492,7 +7491,6 @@
      "resolved": "https://registry.npmjs.org/sharp/-/sharp-0.35.3.tgz",
      "integrity": "sha512-ej0zVHuZGHCiABXcNxeYhpRnPNPAcvbG8RMdBAhDAxLKkCRVSpK3Iyu7qbqw3JMzoj0REeM6f3tJLtVwl0023Q==",
      "license": "Apache-2.0",
-      "optional": true,
      "dependencies": {
        "@img/colour": "^1.1.0",
        "detect-libc": "^2.1.2",
@@ -7542,7 +7540,6 @@
      "resolved": "https://registry.npmjs.org/semver/-/semver-7.8.5.tgz",
      "integrity": "sha512-Y7/KDsb8LjooZpwaqGyulO6DQlksgCncchHGk+sZIY4SBvUocMBEFH5Ur1fI4dV+Jvl0w6cjvucaIi40puRioA==",
      "license": "ISC",
-      "optional": true,
      "bin": {
        "semver": "bin/semver.js"
      },
--- a/package.json
+++ b/package.json
@@ -7,7 +7,7 @@
  },
  "scripts": {
    "dev": "next dev",
-    "build": "next build",
+    "build": "next build --webpack",
    "start": "next start",
    "test": "npm run build && node --import tsx --test tests/*.test.mjs",
    "lint": "eslint . --ignore-pattern dist --ignore-pattern .next --ignore-pattern koc-portal/out",
@@ -26,6 +26,7 @@
    "node-cron": "^4.2.1",
    "react": "19.2.6",
    "react-dom": "19.2.6",
+    "sharp": "^0.35.3",
    "zod": "^4.4.3"
  },
  "devDependencies": {
--- a/public/KOC资源导入模板.xlsx
+++ b/public/KOC资源导入模板.xlsx
--- a/scripts/setup.sh
+++ b/scripts/setup.sh
@@ -0,0 +1,7 @@
+#!/bin/sh
+set -eu
+
+APP_DIR=$(CDPATH= cd -- "$(dirname -- "$0")/.." && pwd)
+cd "$APP_DIR"
+npm ci
+npm run build
--- a/scripts/start.sh
+++ b/scripts/start.sh
@@ -0,0 +1,8 @@
+#!/bin/sh
+set -eu
+
+APP_DIR=$(CDPATH= cd -- "$(dirname -- "$0")/.." && pwd)
+cd "$APP_DIR"
+export HOSTNAME=0.0.0.0
+export PORT="${PORT:-9000}"
+exec node .next/standalone/server.js
--- a/tests/feishu-client.test.mjs
+++ b/tests/feishu-client.test.mjs
@@ -120,6 +120,55 @@ test("resolves a wiki sheet and imports title, body, tags, and all images", asyn
  assert.equal(valuesCall.url.searchParams.get("ranges"), "sheet-one!A1:J20");
 });

+test("imports video attachments from a Feishu video task sheet", async () => {
+  clearFeishuAccessTokenCacheForTests();
+  const { fetchImpl } = fakeFeishu();
+  const videoFetch = async (input, init = {}) => {
+    const url = new URL(String(input));
+    if (url.pathname.endsWith("/values_batch_get")) {
+      return apiResponse({
+        valueRanges: [
+          {
+            values: [
+              ["标题", "内容（标题+正文+tag）", "视频"],
+              [
+                "一条测试视频",
+                "一条测试视频\n视频正文 #测试",
+                {
+                  type: "attachment",
+                  fileToken: "video-token-one",
+                  text: "demo.mp4",
+                  mimeType: "video/mp4",
+                  size: 1024,
+                },
+              ],
+            ],
+          },
+        ],
+      });
+    }
+    return fetchImpl(input, init);
+  };
+  const source = await readFeishuSource(
+    "https://tenant.feishu.cn/wiki/wiki-test",
+    bindings,
+    videoFetch,
+  );
+
+  assert.equal(source.rows.length, 1);
+  assert.equal(source.rows[0].title, "一条测试视频");
+  assert.equal(source.rows[0].body, "一条测试视频\n视频正文 #测试");
+  assert.deepEqual(source.rows[0].videos, [
+    {
+      index: 1,
+      fileToken: "video-token-one",
+      name: "demo.mp4",
+      mimeType: "video/mp4",
+      size: 1024,
+    },
+  ]);
+});
+
 test("requires an exact sheet link when a workbook has multiple visible sheets", async () => {
  clearFeishuAccessTokenCacheForTests();
  const { fetchImpl } = fakeFeishu({
--- a/tests/mcp-collection-client.test.mjs
+++ b/tests/mcp-collection-client.test.mjs
@@ -1,7 +1,9 @@
 import assert from "node:assert/strict";
 import test from "node:test";
 import {
+  collectMetricsFromMcp,
  collectXhsMetricsFromMcp,
+  resolveAccountProfileFromMcp,
  resolveCollectionMcpConfig,
  resolveXhsPublicAccountDetails,
  resolveXhsPublicAccountId,
@@ -40,10 +42,17 @@ function toolEnvelope(payload, isError = false) {
  };
 }

-function createFakeMcp(toolResults) {
+function createFakeMcp(toolResults, redirects = {}) {
  let toolIndex = 0;
  const calls = [];
  const fetchImpl = async (url, init) => {
+    if (!init?.body) {
+      const requestUrl = String(url);
+      calls.push({ url: requestUrl, body: null, headers: new Headers(init?.headers) });
+      const location = redirects[requestUrl];
+      if (!location) throw new Error(`Unexpected public request: ${requestUrl}`);
+      return new Response("", { status: 302, headers: { location } });
+    }
    const body = JSON.parse(init.body);
    calls.push({ url: String(url), body, headers: new Headers(init.headers) });
    if (body.method === "initialize") {
@@ -103,6 +112,7 @@ test("collects likes, comments and favorites from the verified MCP shape", async
    likes: 483,
    comments: 41,
    collects: 519,
+    shares: 0,
  });
  assert.equal(calls.length, 3);
  assert.equal(calls[2].body.params.name, "fetch_content_detail");
@@ -147,7 +157,7 @@ test("collects through a stateless MCP server without a session header", async (
    fetchImpl,
  );

-  assert.deepEqual(result, { likes: 12, comments: 3, collects: 6 });
+  assert.deepEqual(result, { likes: 12, comments: 3, collects: 6, shares: 0 });
  assert.deepEqual(
    calls.map((call) => call.body.method),
    ["initialize", "tools/call"],
@@ -217,19 +227,24 @@ test("resolves the real XHS account profile from a submitted note link", async (
    redId: "94329495984",
    ipLocation: "重庆",
    followers: 734,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
  });
-  assert.equal(calls.length, 4);
+  const mcpCalls = calls.filter((call) => call.body?.method);
+  assert.equal(mcpCalls.length, 4);
  assert.equal(
-    calls[2].body.params.name,
-    "collect_xhs_wen_note_detail",
+    mcpCalls[2].body.params.name,
+    "fetch_content_detail",
  );
  assert.equal(
-    calls[2].body.params.arguments.request.note_id,
-    "6a671108000000000f004bef",
+    mcpCalls[2].body.params.arguments.request.link,
+    "https://www.xiaohongshu.com/discovery/item/6a671108000000000f004bef",
  );
-  assert.equal(calls[3].body.params.name, "parse_xhs_user_summary");
+  assert.equal(mcpCalls[3].body.params.name, "parse_xhs_user_summary");
  assert.equal(
-    calls[3].body.params.arguments.request.url,
+    mcpCalls[3].body.params.arguments.request.url,
    "https://www.xiaohongshu.com/user/profile/6905cbca0000000037009f49",
  );
  assert.equal(
@@ -253,7 +268,17 @@ test("resolves followers directly from the supported XHS user summary tool", asy
            ipLocation: "福建",
            nickname: "555 五",
            userId: "1020668113",
+            gender: "女",
+            desc: "分享城市周末与美食",
+            tags: ["本地生活"],
          },
+          notes: [
+            {
+              note_id: "note-1",
+              title: "长沙湘菜探店",
+              url: "https://www.xiaohongshu.com/explore/note-1",
+            },
+          ],
        },
      },
    }),
@@ -273,6 +298,10 @@ test("resolves followers directly from the supported XHS user summary tool", asy
    followers: 6,
    redId: "1020668113",
    ipLocation: "福建",
+    gender: "女",
+    bio: "分享城市周末与美食",
+    recentNoteTitles: ["长沙湘菜探店"],
+    providerTags: ["本地生活"],
  });
  assert.equal(calls[2].body.params.name, "parse_xhs_user_summary");
 });
@@ -338,8 +367,8 @@ test("resolves an xhslink short URL before requesting the author profile", async
    "https://www.xiaohongshu.com/user/profile/6905cbca0000000037009f49",
  );
  assert.equal(
-    fakeMcp.calls[2].body.params.arguments.request.note_id,
-    "6a572da40000000021018bd2",
+    fakeMcp.calls[2].body.params.arguments.request.link,
+    "http://xhslink.cn/o/AJFyP5dnj7O",
  );
 });

@@ -400,10 +429,14 @@ test("uses the public note author when MCP profile lookup fails", async () => {
    redId: "1020668113",
    ipLocation: "待识别",
    followers: 6,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
  });
  assert.equal(
    fakeMcp.calls[2].body.params.name,
-    "collect_xhs_wen_note_detail",
+    "fetch_content_detail",
  );
 });

@@ -431,27 +464,16 @@ test("reads the user-visible Xiaohongshu number from a public profile", async ()
  });
 });

-test("falls back to parse_xhs_note when the primary tool fails", async () => {
+test("treats empty interaction counters from the current detail tool as zero", async () => {
  const { calls, fetchImpl } = createFakeMcp([
-    toolEnvelope(
-      {
-        response: {
-          code: 400,
-          success: false,
-          msg: "获取内容详情失败",
-          data: null,
-        },
-      },
-      true,
-    ),
    toolEnvelope({
      response: {
        code: 200,
        success: true,
        data: {
-          likes: "1.2万",
-          comments: 32,
-          collects: "2,345",
+          likes: "2",
+          comments: "",
+          collects: "",
        },
      },
    }),
@@ -466,11 +488,42 @@ test("falls back to parse_xhs_note when the primary tool fails", async () => {
  );

  assert.deepEqual(result, {
-    likes: 12_000,
-    comments: 32,
-    collects: 2_345,
+    likes: 2,
+    comments: 0,
+    collects: 0,
+    shares: 0,
  });
-  assert.equal(calls[3].body.params.name, "parse_xhs_note");
+  assert.equal(calls.length, 3);
+  assert.equal(calls[2].body.params.name, "fetch_content_detail");
+});
+
+test("surfaces current detail tool failures without calling removed tools", async () => {
+  const { calls, fetchImpl } = createFakeMcp([
+    toolEnvelope(
+      {
+        response: {
+          code: 400,
+          success: false,
+          msg: "获取内容详情失败",
+          data: null,
+        },
+      },
+      true,
+    ),
+  ]);
+
+  await assert.rejects(
+    collectXhsMetricsFromMcp(
+      "https://www.xiaohongshu.com/explore/test",
+      {
+        endpoint: "https://collector.example/mcp?key=test-key",
+      },
+      fetchImpl,
+    ),
+    /获取内容详情失败/,
+  );
+  assert.equal(calls.length, 3);
+  assert.equal(calls[2].body.params.name, "fetch_content_detail");
 });

 test("requires the MCP key without sending a network request", async () => {
@@ -551,6 +604,237 @@ test("rebuilds the MCP session after a gateway session miss", async () => {
    fetchImpl,
  );

