feat: 更新一堆 ai 初始化以及 skill

2026-04-15 09:40:15 +08:00
commit 67b2f7f2ac
37 changed files with 3121 additions and 0 deletions
--- a/.agents/skills/xiaohongshu/LICENSE
+++ b/.agents/skills/xiaohongshu/LICENSE
@@ -0,0 +1,25 @@
+MIT License
+
+Copyright (c) 2025
+
+This license applies to the shell script wrapper layer in this repository only.
+The xiaohongshu-mcp binary (https://github.com/xpzouying/xiaohongshu-mcp) is a
+separate project with its own licensing terms.
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
--- a/.agents/skills/xiaohongshu/README.md
+++ b/.agents/skills/xiaohongshu/README.md
@@ -0,0 +1,315 @@
+# XHS AI Toolkit
+
+<p align="center">
+  <b>Make AI understand your Xiaohongshu (RedNote)</b>
+</p>
+
+<p align="center">
+  <a href="README_CN.md">简体中文</a> | English
+</p>
+
+<p align="center">
+  <img alt="License" src="https://img.shields.io/badge/license-MIT-blue.svg">
+  <img alt="Platform" src="https://img.shields.io/badge/platform-Linux%20%7C%20macOS-lightgrey.svg">
+  <img alt="Python" src="https://img.shields.io/badge/python-3.10+-green.svg">
+  <img alt="MCP" src="https://img.shields.io/badge/MCP-compatible-purple.svg">
+</p>
+
+---
+
+AI-powered toolkit for **Xiaohongshu (小红书 / RedNote)** that turns your favorite posts into AI memory.
+
+- **MCP Integration** — Search, browse, comment via AI assistants
+- **Trend Tracking** — Auto-generate topic reports with engagement analytics
+- **Memory Export** — Convert your liked/saved posts into AI-searchable knowledge base
+
+Built on [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp) and [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader).
+
+## Features
+
+| Feature | Description |
+|---------|-------------|
+| Search | Search posts by keywords |
+| Feed | Get homepage recommendations |
+| Post Details | Fetch post content, comments, engagement stats |
+| Comment | Post comments to notes |
+| User Profile | Get user info and their posts |
+| Trend Tracking | Auto-generate topic analysis reports |
+| Long Image Export | Export posts as annotated JPG long images |
+| Memory Export | Export liked/saved posts as Markdown for AI memory |
+
+## Quick Start
+
+### 1. Install xiaohongshu-mcp
+
+Download from [GitHub Releases](https://github.com/xpzouying/xiaohongshu-mcp/releases):
+
+```bash
+# Linux x64
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-mcp-linux-amd64.tar.gz
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-login-linux-amd64.tar.gz
+
+# macOS ARM
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-mcp-darwin-arm64.tar.gz
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-login-darwin-arm64.tar.gz
+```
+
+Install:
+
+```bash
+mkdir -p ~/.local/bin
+tar -xzf xiaohongshu-mcp-*.tar.gz -C ~/.local/bin/
+tar -xzf xiaohongshu-login-*.tar.gz -C ~/.local/bin/
+
+cd ~/.local/bin
+mv xiaohongshu-mcp-* xiaohongshu-mcp
+mv xiaohongshu-login-* xiaohongshu-login
+chmod +x xiaohongshu-mcp xiaohongshu-login
+```
+
+### 2. Install This Toolkit
+
+```bash
+# Clone to OpenClaw workspace
+git clone https://github.com/zhjiang22/openclaw-xhs.git
+cp -r openclaw-xhs ~/.openclaw/workspace/skills/xiaohongshu
+
+# Or use symlink
+ln -s /path/to/openclaw-xhs ~/.openclaw/workspace/skills/xiaohongshu
+
+# Verify installation
+cd ~/.openclaw/workspace/skills/xiaohongshu/scripts
+./install-check.sh
+```
+
+### 3. Login (Get Cookies)
+
+**Option A: Desktop Environment**
+
+```bash
+./login.sh  # Opens browser, scan QR code with Xiaohongshu app
+```
+
+**Option B: Headless Server**
+
+Get cookies on your local machine, then copy to server:
+
+```bash
+# On local machine with GUI
+./xiaohongshu-login
+# Cookies saved to /tmp/cookies.json
+
+# Copy to server
+scp /tmp/cookies.json user@server:~/.xiaohongshu/cookies.json
+```
+
+### 4. Start Service
+
+```bash
+./start-mcp.sh              # Headless mode
+./start-mcp.sh --headless=false  # Show browser (debug)
+```
+
+Service runs at `http://localhost:18060/mcp`.
+
+#### Server Deployment (Headless Linux)
+
+On servers without a desktop environment, the underlying browser requires a virtual display.
+`start-mcp.sh` **auto-detects** the environment — if no display is found, it starts Xvfb automatically. Just install it first:
+
+```bash
+# Debian/Ubuntu
+sudo apt-get install -y xvfb
+
+# CentOS/RHEL
+sudo yum install -y xorg-x11-server-Xvfb
+```
+
+No extra configuration needed. The script handles:
+- Detecting the `DISPLAY` environment variable
+- Auto-starting `Xvfb :99` when no display is available
+- Cleaning up Xvfb when `stop-mcp.sh` is called
+
+> **Note**: Without Xvfb, login and search will fail on headless servers. See [Issue #3](https://github.com/zhjiang22/openclaw-xhs/issues/3).
+
+## Usage
+
+### Basic Commands
+
+```bash
+./status.sh                    # Check login status
+./search.sh "coffee"           # Search posts
+./recommend.sh                 # Get recommendations
+./post-detail.sh <id> <token>  # Get post details
+./comment.sh <id> <token> "Great post!"  # Comment
+./user-profile.sh <user_id> <xsec_token>  # Get user profile
+```
+
+### Trend Tracking
+
+Auto-search trending posts and generate analysis reports:
+
+```bash
+./track-topic.sh