-  assert.deepEqual(result, { likes: 8, comments: 2, collects: 5 });
+  assert.deepEqual(result, { likes: 8, comments: 2, collects: 5, shares: 0 });
  assert.equal(initializeCount, 2);
 });
+
+test("collects Douyin likes, favorites, shares and comments", async () => {
+  const { calls, fetchImpl } = createFakeMcp([
+    toolEnvelope({
+      response: {
+        code: 200,
+        success: true,
+        data: {
+          digg_count: "120",
+          collect_count: "30",
+          share_count: "8",
+          comment_count: "12",
+        },
+      },
+    }),
+  ]);
+
+  const result = await collectMetricsFromMcp(
+    "https://www.douyin.com/video/7520000000000000000",
+    "抖音",
+    { endpoint: "https://collector.example/mcp", key: "test-key" },
+    fetchImpl,
+  );
+
+  assert.deepEqual(result, {
+    likes: 120,
+    collects: 30,
+    shares: 8,
+    comments: 12,
+  });
+  assert.equal(calls[2].body.params.name, "fetch_content_detail");
+  assert.equal(calls[2].body.params.arguments.request.plant, "dy");
+});
+
+test("maps the current Douyin MCP metric field names", async () => {
+  const { fetchImpl } = createFakeMcp([
+    toolEnvelope({
+      response: {
+        code: 200,
+        success: true,
+        data: {
+          liked_count: "5682",
+          collected_count: "565",
+          share_count: "6878",
+          comment_count: "332",
+        },
+      },
+    }),
+  ]);
+
+  const result = await collectMetricsFromMcp(
+    "https://v.douyin.com/5O5VpgomO2U/",
+    "抖音",
+    { endpoint: "https://collector.example/mcp", key: "test-key" },
+    fetchImpl,
+  );
+
+  assert.deepEqual(result, {
+    likes: 5682,
+    collects: 565,
+    shares: 6878,
+    comments: 332,
+  });
+});
+
+test("resolves a Douyin account from a submitted work link", async () => {
+  const secUid = "MS4wLjABAAAA-test-profile-123456";
+  const { calls, fetchImpl } = createFakeMcp([
+    toolEnvelope({
+      response: {
+        code: 200,
+        success: true,
+        data: {
+          author: {
+            nickname: "抖音作者",
+            sec_uid: secUid,
+            unique_id: "douyin-123",
+          },
+        },
+      },
+    }),
+    toolEnvelope({
+      response: {
+        code: 200,
+        success: true,
+        data: {
+          user: {
+            nickname: "抖音作者",
+            unique_id: "douyin-123",
+            sec_uid: secUid,
+            follower_count: "1.5万",
+            ip_location: "上海",
+          },
+        },
+      },
+    }),
+  ]);
+
+  const profile = await resolveAccountProfileFromMcp(
+    "https://www.douyin.com/video/7520000000000000000",
+    "抖音",
+    "回填昵称",
+    { endpoint: "https://collector.example/mcp", key: "test-key" },
+    fetchImpl,
+  );
+
+  assert.deepEqual(profile, {
+    platformUid: secUid,
+    nickname: "抖音作者",
+    profileUrl: `https://www.douyin.com/user/${secUid}`,
+    redId: "douyin-123",
+    ipLocation: "上海",
+    followers: 15_000,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
+  });
+  assert.equal(calls[2].body.params.arguments.request.plant, "dy");
+  assert.equal(calls[3].body.params.name, "parse_dy_user_summary");
+});
+
+test("does not treat a Douyin short-link device id as the author sec_uid", async () => {
+  const shortLink = "https://v.douyin.com/5O5VpgomO2U/";
+  const { calls, fetchImpl } = createFakeMcp(
+    [
+      toolEnvelope({
+        response: {
+          code: 200,
+          success: true,
+          data: {
+            nickname: "搞怪噜噜😜",
+            user_id: "4065311277723529",
+          },
+        },
+      }),
+    ],
+    {
+      [shortLink]: "https://www.iesdouyin.com/share/video/7671270545631842038/?did=MS4wLjABAAAA-device-token&with_sec_did=1",
+    },
+  );
+
+  const profile = await resolveAccountProfileFromMcp(
+    shortLink,
+    "抖音",
+    "回填昵称",
+    { endpoint: "https://collector.example/mcp", key: "test-key" },
+    fetchImpl,
+  );
+
+  assert.deepEqual(profile, {
+    platformUid: "4065311277723529",
+    nickname: "搞怪噜噜😜",
+    profileUrl: "",
+    redId: "",
+    ipLocation: "待识别",
+    followers: null,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
+  });
+  assert.equal(calls.find((call) => call.body === null)?.url, shortLink);
+  assert.equal(
+    calls.find((call) => call.body?.params?.name === "parse_dy_user_summary"),
+    undefined,
+  );
+});
+
+test("resolves a Douyin profile only when the public redirect exposes sec_uid", async () => {
+  const shortLink = "https://v.douyin.com/author-sec-uid/";
+  const secUid = "MS4wLjABAAAAOqL4Jdu8htr7EWCDAyIr5z_7uvCAxhj-GOzCWg5zn8bDiKOp3WPw7lWkTyHvZpMY";
+  const { calls, fetchImpl } = createFakeMcp(
+    [
+      toolEnvelope({
+        response: {
+          code: 200,
+          success: true,
+          data: {
+            nickname: "抖音作者",
+            user_id: "4065311277723529",
+          },
+        },
+      }),
+      toolEnvelope({
+        response: {
+          code: 200,
+          success: true,
+          data: {
+            user: {
+              nickname: "抖音作者",
+              unique_id: "douyin-987",
+              sec_uid: secUid,
+              follower_count: "2.3万",
+              ip_location: "广东",
+            },
+          },
+        },
+      }),
+    ],
+    {
+      [shortLink]: `https://www.iesdouyin.com/share/video/7671270545631842038/?sec_uid=${secUid}`,
+    },
+  );
+
+  const profile = await resolveAccountProfileFromMcp(
+    shortLink,
+    "抖音",
+    "回填昵称",
+    { endpoint: "https://collector.example/mcp", key: "test-key" },
+    fetchImpl,
+  );
+
+  assert.deepEqual(profile, {
+    platformUid: secUid,
+    nickname: "抖音作者",
+    profileUrl: `https://www.douyin.com/user/${secUid}`,
+    redId: "douyin-987",
+    ipLocation: "广东",
+    followers: 23_000,
+    gender: "",
+    bio: "",
+    recentNoteTitles: [],
+    providerTags: [],
+  });
+  assert.equal(
+    calls.find((call) => call.body?.params?.name === "parse_dy_user_summary")
+      ?.body.params.arguments.request.url,
+    `https://www.douyin.com/user/${secUid}`,
+  );
+});
--- a/tests/partner-batch-upload.test.mjs
+++ b/tests/partner-batch-upload.test.mjs
@@ -0,0 +1,57 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { strToU8, unzipSync, zipSync } from "fflate";
+import { compactPartnerBatchWorkbookForUpload } from "../koc-portal/app/batch-workbook-upload.ts";
+
+const imageBytes = (marker, size) => {
+  const bytes = new Uint8Array(size);
+  bytes.set([0x89, 0x50, 0x4e, 0x47, marker]);
+  for (let index = 5; index < bytes.length; index += 1) bytes[index] = marker;
+  return bytes;
+};
+
+test("slims oversized WPS workbooks without removing backfill screenshots", () => {
+  const worksheet = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+      <sheetData>
+        <row r="1">
+          <c r="A1" t="inlineStr"><is><t>序号（不能改）</t></is></c>
+          <c r="D1" t="inlineStr"><is><t>图片1</t></is></c>
+          <c r="F1" t="inlineStr"><is><t>笔记截图</t></is></c>
+          <c r="G1" t="inlineStr"><is><t>数据分析截图（单篇笔记数据分析截图）</t></is></c>
+        </row>
+        <row r="2">
+          <c r="D2" t="str"><f>_xlfn.DISPIMG(&quot;SOURCE&quot;,1)</f><v>=DISPIMG(&quot;SOURCE&quot;,1)</v></c>
+          <c r="F2" t="str"><f>_xlfn.DISPIMG(&quot;PUBLISH&quot;,1)</f><v>=DISPIMG(&quot;PUBLISH&quot;,1)</v></c>
+        </row>
+      </sheetData>
+    </worksheet>`;
+  const cellImages = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <etc:cellImages xmlns:etc="http://www.wps.cn/officeDocument/2017/etCustomData" xmlns:xdr="http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing" xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships">
+      <etc:cellImage><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="1" name="SOURCE"/></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId1"/></xdr:blipFill></xdr:pic></etc:cellImage>
+      <etc:cellImage><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="2" name="PUBLISH"/></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId2"/></xdr:blipFill></xdr:pic></etc:cellImage>
+    </etc:cellImages>`;
+  const relationships = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">
+      <Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image" Target="media/source.png"/>
+      <Relationship Id="rId2" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image" Target="media/publish.png"/>
+    </Relationships>`;
+  const workbook = zipSync(
+    {
+      "xl/worksheets/sheet1.xml": strToU8(worksheet),
+      "xl/cellimages.xml": strToU8(cellImages),
+      "xl/_rels/cellimages.xml.rels": strToU8(relationships),
+      "xl/media/source.png": [imageBytes(1, 2_000_000), { level: 0 }],
+      "xl/media/publish.png": [imageBytes(2, 2_000), { level: 0 }],
+    },
+    { level: 0 },
+  );
+
+  const compacted = compactPartnerBatchWorkbookForUpload(workbook);
+  const entries = unzipSync(compacted.bytes);
+  assert.equal(entries["xl/media/source.png"], undefined);
+  assert.deepEqual(entries["xl/media/publish.png"], imageBytes(2, 2_000));
+  assert.equal(compacted.removedMediaCount, 1);
+  assert.equal(compacted.preservedScreenshotCount, 1);
+  assert.ok(compacted.bytes.byteLength < workbook.byteLength / 10);
+});
--- a/tests/partner-batch-workbook.test.mjs
+++ b/tests/partner-batch-workbook.test.mjs
@@ -0,0 +1,327 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { strFromU8, strToU8, unzipSync, zipSync } from "fflate";
+import { buildRecoveryWorkbook } from "../lib/recovery-workbook.ts";
+import {
+  PARTNER_BATCH_HEADERS,
+  buildPartnerBatchWorkbookColumns,
+  parsePartnerBatchWorkbook,
+  resolvePartnerWorkbookOrigin,
+} from "../lib/partner-batch-workbook.ts";
+import { hasMp4FileSignature } from "../lib/video-file.ts";
+
+const png = Uint8Array.from([
+  0x89, 0x50, 0x4e, 0x47, 0x0d, 0x0a, 0x1a, 0x0a,
+  0x00, 0x00, 0x00, 0x0d, 0x49, 0x48, 0x44, 0x52,
+  0x00, 0x00, 0x00, 0x01, 0x00, 0x00, 0x00, 0x01,
+]);
+
+test("round-trips hidden assignment IDs and embedded backfill screenshots", () => {
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "批量回填",
+    headers: [...PARTNER_BATCH_HEADERS],
+    columnWidths: [10, 20, 40, 30, 40, 24, 28, 20, 20, 20],
+    hiddenColumns: [7, 8, 9],
+    rows: [
+      {
+        cells: [
+          1,
+          "测试笔记",
+          "正文 #话题",
+          "",
+          "https://www.xiaohongshu.com/explore/1234567890abcdef",
+          "",
+          "",
+          "distribution-1",
+          "",
+          "",
+        ],
+        images: [
+          {
+            column: 3,
+            image: { bytes: png, contentType: "image/png", description: "原图" },
+          },
+          {
+            column: 5,
+            image: { bytes: png, contentType: "image/png", description: "笔记截图" },
+          },
+          {
+            column: 6,
+            image: { bytes: png, contentType: "image/png", description: "数据截图" },
+          },
+        ],
+      },
+    ],
+  });
+  const rows = parsePartnerBatchWorkbook(workbook);
+  assert.equal(rows.length, 1);
+  assert.equal(rows[0].distributionId, "distribution-1");
+  assert.equal(rows[0].title, "测试笔记");
+  assert.match(rows[0].publishUrl, /xiaohongshu\.com/);
+  assert.equal(rows[0].publishScreenshot?.contentType, "image/png");
+  assert.equal(rows[0].creatorScreenshot?.contentType, "image/png");
+
+  const entries = unzipSync(workbook);
+  const worksheet = strFromU8(entries["xl/worksheets/sheet1.xml"]);
+  assert.match(worksheet, /序号（不能改）/);
+  assert.doesNotMatch(worksheet, /张原图（见图）|已回填（见图）|请插入笔记截图/);
+  assert.match(worksheet, /<drawing r:id="rId1"\/>/);
+  assert.doesNotMatch(worksheet, /#VALUE!/);
+  const drawing = strFromU8(entries["xl/drawings/drawing1.xml"]);
+  assert.equal((drawing.match(/<xdr:twoCellAnchor editAs="twoCell">/g) ?? []).length, 3);
+  assert.match(drawing, /<xdr:col>3<\/xdr:col>/);
+  assert.match(drawing, /<xdr:col>5<\/xdr:col>/);
+  assert.match(drawing, /<xdr:col>6<\/xdr:col>/);
+  assert.match(worksheet, /min="8" max="8"[^>]*hidden="1"/);
+});
+
+test("accepts the legacy sequence header for previously exported workbooks", () => {
+  const legacyHeaders = [...PARTNER_BATCH_HEADERS];
+  legacyHeaders[0] = "序号";
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "批量回填",
+    headers: legacyHeaders,
+    columnWidths: [10, 20, 40, 30, 40, 24, 28, 20, 20, 20],
+    hiddenColumns: [7, 8, 9],
+    rows: [
+      {
+        cells: [
+          1,
+          "旧模板笔记",
+          "正文",
+          "",
+          "",
+          "",
+          "",
+          "distribution-legacy",
+          "",
+          "",
+        ],
+        images: [],
+      },
+    ],
+  });
+  const rows = parsePartnerBatchWorkbook(workbook);
+  assert.equal(rows[0].sequence, "1");
+  assert.equal(rows[0].distributionId, "distribution-legacy");
+});
+
+test("imports dynamic source image columns without confusing screenshot columns", () => {
+  const headers = [
+    "序号（不能改）",
+    "标题",
+    "笔记内容（正文+话题）",
+    "图片1",
+    "图片2",
+    "发布链接",
+    "笔记截图",
+    "数据分析截图（单篇笔记数据分析截图）",
+    "_系统笔记ID",
+    "_原笔记截图",
+    "_原数据分析截图",
+  ];
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "批量回填",
+    headers,
+    columnWidths: headers.map(() => 20),
+    hiddenColumns: [8, 9, 10],
+    rows: [
+      {
+        cells: [
+          1,
+          "多图笔记",
+          "正文",
+          "",
+          "",
+          "https://www.xiaohongshu.com/explore/dynamic",
+          "",
+          "",
+          "distribution-dynamic",
+          "",
+          "",
+        ],
+        images: [
+          {
+            column: 3,
+            image: { bytes: png, contentType: "image/png", description: "原图1" },
+          },
+          {
+            column: 4,
+            image: { bytes: png, contentType: "image/png", description: "原图2" },
+          },
+          {
+            column: 6,
+            image: { bytes: png, contentType: "image/png", description: "笔记截图" },
+          },
+          {
+            column: 7,
+            image: { bytes: png, contentType: "image/png", description: "数据截图" },
+          },
+        ],
+      },
+    ],
+  });
+  const rows = parsePartnerBatchWorkbook(workbook);
+  assert.equal(rows[0].distributionId, "distribution-dynamic");
+  assert.equal(rows[0].publishScreenshot?.contentType, "image/png");
+  assert.equal(rows[0].creatorScreenshot?.contentType, "image/png");
+});
+
+test("imports screenshots saved by WPS as DISPIMG cell images", () => {
+  const sourceImage = Uint8Array.from([...png, 1]);
+  const publishScreenshot = Uint8Array.from([...png, 2]);
+  const creatorScreenshot = Uint8Array.from([...png, 3]);
+  const worksheet = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <worksheet xmlns="http://schemas.openxmlformats.org/spreadsheetml/2006/main">
+      <sheetData>
+        <row r="1">
+          <c r="A1" t="inlineStr"><is><t>序号（不能改）</t></is></c>
+          <c r="B1" t="inlineStr"><is><t>标题</t></is></c>
+          <c r="C1" t="inlineStr"><is><t>笔记内容（正文+话题）</t></is></c>
+          <c r="D1" t="inlineStr"><is><t>图片1</t></is></c>
+          <c r="E1" t="inlineStr"><is><t>发布链接</t></is></c>
+          <c r="F1" t="inlineStr"><is><t>笔记截图</t></is></c>
+          <c r="G1" t="inlineStr"><is><t>数据分析截图（单篇笔记数据分析截图）</t></is></c>
+          <c r="H1" t="inlineStr"><is><t>_系统笔记ID</t></is></c>
+          <c r="I1" t="inlineStr"><is><t>_原笔记截图</t></is></c>
+          <c r="J1" t="inlineStr"><is><t>_原数据分析截图</t></is></c>
+        </row>
+        <row r="2">
+          <c r="A2"><v>1</v></c>
+          <c r="B2" t="inlineStr"><is><t>WPS 笔记</t></is></c>
+          <c r="D2" t="str"><f>_xlfn.DISPIMG(&quot;SOURCE&quot;,1)</f><v>=DISPIMG(&quot;SOURCE&quot;,1)</v></c>
+          <c r="E2" t="inlineStr"><is><t>https://www.xiaohongshu.com/explore/wps</t></is></c>
+          <c r="F2" t="str"><f>_xlfn.DISPIMG(&quot;PUBLISH&quot;,1)</f><v>=DISPIMG(&quot;PUBLISH&quot;,1)</v></c>
+          <c r="G2" t="str"><f>_xlfn.DISPIMG(&quot;CREATOR&quot;,1)</f><v>=DISPIMG(&quot;CREATOR&quot;,1)</v></c>
+          <c r="H2" t="inlineStr"><is><t>distribution-wps</t></is></c>
+        </row>
+      </sheetData>
+    </worksheet>`;
+  const cellImages = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <etc:cellImages xmlns:etc="http://www.wps.cn/officeDocument/2017/etCustomData" xmlns:xdr="http://schemas.openxmlformats.org/drawingml/2006/spreadsheetDrawing" xmlns:a="http://schemas.openxmlformats.org/drawingml/2006/main" xmlns:r="http://schemas.openxmlformats.org/officeDocument/2006/relationships">
+      <etc:cellImage><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="1" name="SOURCE"/></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId1"/></xdr:blipFill></xdr:pic></etc:cellImage>
+      <etc:cellImage><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="2" name="PUBLISH"/></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId2"/></xdr:blipFill></xdr:pic></etc:cellImage>
+      <etc:cellImage><xdr:pic><xdr:nvPicPr><xdr:cNvPr id="3" name="CREATOR"/></xdr:nvPicPr><xdr:blipFill><a:blip r:embed="rId3"/></xdr:blipFill></xdr:pic></etc:cellImage>
+    </etc:cellImages>`;
+  const relationships = `<?xml version="1.0" encoding="UTF-8" standalone="yes"?>
+    <Relationships xmlns="http://schemas.openxmlformats.org/package/2006/relationships">
+      <Relationship Id="rId1" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image" Target="media/source.png"/>
+      <Relationship Id="rId2" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image" Target="media/publish.png"/>
+      <Relationship Id="rId3" Type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image" Target="media/creator.png"/>
+    </Relationships>`;
+  const workbook = zipSync({
+    "xl/worksheets/sheet1.xml": strToU8(worksheet),
+    "xl/cellimages.xml": strToU8(cellImages),
+    "xl/_rels/cellimages.xml.rels": strToU8(relationships),
+    "xl/media/source.png": sourceImage,
+    "xl/media/publish.png": publishScreenshot,
+    "xl/media/creator.png": creatorScreenshot,
+  });
+
+  const rows = parsePartnerBatchWorkbook(workbook);
+  assert.equal(rows.length, 1);
+  assert.equal(rows[0].distributionId, "distribution-wps");
+  assert.deepEqual(rows[0].publishScreenshot?.bytes, publishScreenshot);
+  assert.deepEqual(rows[0].creatorScreenshot?.bytes, creatorScreenshot);
+});
+
+test("builds video-task workbooks with video columns and no source image columns", () => {
+  const columns = buildPartnerBatchWorkbookColumns({
+    contentFormat: "video",
+    maxSourceImages: 3,
+    maxSourceVideos: 2,
+  });
+  assert.deepEqual(columns.headers.slice(0, 8), [
+    "序号（不能改）",
+    "标题",
+    "笔记内容（正文+话题）",
+    "视频1",
+    "视频2",
+    "发布链接",
+    "笔记截图",
+    "数据分析截图（单篇笔记数据分析截图）",
+  ]);
+  assert.equal(columns.headers.some((header) => /^图片\d+$/.test(header)), false);
+
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "视频批量回填",
+    headers: columns.headers,
+    columnWidths: columns.columnWidths,
+    hiddenColumns: [
+      columns.systemColumn,
+      columns.systemColumn + 1,
+      columns.systemColumn + 2,
+    ],
+    rows: [
+      {
+        cells: [
+          1,
+          "视频笔记",
+          "视频正文 #测试",
+          "下载视频1",
+          "下载视频2",
+          "",
+          "",
+          "",
+          "distribution-video",
+          "",
+          "",
+        ],
+        images: [],
+        hyperlinks: [
+          {
+            column: columns.sourceVideoStartColumn,
+            url: "https://koc.example.com/api/partner-image?kind=video&download=1",
+          },
+        ],
+      },
+    ],
+  });
+  const entries = unzipSync(workbook);
+  const worksheet = strFromU8(entries["xl/worksheets/sheet1.xml"]);
+  const relationships = strFromU8(
+    entries["xl/worksheets/_rels/sheet1.xml.rels"],
+  );
+  assert.match(worksheet, /视频1/);
+  assert.doesNotMatch(worksheet, /图片1/);
+  assert.match(relationships, /https:\/\/koc\.example\.com\/api\/partner-image/);
+  assert.match(relationships, /download=1/);
+});
+
+test("uses the configured public origin before proxy or container addresses", () => {
+  const request = new Request("http://app:3000/api/partner-batch-workbook", {
+    headers: {
+      host: "app:3000",
+      "x-forwarded-host": "internal-proxy:8080",
+      "x-forwarded-proto": "https",
+    },
+  });
+  assert.equal(
+    resolvePartnerWorkbookOrigin(request, "https://koc.example.com/koc/"),
+    "https://koc.example.com",
+  );
+});
+
+test("preserves a forwarded non-standard port when no origin is configured", () => {
+  const request = new Request("http://app:3000/api/partner-batch-workbook", {
+    headers: {
+      host: "app:3000",
+      "x-forwarded-host": "localhost:8080",
+      "x-forwarded-proto": "http",
+    },
+  });
+  assert.equal(resolvePartnerWorkbookOrigin(request), "http://localhost:8080");
+});
+
+test("recognizes MP4 bytes instead of trusting a response content type", () => {
+  const mp4Header = Uint8Array.from([
+    0x00, 0x00, 0x00, 0x18,
+    0x66, 0x74, 0x79, 0x70,
+    0x69, 0x73, 0x6f, 0x6d,
+    0x00, 0x00, 0x02, 0x00,
+    0x69, 0x73, 0x6f, 0x6d,
+    0x6d, 0x70, 0x34, 0x32,
+  ]);
+  assert.equal(hasMp4FileSignature(mp4Header), true);
+  assert.equal(hasMp4FileSignature(new TextEncoder().encode("not a video")), false);
+});
--- a/tests/recovery-workbook.test.mjs
+++ b/tests/recovery-workbook.test.mjs
@@ -16,7 +16,7 @@ test("creates an xlsx archive whose pictures are embedded in worksheet cells", (
    columnWidths: [8, 24, 24, 24],
    rows: [
      {
-        cells: [1, "测试笔记", "见图", "见图"],
+        cells: [1, "测试笔记", "", ""],
        images: [
          {
            column: 2,
@@ -43,12 +43,24 @@ test("creates an xlsx archive whose pictures are embedded in worksheet cells", (
    ],
  });
  const archive = unzipSync(workbook);
+  const worksheet = strFromU8(archive["xl/worksheets/sheet1.xml"]);

  assert.ok(archive["xl/media/image1.png"]);
  assert.ok(archive["xl/media/image2.png"]);
-  assert.match(strFromU8(archive["xl/worksheets/sheet1.xml"]), /<drawing r:id="rId1"\/>/);
-  assert.match(strFromU8(archive["xl/drawings/drawing1.xml"]), /oneCellAnchor/);
-  assert.match(strFromU8(archive["xl/drawings/_rels/drawing1.xml.rels"]), /image2\.png/);
+  assert.match(worksheet, /<drawing r:id="rId1"\/>/);
+  assert.doesNotMatch(worksheet, /#VALUE!/);
+  const drawing = strFromU8(archive["xl/drawings/drawing1.xml"]);
+  assert.equal((drawing.match(/<xdr:twoCellAnchor editAs="twoCell">/g) ?? []).length, 2);
+  assert.match(drawing, /<xdr:col>2<\/xdr:col>/);
+  assert.match(drawing, /<xdr:col>3<\/xdr:col>/);
+  assert.match(
+    strFromU8(archive["xl/drawings/_rels/drawing1.xml.rels"]),
+    /image2\.png/,
+  );
+  assert.doesNotMatch(
+    worksheet,
+    /见图/,
+  );
 });

 test("creates clickable external hyperlinks for resource exports", () => {
--- a/tests/rendered-html.test.mjs
+++ b/tests/rendered-html.test.mjs
@@ -23,6 +23,30 @@ test("builds the KOC LOOP product shell", async () => {
  await access(new URL("../.next/static", import.meta.url));
 });

+test("keeps distribution filters as always-visible fuzzy search fields", async () => {
+  const [adminApp, globalCss] = await Promise.all([
+    readFile(new URL("../app/admin-app.tsx", import.meta.url), "utf8"),
+    readFile(new URL("../app/globals.css", import.meta.url), "utf8"),
+  ]);
+
+  assert.match(globalCss, /\.distribution-task-search\s*\{[^}]*flex-direction:\s*row/s);
+  assert.match(globalCss, /\.distribution-task-search\s*\{[^}]*margin:\s*0/s);
+  assert.match(globalCss, /\.distribution-task-search\s*>\s*span\s*\{[^}]*flex:\s*0\s+0\s+18px/s);
+  assert.match(globalCss, /\.distribution-task-filter-combobox\.brand\s*\{[^}]*flex-basis:\s*176px/s);
+  assert.match(globalCss, /\.distribution-task-filter-menu\s*\{[^}]*position:\s*absolute[^}]*z-index:\s*60/s);
+  assert.match(globalCss, /\.distribution-task-filter-input\s*\{[^}]*display:\s*flex[^}]*margin:\s*0/s);
+  assert.match(adminApp, /aria-expanded=\{openTaskFilter === filter\.key\}/);
+  assert.match(adminApp, /role="combobox"/);
+  assert.match(adminApp, /role="listbox"/);
+  assert.match(adminApp, /placeholder=\{`搜索\$\{filter\.label\}`\}/);
+  assert.match(adminApp, /setTaskFilterValue\(filter\.key, event\.target\.value\)/);
+  assert.doesNotMatch(adminApp, /distribution-task-filter-trigger/);
+  assert.doesNotMatch(adminApp, /distribution-task-filter-menu-search/);
+  assert.doesNotMatch(adminApp, /taskFilterQuery/);
+  assert.doesNotMatch(adminApp, /distribution-task-option-panel/);
+  assert.doesNotMatch(adminApp, /<select value=\{contentTypeFilter\}/);
+});
+
 test("stacks user management and securely removes departed accounts", async () => {
  const [usersPage, usersRoute, globalCss] = await Promise.all([
    readFile(new URL("../app/users-page.tsx", import.meta.url), "utf8"),
@@ -137,8 +161,8 @@ test("issues external task links and supports one-to-one note submissions", asyn
  assert.match(partnerRoute, /claimantIdentifier\.canonical/);
  assert.match(partnerRoute, /legacyPartnerId/);
  assert.match(partnerRoute, /微信号或手机号/);
-  assert.match(partnerRoute, /extractXhsPublishUrl/);
-  assert.match(partnerRoute, /小红书长链或短链/);
+  assert.match(partnerRoute, /extractPublishUrl/);
+  assert.match(partnerRoute, /task\.platform/);
  assert.match(partnerRoute, /请填写发布链接/);
  assert.doesNotMatch(partnerRoute, /请填写发布账号昵称和发布链接/);
  assert.match(partnerRoute, /没有找到领取记录/);
@@ -147,11 +171,16 @@ test("issues external task links and supports one-to-one note submissions", asyn
  assert.match(partnerRoute, /enrichDistributionAccount/);
  assert.match(partnerRoute, /runInBackground\(enrichment/);
  assert.match(accountEnrichment, /profile_url = excluded\.profile_url/);
-  assert.match(accountEnrichment, /resolveXhsAccountProfileFromMcp/);
-  assert.match(accountEnrichment, /resolveXhsProfileDetailsFromMcp/);
+  assert.match(accountEnrichment, /resolveAccountProfileFromMcp/);
+  assert.match(accountEnrichment, /resolveProfileDetailsFromMcp/);
  assert.match(accountEnrichment, /isVerifiedXhsProfileUrl/);
  assert.match(accountEnrichment, /endsWith\("\.xiaohongshu\.com"\)/);
  assert.match(accountEnrichment, /DELETE FROM accounts/);
+  assert.match(accountEnrichment, /WHERE platform = \? AND platform_uid = \?/);
+  assert.match(accountEnrichment, /existingAccount\?\.id \|\| canonicalAccountId/);
+  assert.match(accountEnrichment, /cl\.claimant_name AS claimant_contact/);
+  assert.match(accountEnrichment, /current_contact = CASE/);
+  assert.match(accountEnrichment, /!row\.resolved_account_id/);
  assert.match(accountEnrichment, /a\.followers/);
  assert.match(accountEnrichment, /followers = CASE/);
  assert.match(accountEnrichment, /Number\(row\.followers \?\? 0\) === 0/);
@@ -171,6 +200,11 @@ test("issues external task links and supports one-to-one note submissions", asyn
  assert.match(imageRoute, /creator-center\//);
  assert.match(imageRoute, /imageKind === "publish"/);
  assert.match(imageRoute, /imageKind === "creator"/);
+  assert.match(imageRoute, /isMutableEvidence/);
+  assert.match(imageRoute, /"private, no-store"/);
+  assert.match(imageRoute, /Content-Type", "video\/mp4"/);
+  assert.match(imageRoute, /video-\$\{imageIndex\}\.mp4/);
+  assert.match(imageRoute, /downloadRequested \? "attachment" : "inline"/);
  assert.match(imageRoute, /cl\.claim_token/);
  assert.match(imageUploadRoute, /isAdminRequest/);
  assert.match(cors, /KOC_PORTAL_URL/);
@@ -178,6 +212,17 @@ test("issues external task links and supports one-to-one note submissions", asyn
  assert.match(cors, /X-KOC-Upload-Kind/);
  assert.match(cors, /Access-Control-Allow-Origin/);
  assert.match(adminApp, /hasCreatorMetrics/);
+  assert.match(adminApp, /function PlatformBadge/);
+  assert.match(adminApp, /function resourceProfileLink/);
+  assert.match(adminApp, /搜索主页/);
+  assert.match(adminApp, /latest_publish_url/);
+  assert.match(adminApp, /通过作品查看主页/);
+  assert.match(adminApp, /platform-logo/);
+  assert.match(adminApp, /按任务名称模糊搜索/);
+  assert.match(adminApp, /全部品牌\/项目/);
+  assert.match(adminApp, /全部内容类型/);
+  assert.match(adminApp, /全部平台/);
+  assert.match(adminApp, /task-scope-subline/);
  assert.match(adminApp, /待KOC填写数据/);
  assert.match(adminApp, /AdminImageLightbox/);
  assert.match(adminApp, /CreatorScreenshotPreview/);
@@ -195,6 +240,43 @@ test("issues external task links and supports one-to-one note submissions", asyn
  assert.match(migration, /claim_token/);
 });

+test("exports and imports claim-bound Excel backfill workbooks", async () => {
+  const [route, parser, workbook, imageNormalizer, nginx] = await Promise.all([
+    readFile(new URL("../app/api/partner-batch-workbook/route.ts", import.meta.url), "utf8"),
+    readFile(new URL("../lib/partner-batch-workbook.ts", import.meta.url), "utf8"),
+    readFile(new URL("../lib/recovery-workbook.ts", import.meta.url), "utf8"),
+    readFile(new URL("../lib/workbook-image.ts", import.meta.url), "utf8"),
+    readFile(new URL("../deploy/nginx/koc-loop.conf", import.meta.url), "utf8"),
+  ]);
+
+  assert.match(route, /批量回填/);
+  assert.match(route, /publish-evidence/);
+  assert.match(route, /creator-center/);
+  assert.match(route, /extractPublishUrl/);
+  assert.match(route, /findAccess/);
+  assert.match(route, /isDifferentFromStoredImage/);
+  assert.match(route, /rowIndex \+ 1/);
+  assert.match(route, /buildPartnerBatchWorkbookColumns/);
+  assert.match(route, /resolvePartnerWorkbookOrigin/);
+  assert.match(route, /download: "1"/);
+  assert.match(route, /columns\.sourceImageStartColumn \+ index/);
+  assert.match(parser, /_系统笔记ID/);
+  assert.match(parser, /笔记截图/);
+  assert.match(parser, /数据分析截图（单篇笔记数据分析截图）/);
+  assert.match(parser, /parseImages/);
+  assert.match(workbook, /hiddenColumns/);
+  assert.match(workbook, /offsetX/);
+  assert.doesNotMatch(workbook, /value \|\| "见图"/);
+  assert.match(imageNormalizer, /\.rotate\(\)/);
+  assert.match(imageNormalizer, /\.png\(/);
+  assert.match(nginx, /client_max_body_size 85m/);
+  assert.equal((nginx.match(/proxy_set_header Host \$http_host;/g) ?? []).length, 2);
+  assert.equal(
+    (nginx.match(/proxy_set_header X-Forwarded-Host \$http_host;/g) ?? []).length,
+    2,
+  );
+});
+
 test("supports task collection schedules and latest public metrics", async () => {
  const [
    adminApp,
@@ -206,6 +288,7 @@ test("supports task collection schedules and latest public metrics", async () =>
    migration,
    accountMigration,
    compose,
+    mcpClient,
  ] = await Promise.all([
    readFile(new URL("../app/admin-app.tsx", import.meta.url), "utf8"),
    readFile(new URL("../app/api/action/route.ts", import.meta.url), "utf8"),
@@ -216,6 +299,7 @@ test("supports task collection schedules and latest public metrics", async () =>
    readFile(new URL("../mysql/0001_init.sql", import.meta.url), "utf8"),
    readFile(new URL("../mysql/0001_init.sql", import.meta.url), "utf8"),
    readFile(new URL("../docker-compose.self-hosted.yml", import.meta.url), "utf8"),
+    readFile(new URL("../lib/mcp-collection-client.ts", import.meta.url), "utf8"),
  ]);

  for (const label of [
@@ -238,12 +322,19 @@ test("supports task collection schedules and latest public metrics", async () =>
  assert.match(adminApp, /sortWithNullsLast/);
  assert.match(adminApp, /内容 \/ 发布账号/);
  assert.match(adminApp, /recovery-title-link/);
-  assert.match(adminApp, /打开小红书笔记/);
+  assert.match(adminApp, /打开\$\{selectedTask\.platform\}作品/);
  assert.match(adminApp, /target="_blank"/);
  assert.match(adminApp, /noopener noreferrer/);
-  assert.match(adminApp, /const noteUrl = xhsPublishUrl\(item\.publish_url\)/);
+  assert.match(adminApp, /const noteUrl = publicPublishUrl\(item\.publish_url\)/);
  assert.match(adminApp, /noteUrl \? \(/);
+  assert.match(adminApp, /updateDistributionPublishUrl/);
+  assert.match(adminApp, /填写链接/);
+  assert.match(adminApp, /更新链接/);
  assert.match(adminApp, /hostname === "xhslink\.cn"/);
+  assert.match(actionRoute, /update_distribution_publish_url/);
+  assert.match(actionRoute, /extractPublishUrl/);
+  assert.match(actionRoute, /DELETE FROM collection_runs WHERE distribution_id = \?/);
+  assert.match(actionRoute, /enrichDistributionAccount/);
  assert.match(actionRoute, /save_collection_schedule/);
  assert.match(actionRoute, /collect_now/);
  assert.match(actionRoute, /backfill_account_profiles/);
@@ -253,7 +344,11 @@ test("supports task collection schedules and latest public metrics", async () =>
  assert.match(actionRoute, /createCollectionRunTasks/);
  assert.match(bootstrapRoute, /runDueScheduledCollections/);
  assert.match(collectionService, /INSERT OR IGNORE INTO collection_runs/);
-  assert.match(collectionService, /collectXhsMetricsFromMcp/);
+  assert.match(
+    collectionService,
+    /run\.status === "success" && source !== "manual"/,
+  );
+  assert.match(collectionService, /collectMetricsFromMcp/);
  assert.doesNotMatch(collectionService, /hashText/);
  assert.match(collectionService, /runScheduledCollections/);
  assert.match(collectionService, /runDueScheduledCollections/);
@@ -272,6 +367,10 @@ test("supports task collection schedules and latest public metrics", async () =>
  assert.match(migration, /collection_runs_distribution_date_idx/);
  assert.match(accountMigration, /public_account_id/);
  assert.match(compose, /ENABLE_SCHEDULER/);
+  assert.match(mcpClient, /"fetch_content_detail"/);
+  assert.match(mcpClient, /"parse_xhs_user_summary"/);
+  assert.doesNotMatch(mcpClient, /"parse_xhs_note"/);
+  assert.doesNotMatch(mcpClient, /"collect_xhs_wen_note_detail"/);
 });

 test("supports fixed screenshot collection tasks without publish metrics", async () => {
@@ -314,7 +413,7 @@ test("exports complete task recovery data to Excel with embedded images", async
  assert.match(adminApp, /导出全部数据/);
  assert.match(adminApp, /\/api\/recovery-export/);
  assert.match(adminApp, /图片和截图已嵌入表格/);
-  assert.match(exportRoute, /小红书昵称/);
+  assert.match(exportRoute, /`\$\{task\.platform\}昵称`/);
  assert.match(exportRoute, /曝光量-实际（第7天）/);
  assert.match(exportRoute, /阅读量-实际（第7天）/);
  assert.match(exportRoute, /publish_screenshot_key/);
@@ -323,8 +422,9 @@ test("exports complete task recovery data to Excel with embedded images", async
  assert.match(exportRoute, /isAdminRequest/);
  assert.match(exportRoute, /consumeMcpExportToken/);
  assert.match(workbook, /xl\/drawings\/drawing1\.xml/);
+  assert.match(workbook, /twoCellAnchor editAs="twoCell"/);
  assert.match(workbook, /xl\/media\/image/);
-  assert.match(workbook, /oneCellAnchor/);
+  assert.match(workbook, /relationships\/image/);
 });

 test("provides simple username-password login and three server-enforced roles", async () => {
@@ -393,6 +493,7 @@ test("filters and exports the current KOC resource result set", async () => {
  ]);

  assert.match(adminApp, /搜索账号名称 \/ 账号ID/);
+  assert.match(adminApp, /当前联系人/);
  assert.match(adminApp, /搜索IP地区/);
  assert.match(adminApp, /搜索合作来源/);
  assert.match(adminApp, /ipLocation\.includes\(ipKeyword\)/);
@@ -401,6 +502,7 @@ test("filters and exports the current KOC resource result set", async () => {
  assert.match(adminApp, /filteredAccounts\.map\(\(item\) => item\.account\.id\)/);
  assert.match(exportRoute, /小红书号\/抖音号/);
  assert.match(exportRoute, /历史合作来源/);
+  assert.match(exportRoute, /当前联系人/);
  assert.match(exportRoute, /合作社资源 · 不可直联/);
  assert.match(exportRoute, /isManagerRequest/);
  assert.match(exportRoute, /consumeMcpExportToken/);
@@ -408,21 +510,61 @@ test("filters and exports the current KOC resource result set", async () => {
 });

 test("imports existing KOC resources through a validated spreadsheet preview", async () => {
-  const [adminApp, importRoute, resourceParser, accountMigration] = await Promise.all([
+  const [adminApp, globalCss, importRoute, resourceParser, resourceWrite, accountMigration, profileMigration, contactMigration] = await Promise.all([
    readFile(new URL("../app/admin-app.tsx", import.meta.url), "utf8"),
+    readFile(new URL("../app/globals.css", import.meta.url), "utf8"),
    readFile(new URL("../app/api/resources-import/route.ts", import.meta.url), "utf8"),
    readFile(new URL("../lib/resource-import.ts", import.meta.url), "utf8"),
+    readFile(new URL("../lib/resource-write.ts", import.meta.url), "utf8"),
    readFile(new URL("../mysql/0002_resource_import.sql", import.meta.url), "utf8"),
+    readFile(new URL("../mysql/0006_account_profile_tags.sql", import.meta.url), "utf8"),
+    readFile(new URL("../mysql/0007_account_current_contact.sql", import.meta.url), "utf8"),
  ]);
  assert.match(adminApp, /下载导入模板/);
  assert.match(adminApp, /校验并预览/);
  assert.match(adminApp, /确认导入/);
  assert.match(importRoute, /isManagerRequest/);
  assert.match(importRoute, /mode !== "commit"/);
-  assert.match(resourceParser, /RESOURCE_IMPORT_MAX_ROWS = 100/);
-  assert.match(resourceParser, /当前自动解析仅支持小红书账号主页/);
-  assert.match(importRoute, /resolveXhsProfileDetailsFromMcp/);
+  assert.match(resourceParser, /RESOURCE_IMPORT_MAX_ROWS = 10_000/);
+  assert.match(resourceParser, /RESOURCE_IMPORT_MAX_BYTES = 20 \* 1024 \* 1024/);
+  assert.match(adminApp, /单次最多 10,000 个账号，文件不超过 20MB/);
+  assert.match(resourceWrite, /RESOURCE_IMPORT_DB_BATCH_SIZE = 100/);
+  assert.match(importRoute, /bulk resource profile enrichment/);
+  assert.match(adminApp, /异常数据将自动跳过，不会导入/);
+  assert.match(
+    adminApp,
+    /summary\.create \+ importPreview\.summary\.update === 0 \|\| importWorking/,
+  );
+  assert.doesNotMatch(
+    adminApp,
+    /disabled=\{importPreview\.summary\.error > 0 \|\| importWorking\}/,
+  );
+  assert.match(importRoute, /const importableRows = analyzed\.filter/);
+  assert.match(importRoute, /跳过 \$\{summary\.error\} 条异常数据/);
+  assert.match(importRoute, /previewAnalyzedRows\(analyzed\)/);
+  assert.match(resourceParser, /当前自动解析仅支持小红书或抖音账号主页/);
+  assert.match(resourceWrite, /resolveProfileDetailsFromMcp/);
  assert.match(accountMigration, /cooperation_source/);
+  assert.match(profileMigration, /ADD COLUMN gender/);
+  assert.match(profileMigration, /ADD COLUMN bio/);
+  assert.match(profileMigration, /ADD COLUMN tags/);
+  assert.match(contactMigration, /ADD COLUMN current_contact/);
+  assert.match(adminApp, /gender-icon male/);
+  assert.match(adminApp, /gender-icon female/);
+  assert.match(adminApp, /resource-profile-avatar/);
+  assert.match(adminApp, /resource-account-number/);
+  assert.match(adminApp, /resource-platform-line/);
+  assert.match(adminApp, /resource-latest/);
+  assert.match(adminApp, /合作来源/);
+  assert.match(adminApp, /待打标/);
+  assert.doesNotMatch(adminApp, /className="verified-dot"/);
+  assert.match(globalCss, /\.resource-tags\s*\{[^}]*margin-bottom:\s*auto/s);
+  assert.match(globalCss, /\.resource-card-foot\s*\{[^}]*margin-top:\s*10px/s);
+  assert.match(adminApp, /resource-tags/);
+  assert.match(adminApp, /最多 5 个/);
+  assert.match(resourceParser, /gender: \["性别"\]/);
+  assert.match(resourceParser, /bio: \["简介"/);
+  assert.match(resourceParser, /tags: \["标签"/);
 });

 test("supports anonymous partner delegation without creating a second data flow", async () => {
--- a/tests/resource-import.test.mjs
+++ b/tests/resource-import.test.mjs
@@ -1,7 +1,9 @@
 import assert from "node:assert/strict";
 import test from "node:test";
+import { strFromU8, strToU8, unzipSync, zipSync } from "fflate";
 import { buildRecoveryWorkbook } from "../lib/recovery-workbook.ts";
 import {
+  RESOURCE_IMPORT_MAX_ROWS,
  mergeCooperationSources,
  normalizeProfileUrl,
  parseResourceFollowers,
@@ -10,6 +12,72 @@ import {
  resourcePlatformUid,
 } from "../lib/resource-import.ts";

+test("accepts several thousand accounts in one import file", () => {
+  const csv = [
+    "账号主页,账号昵称,账号ID,IP属地,粉丝数,合作来源",
+    ...Array.from({ length: 3_000 }, (_, index) => {
+      const id = String(index + 1).padStart(6, "0");
+      return `https://www.xiaohongshu.com/user/profile/bulk${id},账号${id},${id},上海,100,批量资源`;
+    }),
+  ].join("\n");
+  const rows = parseResourceImportFile(
+    "bulk-resources.csv",
+    new TextEncoder().encode(csv),
+  );
+
+  assert.equal(RESOURCE_IMPORT_MAX_ROWS, 10_000);
+  assert.equal(rows.length, 3_000);
+  assert.equal(rows[0].rowNumber, 2);
+  assert.equal(rows.at(-1)?.rowNumber, 3_001);
+  assert.equal(rows.every((row) => row.errors.length === 0), true);
+});
+
+test("parses several thousand accounts from an XLSX workbook", () => {
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "KOC资源导入",
+    headers: ["账号主页", "账号昵称", "账号ID", "IP属地", "粉丝数", "合作来源"],
+    columnWidths: [48, 24, 20, 16, 14, 24],
+    rows: Array.from({ length: 3_000 }, (_, index) => {
+      const id = String(index + 1).padStart(6, "0");
+      return {
+        cells: [
+          `https://www.xiaohongshu.com/user/profile/xlsx${id}`,
+          `账号${id}`,
+          id,
+          "北京",
+          200,
+          "Excel批量资源",
+        ],
+        images: [],
+      };
+    }),
+  });
+  const rows = parseResourceImportFile("bulk-resources.xlsx", workbook);
+
+  assert.equal(rows.length, 3_000);
+  assert.equal(rows[0].profileUrl.endsWith("xlsx000001"), true);
+  assert.equal(rows.at(-1)?.profileUrl.endsWith("xlsx003000"), true);
+});
+
+test("keeps a bounded 10,000-row safety limit", () => {
+  const csv = [
+    "账号主页",
+    ...Array.from(
+      { length: RESOURCE_IMPORT_MAX_ROWS + 1 },
+      (_, index) => `https://www.douyin.com/user/bulk-account-${index + 1}`,
+    ),
+  ].join("\n");
+
+  assert.throws(
+    () =>
+      parseResourceImportFile(
+        "too-many-resources.csv",
+        new TextEncoder().encode(csv),
+      ),
+    /单次最多导入 10000 个账号/,
+  );
+});
+
 test("parses CSV resources and normalizes public profile data", () => {
  const csv = [
    "账号主页,合作来源",
@@ -26,6 +94,9 @@ test("parses CSV resources and normalizes public profile data", () => {
    ipLocation: "",
    followers: 0,
    followersResolved: false,
+    gender: "",
+    bio: "",
+    tags: [],
    cooperationSource: "林林KOC社群",
    errors: [],
  });
@@ -34,8 +105,8 @@ test("parses CSV resources and normalizes public profile data", () => {

 test("uses optional account fields directly and only requires the profile URL", () => {
  const csv = [
-    "账号链接,账号昵称,账号ID,IP属地,粉丝数,合作来源",
-    "https://www.xiaohongshu.com/user/profile/abc123,番茄不炒蛋,4171542126,江西,10+,历史资源",
+    "账号链接,账号昵称,账号ID,IP属地,粉丝数,性别,简介,标签,合作来源",
+    "https://www.xiaohongshu.com/user/profile/abc123,番茄不炒蛋,4171542126,江西,10+,女,分享江西本地生活,本地生活、美食探店,历史资源",
  ].join("\n");
  const [row] = parseResourceImportFile(
    "resources.csv",
@@ -46,9 +117,36 @@ test("uses optional account fields directly and only requires the profile URL",
  assert.equal(row.ipLocation, "江西");
  assert.equal(row.followers, 10);
  assert.equal(row.followersResolved, true);
+  assert.equal(row.gender, "女");
+  assert.equal(row.bio, "分享江西本地生活");
+  assert.deepEqual(row.tags, ["本地生活", "美食探店"]);
  assert.deepEqual(resourceImportMissingFields(row), []);
 });

+test("validates optional gender", () => {
+  const csv = [
+    "账号链接,性别,标签",
+    "https://www.xiaohongshu.com/user/profile/abc123,其他,美食探店",
+  ].join("\n");
+  const [row] = parseResourceImportFile(
+    "resources.csv",
+    new TextEncoder().encode(csv),
+  );
+  assert.match(row.errors.join("；"), /性别格式不正确/);
+});
+
+test("rejects more than five tags in one optional tag cell", () => {
+  const csv = [
+    "账号链接,标签",
+    'https://www.xiaohongshu.com/user/profile/abc123,"美食探店,旅游出行,数码产品,本地生活,婚嫁备婚,美妆护肤"',
+  ].join("\n");
+  const [row] = parseResourceImportFile(
+    "resources.csv",
+    new TextEncoder().encode(csv),
+  );
+  assert.match(row.errors.join("；"), /最多填写 5 个标签/);
+});
+
 test("normalizes common follower formats and identifies missing enrichment fields", () => {
  assert.deepEqual(parseResourceFollowers("1.3万"), {
    value: 13_000,
@@ -86,11 +184,49 @@ test("parses the first matching worksheet from an XLSX workbook", () => {
  assert.equal(rows[0].followersResolved, false);
 });

+test("keeps values in their columns after a self-closing blank XLSX cell", () => {
+  const workbook = buildRecoveryWorkbook({
+    sheetName: "KOC资源导入",
+    headers: ["账号主页", "账号昵称", "账号ID", "IP属地", "粉丝数", "合作来源"],
+    columnWidths: [48, 24, 20, 16, 14, 24],
+    rows: [
+      {
+        cells: [
+          "https://www.xiaohongshu.com/user/profile/blank-ip-cell",
+          "空白IP账号",
+          "123456789",
+          "",
+          10,
+          "",
+        ],
+        images: [],
+      },
+    ],
+  });
+  const entries = unzipSync(workbook);
+  const sheetPath = "xl/worksheets/sheet1.xml";
+  const sheetXml = strFromU8(entries[sheetPath]);
+  entries[sheetPath] = strToU8(
+    sheetXml.replace('<c r="E2"', '<c r="D2"/><c r="E2"'),
+  );
+
+  const [row] = parseResourceImportFile(
+    "self-closing-blank.xlsx",
+    zipSync(entries),
+  );
+  assert.equal(row.ipLocation, "");
+  assert.equal(row.followers, 10);
+  assert.equal(row.followersResolved, true);
+  assert.deepEqual(row.errors, []);
+});
+
 test("reports invalid required fields without hiding valid rows", () => {
-  const csv = "账号主页,合作来源\n,历史资源\nhttps://www.douyin.com/user/demo,社群";
+  const csv = "账号主页,合作来源\n,历史资源\nhttps://www.douyin.com/user/demo,社群\nhttps://example.com/user/demo,其他";
  const rows = parseResourceImportFile("resources.csv", new TextEncoder().encode(csv));
  assert.match(rows[0].errors.join("；"), /账号主页不能为空/);
-  assert.match(rows[1].errors.join("；"), /仅支持小红书账号主页/);
+  assert.equal(rows[1].platform, "抖音");
+  assert.equal(rows[1].errors.length, 0);
+  assert.match(rows[2].errors.join("；"), /仅支持小红书或抖音账号主页/);
 });

 test("rejects invalid optional follower values without requiring other optional fields", () => {
@@ -105,6 +241,18 @@ test("rejects invalid optional follower values without requiring other optional
  assert.match(row.errors.join("；"), /粉丝数格式不正确/);
 });

+test("rejects a numeric value entered as an IP location", () => {
+  const csv = [
+    "账号链接,IP属地,粉丝数",
+    "https://www.xiaohongshu.com/user/profile/abc123,10,100",
+  ].join("\n");
+  const [row] = parseResourceImportFile(
+    "resources.csv",
+    new TextEncoder().encode(csv),
+  );
+  assert.match(row.errors.join("；"), /IP属地格式不正确/);
+});
+
 test("normalizes profile URLs and merges cooperation sources", () => {
  assert.equal(
    normalizeProfileUrl("https://www.xiaohongshu.com/user/profile/abc/?foo=1#top"),
--- a/tests/wecom-client.test.mjs
+++ b/tests/wecom-client.test.mjs
@@ -0,0 +1,607 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import {
+  clearWecomAccessTokenCacheForTests,
+  createGroupMsgTemplate,
+  listCustomerGroupChats,
+  listExternalContacts,
+  resolveWecomConfig,
+  sendWecomAppMessage,
+  sendWecomRobotMessage,
+  WecomClientError,
+} from "../lib/wecom-client.ts";
+
+const baseBindings = {
+  WECOM_CORP_ID: "corp-test",
+  WECOM_AGENT_ID: "1000001",
+  WECOM_SECRET: "secret-test",
+  WECOM_ROBOT_WEBHOOK:
+    "https://qyapi.weixin.qq.com/cgi-bin/webhook/send?key=robot-test",
+  WECOM_NOTIFY_DUE_DAYS: "3",
+};
+
+function fakeWecom(options = {}) {
+  const calls = [];
+  const fetchImpl = async (input, init = {}) => {
+    const url = new URL(String(input));
+    calls.push({ url, init });
+    if (url.pathname.endsWith("/cgi-bin/gettoken")) {
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        access_token: "token-test",
+        expires_in: 7200,
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/webhook/send")) {
+      return Response.json({
+        errcode: options.robotErrcode ?? 0,
+        errmsg: options.robotErrmsg ?? "ok",
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/message/send")) {
+      const body = JSON.parse(String(init.body ?? "{}"));
+      const invalid = options.invalidUser
+        ? body.touser.split("|").filter((u) => u === options.invalidUser)
+        : [];
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        invaliduser: invalid.join("|"),
+      });
+    }
+    return new Response("not found", { status: 404 });
+  };
+  return { calls, fetchImpl };
+}
+
+test.beforeEach(() => {
+  clearWecomAccessTokenCacheForTests();
+});
+
+test("resolveWecomConfig applies dueDays and trims values", () => {
+  const config = resolveWecomConfig({
+    WECOM_CORP_ID: "  corp  ",
+    WECOM_AGENT_ID: "  10  ",
+    WECOM_SECRET: "  s  ",
+    WECOM_ROBOT_WEBHOOK: "  https://hook  ",
+    WECOM_NOTIFY_DUE_DAYS: "5",
+  });
+  assert.equal(config.corpId, "corp");
+  assert.equal(config.agentId, "10");
+  assert.equal(config.secret, "s");
+  assert.equal(config.robotWebhook, "https://hook");
+  assert.equal(config.dueDays, 5);
+});
+
+test("resolveWecomConfig falls back to default dueDays for bad input", () => {
+  assert.equal(resolveWecomConfig({ WECOM_NOTIFY_DUE_DAYS: "0" }).dueDays, 3);
+  assert.equal(resolveWecomConfig({ WECOM_NOTIFY_DUE_DAYS: "abc" }).dueDays, 3);
+  assert.equal(
+    resolveWecomConfig({ WECOM_NOTIFY_DUE_DAYS: "100" }).dueDays,
+    30,
+  );
+});
+
+test("sendWecomRobotMessage posts text payload and returns void", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  await sendWecomRobotMessage(
+    "hello",
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  const hookCall = calls.find((c) =>
+    c.url.pathname.endsWith("/cgi-bin/webhook/send"),
+  );
+  assert.ok(hookCall, "robot webhook was called");
+  const body = JSON.parse(String(hookCall.init.body));
+  assert.equal(body.msgtype, "text");
+  assert.equal(body.text.content, "hello");
+});
+
+test("sendWecomRobotMessage skips silently when webhook is empty", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  const config = resolveWecomConfig({ ...baseBindings, WECOM_ROBOT_WEBHOOK: "" });
+  await sendWecomRobotMessage("hello", config, fetchImpl);
+  assert.equal(
+    calls.filter((c) =>
+      c.url.pathname.endsWith("/cgi-bin/webhook/send"),
+    ).length,
+    0,
+  );
+});
+
+test("sendWecomRobotMessage throws WecomClientError on provider error", async () => {
+  const { fetchImpl } = fakeWecom({
+    robotErrcode: 93000,
+    robotErrmsg: "invalid webhook url",
+  });
+  await assert.rejects(
+    () =>
+      sendWecomRobotMessage(
+        "hello",
+        resolveWecomConfig(baseBindings),
+        fetchImpl,
+      ),
+    (error) => {
+      assert.ok(error instanceof WecomClientError);
+      assert.equal(error.status, 502);
+      return true;
+    },
+  );
+});
+
+test("sendWecomAppMessage fetches access_token then sends to message/send", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  const result = await sendWecomAppMessage(
+    ["user-a", "user-b"],
+    "催办内容",
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.equal(result.sent, 2);
+  assert.equal(result.failed, 0);
+  assert.equal(result.skipped, false);
+  const tokenCall = calls.find((c) =>
+    c.url.pathname.endsWith("/cgi-bin/gettoken"),
+  );
+  assert.ok(tokenCall, "gettoken was called");
+  const sendCall = calls.find((c) =>
+    c.url.pathname.endsWith("/cgi-bin/message/send"),
+  );
+  assert.ok(sendCall, "message/send was called");
+  assert.equal(sendCall.url.searchParams.get("access_token"), "token-test");
+  const body = JSON.parse(String(sendCall.init.body));
+  assert.equal(body.touser, "user-a|user-b");
+  assert.equal(body.agentid, 1000001);
+  assert.equal(body.text.content, "催办内容");
+});
+
+test("sendWecomAppMessage skips when no external user ids", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  const result = await sendWecomAppMessage(
+    [],
+    "催办",
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.equal(result.skipped, true);
+  assert.equal(
+    calls.filter((c) =>
+      c.url.pathname.endsWith("/cgi-bin/message/send"),
+    ).length,
+    0,
+  );
+});
+
+test("sendWecomAppMessage skips when corp credentials are missing", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  const result = await sendWecomAppMessage(
+    ["user-a"],
+    "催办",
+    resolveWecomConfig({
+      ...baseBindings,
+      WECOM_CORP_ID: "",
+      WECOM_AGENT_ID: "",
+      WECOM_SECRET: "",
+    }),
+    fetchImpl,
+  );
+  assert.equal(result.skipped, true);
+  assert.equal(
+    calls.filter((c) =>
+      c.url.pathname.endsWith("/cgi-bin/gettoken"),
+    ).length,
+    0,
+  );
+});
+
+test("sendWecomAppMessage reports failed when provider marks invaliduser", async () => {
+  const { fetchImpl } = fakeWecom({ invalidUser: "user-a" });
+  const result = await sendWecomAppMessage(
+    ["user-a", "user-b"],
+    "催办",
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.equal(result.sent, 1);
+  assert.equal(result.failed, 1);
+});
+
+test("access_token is cached across calls", async () => {
+  const { calls, fetchImpl } = fakeWecom();
+  const config = resolveWecomConfig(baseBindings);
+  await sendWecomAppMessage(["u1"], "a", config, fetchImpl);
+  await sendWecomAppMessage(["u2"], "b", config, fetchImpl);
+  const tokenCalls = calls.filter((c) =>
+    c.url.pathname.endsWith("/cgi-bin/gettoken"),
+  );
+  assert.equal(tokenCalls.length, 1, "access_token cached for second call");
+});
+
+function fakeExternalContacts() {
+  const calls = [];
+  const fetchImpl = async (input, init = {}) => {
+    const url = new URL(String(input));
+    calls.push({ url, init });
+    if (url.pathname.endsWith("/cgi-bin/gettoken")) {
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        access_token: "token-test",
+        expires_in: 7200,
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/get_follow_user_list")) {
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        follow_user: [{ userid: "owner-a" }, { userid: "owner-b" }],
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/list")) {
+      const userid = url.searchParams.get("userid");
+      if (userid === "owner-a") {
+        return Response.json({
+          errcode: 0,
+          errmsg: "ok",
+          external_userid: ["ext-1", "ext-2"],
+        });
+      }
+      if (userid === "owner-b") {
+        return Response.json({
+          errcode: 0,
+          errmsg: "ok",
+          external_userid: ["ext-3"],
+        });
+      }
+      return Response.json({ errcode: 60020, errmsg: "user not found" });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/get")) {
+      const eid = url.searchParams.get("external_userid");
+      if (eid === "ext-1") {
+        return Response.json({
+          errcode: 0,
+          errmsg: "ok",
+          external_contact: {
+            external_userid: "ext-1",
+            name: "张三",
+            avatar: "https://example.com/a.png",
+            corp_fullname: "ACME 公司",
+          },
+        });
+      }
+      if (eid === "ext-2") {
+        return Response.json({
+          errcode: 0,
+          errmsg: "ok",
+          external_contact: {
+            external_userid: "ext-2",
+            name: "李四",
+            avatar: "",
+            corp_fullname: "",
+          },
+        });
+      }
+      if (eid === "ext-3") {
+        return Response.json({
+          errcode: 0,
+          errmsg: "ok",
+          external_contact: {
+            external_userid: "ext-3",
+            name: "王五",
+            avatar: "https://example.com/c.png",
+            corp_fullname: "Other 公司",
+          },
+        });
+      }
+      return Response.json({ errcode: 60111, errmsg: "not found" });
+    }
+    return new Response("not found", { status: 404 });
+  };
+  return { calls, fetchImpl };
+}
+
+test("listExternalContacts flattens follow_user → list → get into a single array", async () => {
+  const { fetchImpl } = fakeExternalContacts();
+  const contacts = await listExternalContacts(
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.equal(contacts.length, 3);
+  const first = contacts.find((c) => c.externalUserId === "ext-1");
+  assert.ok(first);
+  assert.equal(first.name, "张三");
+  assert.equal(first.avatar, "https://example.com/a.png");
+  assert.equal(first.corpName, "ACME 公司");
+  assert.equal(first.ownerUserId, "owner-a");
+  const second = contacts.find((c) => c.externalUserId === "ext-2");
+  assert.ok(second);
+  assert.equal(second.corpName, "李四");
+  const third = contacts.find((c) => c.externalUserId === "ext-3");
+  assert.ok(third);
+  assert.equal(third.ownerUserId, "owner-b");
+});
+
+test("listExternalContacts skips members with non-zero errcode", async () => {
+  const { fetchImpl } = fakeExternalContacts();
+  const contacts = await listExternalContacts(
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  const ownerBIds = contacts
+    .filter((c) => c.ownerUserId === "owner-b")
+    .map((c) => c.externalUserId);
+  assert.deepEqual(ownerBIds, ["ext-3"]);
+});
+
+test("listExternalContacts returns empty array when app credentials missing", async () => {
+  const { calls, fetchImpl } = fakeExternalContacts();
+  const contacts = await listExternalContacts(
+    resolveWecomConfig({
+      ...baseBindings,
+      WECOM_CORP_ID: "",
+      WECOM_AGENT_ID: "",
+      WECOM_SECRET: "",
+    }),
+    fetchImpl,
+  );
+  assert.deepEqual(contacts, []);
+  assert.equal(calls.length, 0, "no API calls without credentials");
+});
+
+function fakeGroupPush(options = {}) {
+  const calls = [];
+  const pages =
+    options.pages ??
+    [
+      {
+        next_cursor: "page-2",
+        group_chat_list: [
+          { chat_id: "wr-normal-1", status: 0 },
+          { chat_id: "wr-dissolved", status: 1 },
+        ],
+      },
+      { next_cursor: "", group_chat_list: [{ chat_id: "wr-normal-2", status: 0 }] },
+    ];
+  const details =
+    options.details ??
+    {
+      "wr-normal-1": { name: "KOC 一群", owner: "zhangsan", member_count: 42 },
+      "wr-normal-2": { name: "KOC 二群", owner: "lisi", member_count: 7 },
+    };
+  const fetchImpl = async (input, init = {}) => {
+    const url = new URL(String(input));
+    calls.push({ url, init });
+    if (url.pathname.endsWith("/cgi-bin/gettoken")) {
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        access_token: "token-test",
+        expires_in: 7200,
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/groupchat/list")) {
+      const body = JSON.parse(String(init.body ?? "{}"));
+      const index = body.cursor ? Number(body.cursor.replace("page-", "")) - 1 : 0;
+      const page = pages[index] ?? { next_cursor: "", group_chat_list: [] };
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        next_cursor: page.next_cursor,
+        group_chat_list: page.group_chat_list,
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/groupchat/get")) {
+      const body = JSON.parse(String(init.body ?? "{}"));
+      const detail = details[body.chat_id];
+      if (!detail) {
+        return Response.json({ errcode: 60111, errmsg: "not found" });
+      }
+      return Response.json({ errcode: 0, errmsg: "ok", group_chat: detail });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template")) {
+      return Response.json({
+        errcode: options.templateErrcode ?? 0,
+        errmsg: options.templateErrmsg ?? "ok",
+        msgid: options.msgid ?? "msgGTEST",
+        fail_list: options.failList ?? [],
+      });
+    }
+    return new Response("not found", { status: 404 });
+  };
+  return { calls, fetchImpl };
+}
+
+test("listCustomerGroupChats paginates groupchat/list and keeps only normal groups", async () => {
+  const { calls, fetchImpl } = fakeGroupPush();
+  const groups = await listCustomerGroupChats(
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.equal(groups.length, 2);
+  const first = groups.find((g) => g.chatId === "wr-normal-1");
+  assert.ok(first);
+  assert.equal(first.name, "KOC 一群");
+  assert.equal(first.ownerUserId, "zhangsan");
+  assert.equal(first.memberCount, 42);
+  const second = groups.find((g) => g.chatId === "wr-normal-2");
+  assert.ok(second);
+  assert.equal(second.ownerUserId, "lisi");
+  assert.equal(
+    groups.some((g) => g.chatId === "wr-dissolved"),
+    false,
+    "non-normal status groups are filtered",
+  );
+  const listCalls = calls.filter((c) =>
+    c.url.pathname.endsWith("/cgi-bin/externalcontact/groupchat/list"),
+  );
+  assert.equal(listCalls.length, 2, "cursor pagination followed");
+  const secondBody = JSON.parse(String(listCalls[1].init.body));
+  assert.equal(secondBody.cursor, "page-2");
+});
+
+test("listCustomerGroupChats skips groups whose detail lookup fails", async () => {
+  const { fetchImpl } = fakeGroupPush({
+    pages: [
+      {
+        next_cursor: "",
+        group_chat_list: [
+          { chat_id: "wr-normal-1", status: 0 },
+          { chat_id: "wr-gone", status: 0 },
+        ],
+      },
+    ],
+    details: {
+      "wr-normal-1": { name: "KOC 一群", owner: "zhangsan", member_count: 42 },
+    },
+  });
+  const groups = await listCustomerGroupChats(
+    resolveWecomConfig(baseBindings),
+    fetchImpl,
+  );
+  assert.deepEqual(
+    groups.map((g) => g.chatId),
+    ["wr-normal-1"],
+  );
+});
+
+test("createGroupMsgTemplate posts group payload with text and link attachment", async () => {
+  const { calls, fetchImpl } = fakeGroupPush();
+  const result = await createGroupMsgTemplate(
+    resolveWecomConfig(baseBindings),
+    {
+      sender: "zhangsan",
+      chatIdList: ["wr-normal-1", "wr-normal-2"],
+      text: "【新任务】测试任务",
+      link: {
+        title: "测试任务",
+        desc: "品牌 A · 3 份",
+        url: "https://portal.example.com/?task=tok-1",
+      },
+    },
+    fetchImpl,
+  );
+  assert.equal(result.msgid, "msgGTEST");
+  assert.deepEqual(result.failList, []);
+  const sendCall = calls.find((c) =>
+    c.url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template"),
+  );
+  assert.ok(sendCall, "add_msg_template was called");
+  assert.equal(sendCall.url.searchParams.get("access_token"), "token-test");
+  const body = JSON.parse(String(sendCall.init.body));
+  assert.equal(body.chat_type, "group");
+  assert.equal(body.sender, "zhangsan");
+  assert.deepEqual(body.chat_id_list, ["wr-normal-1", "wr-normal-2"]);
+  assert.equal(body.text.content, "【新任务】测试任务");
+  assert.equal(body.attachments.length, 1);
+  assert.equal(body.attachments[0].msgtype, "link");
+  assert.equal(body.attachments[0].link.url, "https://portal.example.com/?task=tok-1");
+});
+
+test("createGroupMsgTemplate omits text and attachments keys when absent", async () => {
+  const { calls, fetchImpl } = fakeGroupPush();
+  await createGroupMsgTemplate(
+    resolveWecomConfig(baseBindings),
+    {
+      sender: "zhangsan",
+      chatIdList: ["wr-normal-1"],
+      link: { title: "只有图文", url: "https://portal.example.com/?task=tok-1" },
+    },
+    fetchImpl,
+  );
+  const sendCall = calls.find((c) =>
+    c.url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template"),
+  );
+  assert.ok(sendCall, "add_msg_template was called");
+  const body = JSON.parse(String(sendCall.init.body));
+  assert.equal("text" in body, false);
+  assert.equal(body.attachments[0].link.title, "只有图文");
+  assert.equal(body.attachments[0].link.desc, "");
+});
+
+test("createGroupMsgTemplate passes through fail_list from provider", async () => {
+  const { fetchImpl } = fakeGroupPush({
+    msgid: "msgFAIL",
+    failList: ["wr-normal-2"],
+  });
+  const result = await createGroupMsgTemplate(
+    resolveWecomConfig(baseBindings),
+    {
+      sender: "zhangsan",
+      chatIdList: ["wr-normal-1", "wr-normal-2"],
+      text: "内容",
+    },
+    fetchImpl,
+  );
+  assert.equal(result.msgid, "msgFAIL");
+  assert.deepEqual(result.failList, ["wr-normal-2"]);
+});
+
+test("createGroupMsgTemplate throws WecomClientError on provider error", async () => {
+  const { fetchImpl } = fakeGroupPush({
+    templateErrcode: 81053,
+    templateErrmsg: "user not in visible scope",
+  });
+  await assert.rejects(
+    () =>
+      createGroupMsgTemplate(
+        resolveWecomConfig(baseBindings),
+        { sender: "zhangsan", chatIdList: ["wr-normal-1"], text: "内容" },
+        fetchImpl,
+      ),
+    (error) => {
+      assert.ok(error instanceof WecomClientError);
+      assert.equal(error.status, 502);
+      assert.match(error.message, /user not in visible scope/);
+      return true;
+    },
+  );
+});
+
+test("createGroupMsgTemplate rejects empty text and link before calling API", async () => {
+  const { calls, fetchImpl } = fakeGroupPush();
+  await assert.rejects(
+    () =>
+      createGroupMsgTemplate(
+        resolveWecomConfig(baseBindings),
+        { sender: "zhangsan", chatIdList: ["wr-normal-1"] },
+        fetchImpl,
+      ),
+    (error) => {
+      assert.ok(error instanceof WecomClientError);
+      assert.equal(error.status, 400);
+      return true;
+    },
+  );
+  assert.equal(
+    calls.filter((c) =>
+      c.url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template"),
+    ).length,
+    0,
+    "no API call for invalid payload",
+  );
+});
+
+test("createGroupMsgTemplate rejects missing sender or empty chat list", async () => {
+  const { fetchImpl } = fakeGroupPush();
+  const config = resolveWecomConfig(baseBindings);
+  await assert.rejects(
+    () =>
+      createGroupMsgTemplate(config, {
+        sender: "",
+        chatIdList: ["wr-normal-1"],
+        text: "内容",
+      }, fetchImpl),
+    /群主/,
+  );
+  await assert.rejects(
+    () =>
+      createGroupMsgTemplate(config, {
+        sender: "zhangsan",
+        chatIdList: [],
+        text: "内容",
+      }, fetchImpl),
+    /客户群列表为空/,
+  );
+});
--- a/tests/wecom-group-push.test.mjs
+++ b/tests/wecom-group-push.test.mjs
@@ -0,0 +1,262 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import {
+  buildTaskGroupPushMessage,
+  groupChatsByOwner,
+  pushTaskToGroupChats,
+  trimToBytes,
+} from "../lib/wecom-group-push-service.ts";
+import {
+  clearWecomAccessTokenCacheForTests,
+  resolveWecomConfig,
+  WecomClientError,
+} from "../lib/wecom-client.ts";
+
+const baseBindings = {
+  WECOM_CORP_ID: "corp-test",
+  WECOM_AGENT_ID: "1000001",
+  WECOM_SECRET: "secret-test",
+};
+
+const baseTask = {
+  id: "task-1",
+  name: "测试任务",
+  brand: "品牌A",
+  quantity: 3,
+  due_at: "2026-09-01",
+  task_type: "content_publish",
+  platform: "小红书",
+  content_format: "image_text",
+  share_token: "tok-123",
+};
+
+test.beforeEach(() => {
+  clearWecomAccessTokenCacheForTests();
+});
+
+test("trimToBytes cuts on UTF-8 boundaries", () => {
+  assert.equal(trimToBytes("abc", 10), "abc");
+  assert.equal(trimToBytes("abcdef", 3), "abc");
+  // 「汉」占 3 字节
+  assert.equal(trimToBytes("a汉b", 4), "a汉");
+  assert.equal(trimToBytes("汉汉汉", 7), "汉汉");
+});
+
+test("buildTaskGroupPushMessage composes text and link with claim url", () => {
+  const message = buildTaskGroupPushMessage(baseTask, "https://portal.example.com/");
+  assert.match(message.text, /【新任务】测试任务/);
+  assert.match(message.text, /品牌：品牌A｜数量：3 份｜截止：2026-09-01/);
+  assert.match(message.text, /平台：小红书｜形式：图文/);
+  assert.match(message.text, /领取链接：https:\/\/portal\.example\.com\/\?task=tok-123/);
+  assert.ok(message.link);
+  assert.equal(message.link.title, "测试任务");
+  assert.equal(message.link.url, "https://portal.example.com/?task=tok-123");
+  assert.match(message.link.desc, /品牌A · 3 份 · 截止 2026-09-01/);
+});
+
+test("buildTaskGroupPushMessage marks video and screenshot tasks", () => {
+  const video = buildTaskGroupPushMessage(
+    { ...baseTask, content_format: "video" },
+    "https://portal.example.com",
+  );
+  assert.match(video.text, /形式：视频/);
+  const screenshot = buildTaskGroupPushMessage(
+    { ...baseTask, task_type: "screenshot_collect" },
+    "https://portal.example.com",
+  );
+  assert.match(screenshot.text, /形式：截图回收/);
+});
+
+test("buildTaskGroupPushMessage drops link when share_token missing", () => {
+  const message = buildTaskGroupPushMessage(
+    { ...baseTask, share_token: null },
+    "https://portal.example.com",
+  );
+  assert.equal(message.link, null);
+  assert.match(message.text, /领取链接生成失败/);
+});
+
+test("buildTaskGroupPushMessage trims long content to byte limits", () => {
+  const longName = "长".repeat(200);
+  const longBrand = "牌".repeat(400);
+  const message = buildTaskGroupPushMessage(
+    { ...baseTask, name: longName, brand: longBrand },
+    "https://portal.example.com",
+  );
+  assert.ok(Buffer.byteLength(message.text, "utf8") <= 4000);
+  assert.ok(Buffer.byteLength(message.link.title, "utf8") <= 128);
+  assert.ok(Buffer.byteLength(message.link.desc, "utf8") <= 512);
+});
+
+test("groupChatsByOwner groups chat ids by owner and skips empty owners", () => {
+  const grouped = groupChatsByOwner([
+    { chat_id: "wr-1", name: "一", owner_user_id: "zhangsan", member_count: 1, status: 0 },
+    { chat_id: "wr-2", name: "二", owner_user_id: "lisi", member_count: 2, status: 0 },
+    { chat_id: "wr-3", name: "三", owner_user_id: "zhangsan", member_count: 3, status: 0 },
+    { chat_id: "wr-4", name: "四", owner_user_id: "  ", member_count: 4, status: 0 },
+  ]);
+  assert.deepEqual(grouped, [
+    { ownerUserId: "zhangsan", chatIds: ["wr-1", "wr-3"] },
+    { ownerUserId: "lisi", chatIds: ["wr-2"] },
+  ]);
+});
+
+function fakeGroupPushDb(taskRow) {
+  const executed = [];
+  const groupRows = [
+    { chat_id: "wr-1", name: "一群", owner_user_id: "zhangsan", member_count: 10, status: 0 },
+    { chat_id: "wr-2", name: "二群", owner_user_id: "lisi", member_count: 5, status: 0 },
+  ];
+  const makeStatement = (sql, params = []) => ({
+    sql,
+    params,
+    bind(...args) {
+      return makeStatement(sql, args);
+    },
+    async run() {
+      executed.push({ sql, params });
+      return {};
+    },
+    async first() {
+      executed.push({ sql, params });
+      if (/FROM tasks WHERE id = \?/.test(sql)) {
+        return params[0] === taskRow.id ? taskRow : null;
+      }
+      return null;
+    },
+    async all() {
+      executed.push({ sql, params });
+      if (/FROM wecom_group_chats/.test(sql)) {
+        return {
+          results: sql.includes(" IN (")
+            ? groupRows.filter((row) => params.includes(row.chat_id))
+            : groupRows,
+        };
+      }
+      return { results: [] };
+    },
+  });
+  return {
+    executed,
+    prepare(sql) {
+      return makeStatement(sql);
+    },
+    async batch(statements) {
+      for (const statement of statements) await statement.run();
+    },
+  };
+}
+
+function fakeTemplateFetch(calls) {
+  return async (input, init = {}) => {
+    const url = new URL(String(input));
+    calls.push({ url, init });
+    if (url.pathname.endsWith("/cgi-bin/gettoken")) {
+      return Response.json({
+        errcode: 0,
+        errmsg: "ok",
+        access_token: "token-test",
+        expires_in: 7200,
+      });
+    }
+    if (url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template")) {
+      return Response.json({ errcode: 0, errmsg: "ok", msgid: "msg-1", fail_list: [] });
+    }
+    return new Response("not found", { status: 404 });
+  };
+}
+
+test("pushTaskToGroupChats batches per owner and records each send", async () => {
+  process.env.KOC_PORTAL_URL = "https://portal.example.com";
+  try {
+    const db = fakeGroupPushDb(baseTask);
+    const calls = [];
+    const summary = await pushTaskToGroupChats(
+      db,
+      { taskId: "task-1", chatIds: ["wr-2", "wr-1"] },
+      resolveWecomConfig(baseBindings),
+      fakeTemplateFetch(calls),
+    );
+    assert.equal(summary.results.length, 2);
+    const senders = summary.results.map((item) => item.sender).sort();
+    assert.deepEqual(senders, ["lisi", "zhangsan"]);
+    const templateCalls = calls.filter((c) =>
+      c.url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template"),
+    );
+    assert.equal(templateCalls.length, 2, "one call per owner");
+    const bodies = templateCalls.map((c) => JSON.parse(String(c.init.body)));
+    const bySender = new Map(bodies.map((body) => [body.sender, body]));
+    assert.deepEqual(bySender.get("zhangsan").chat_id_list, ["wr-1"]);
+    assert.deepEqual(bySender.get("lisi").chat_id_list, ["wr-2"]);
+    assert.match(bySender.get("zhangsan").text.content, /【新任务】测试任务/);
+    assert.equal(
+      bySender.get("zhangsan").attachments[0].link.url,
+      "https://portal.example.com/?task=tok-123",
+    );
+    const inserts = db.executed.filter((item) =>
+      item.sql.includes("INSERT INTO wecom_group_pushes"),
+    );
+    assert.equal(inserts.length, 2);
+  } finally {
+    delete process.env.KOC_PORTAL_URL;
+  }
+});
+
+test("pushTaskToGroupChats rejects unknown chat ids", async () => {
+  const db = fakeGroupPushDb(baseTask);
+  const calls = [];
+  await assert.rejects(
+    () =>
+      pushTaskToGroupChats(
+        db,
+        { taskId: "task-1", chatIds: ["wr-404"] },
+        resolveWecomConfig(baseBindings),
+        fakeTemplateFetch(calls),
+      ),
+    (error) => {
+      assert.ok(error instanceof WecomClientError);
+      assert.equal(error.status, 400);
+      assert.match(error.message, /未同步/);
+      return true;
+    },
+  );
+});
+
+test("pushTaskToGroupChats rejects missing task", async () => {
+  const db = fakeGroupPushDb(baseTask);
+  await assert.rejects(
+    () =>
+      pushTaskToGroupChats(
+        db,
+        { taskId: "task-404", chatIds: ["wr-1"] },
+        resolveWecomConfig(baseBindings),
+        fakeTemplateFetch([]),
+      ),
+    (error) => {
+      assert.ok(error instanceof WecomClientError);
+      assert.equal(error.status, 404);
+      return true;
+    },
+  );
+});
+
+test("pushTaskToGroupChats prefers explicit text over template", async () => {
+  process.env.KOC_PORTAL_URL = "https://portal.example.com";
+  try {
+    const db = fakeGroupPushDb(baseTask);
+    const calls = [];
+    await pushTaskToGroupChats(
+      db,
+      { taskId: "task-1", chatIds: ["wr-1"], text: "运营手动编辑的文案" },
+      resolveWecomConfig(baseBindings),
+      fakeTemplateFetch(calls),
+    );
+    const templateCall = calls.find((c) =>
+      c.url.pathname.endsWith("/cgi-bin/externalcontact/add_msg_template"),
+    );
+    const body = JSON.parse(String(templateCall.init.body));
+    assert.equal(body.text.content, "运营手动编辑的文案");
+  } finally {
+    delete process.env.KOC_PORTAL_URL;
+  }
+});
--- a/tests/wecom-notifier.test.mjs
+++ b/tests/wecom-notifier.test.mjs
@@ -0,0 +1,15 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import { computeDueCutoff } from "../lib/wecom-notifier-service.ts";
+
+test("computeDueCutoff advances by dueDays and crosses month/year", () => {
+  assert.equal(computeDueCutoff("2026-08-18", 3), "2026-08-21");
+  assert.equal(computeDueCutoff("2026-08-30", 3), "2026-09-02");
+  assert.equal(computeDueCutoff("2026-12-30", 3), "2027-01-02");
+  assert.equal(computeDueCutoff("2026-08-18", 0), "2026-08-18");
+  assert.equal(computeDueCutoff("2026-08-18", 10), "2026-08-28");
+});
+
+test("computeDueCutoff returns input unchanged when malformed", () => {
+  assert.equal(computeDueCutoff("not-a-date", 3), "not-a-date");
+});
--- a/tests/workbook-image.test.mjs
+++ b/tests/workbook-image.test.mjs
@@ -0,0 +1,98 @@
+import assert from "node:assert/strict";
+import test from "node:test";
+import sharp from "sharp";
+import { normalizeWorkbookImage } from "../lib/workbook-image.ts";
+
+test("bakes EXIF orientation into exported workbook image pixels", async () => {
+  const source = await sharp({
+    create: {
+      width: 8,
+      height: 4,
+      channels: 3,
+      background: "#e95420",
+    },
+  })
+    .jpeg()
+    .withMetadata({ orientation: 6 })
+    .toBuffer();
+
+  const normalized = await normalizeWorkbookImage({
+    bytes: new Uint8Array(source),
+    contentType: "image/jpeg",
+    width: 8,
+    height: 4,
+    description: "手机照片",
+  });
+  const metadata = await sharp(normalized.bytes).metadata();
+
+  assert.equal(normalized.contentType, "image/png");
+  assert.equal(normalized.width, 4);
+  assert.equal(normalized.height, 8);
+  assert.equal(metadata.width, 4);
+  assert.equal(metadata.height, 8);
+  assert.equal(metadata.orientation, undefined);
+});
+
+test("keeps unsupported image bytes unchanged", async () => {
+  const bytes = Uint8Array.from([1, 2, 3]);
+  const normalized = await normalizeWorkbookImage({
+    bytes,
+    contentType: "application/octet-stream",
+    description: "未知文件",
+  });
+  assert.equal(normalized.bytes, bytes);
+  assert.equal(normalized.contentType, "application/octet-stream");
+});
+
+test("normalizes recognizable images even when storage metadata has no image MIME type", async () => {
+  const source = await sharp({
+    create: {
+      width: 8,
+      height: 4,
+      channels: 3,
+      background: "#22c55e",
+    },
+  })
+    .jpeg()
+    .withMetadata({ orientation: 6 })
+    .toBuffer();
+  const normalized = await normalizeWorkbookImage({
+    bytes: new Uint8Array(source),
+    contentType: "application/octet-stream",
+    description: "方向元数据缺失测试",
+  });
+
+  assert.equal(normalized.contentType, "image/png");
+  assert.equal(normalized.width, 4);
+  assert.equal(normalized.height, 8);
+});
+
+test("can downsize full-resolution source images for compact workbook exports", async () => {
+  const source = await sharp({
+    create: {
+      width: 4000,
+      height: 3000,
+      channels: 3,
+      background: "#d4a72c",
+    },
+  })
+    .png({ compressionLevel: 0 })
+    .toBuffer();
+  const normalized = await normalizeWorkbookImage(
+    {
+      bytes: new Uint8Array(source),
+      contentType: "image/png",
+      width: 4000,
+      height: 3000,
+      description: "批量回填原图",
+    },
+    { maxDimension: 1600, outputFormat: "jpeg", jpegQuality: 82 },
+  );
+  const metadata = await sharp(normalized.bytes).metadata();
+
+  assert.equal(normalized.contentType, "image/jpeg");
+  assert.equal(normalized.width, 1600);
+  assert.equal(normalized.height, 1200);
+  assert.equal(metadata.format, "jpeg");
+  assert.ok(normalized.bytes.byteLength < source.byteLength / 20);
+});
Author	SHA1	Message	Date
ABAPPLO	9697b5890d	feat: 任务发布到企微客户群（企业群发）与资源库单条新增 - 任务中心新增「发布到企微群」：同步客户群清单（wecom_group_chats）、按群主分组创建企业群发任务（add_msg_template）、发送记录落库 wecom_group_pushes，迁移 mysql/0009；lib/wecom-client.ts 补 listCustomerGroupChats/createGroupMsgTemplate - KOC 资源库支持单条新增：app/api/resources-insert + lib/resource-write，拆出资源写入公共逻辑供导入复用；0008 补合作方外部联系人字段 - 环境变量示例补企微凭证与 SEED_DEMO_DATA；next.config 增加 allowedDevOrigins；CLAUDE.md 补充项目说明 - .gitignore 排除 .codegraph/ 与 .ipynb_checkpoints/	2026-08-20 15:16:18 +08:00
林炳松	8f7ea0558d	Merge pull request 'feat: 接入企业微信通知（临期催办 + 群机器人汇总）' (#3 ) from feat/account-tags into main Reviewed-on: #3	2026-08-18 09:12:59 +00:00
ABAPPLO	f2ac751c4c	feat: 接入企业微信通知（临期催办 + 群机器人汇总）新增 lib/wecom-client.ts 与 lib/wecom-notifier-service.ts，支持群机器人 webhook 与应用消息双通道；scheduler 加入临期 N 天催办与每日管理员汇总；action 路由补 bind_wecom_external_id 与 send_test_wecom 两个管理端动作，配套测试。	2026-08-18 17:05:16 +08:00
巫凤萍	cac6c5e83b	fix: 修复回填更新与截图刷新	2026-08-16 22:22:50 +08:00
巫凤萍	74671a9b9f	docs: 更新 main 私有化部署指南	2026-08-15 03:57:50 +08:00
巫凤萍	f37d05dd88	feat: 完善视频任务与 KOC 资源库	2026-08-15 03:53:09 +08:00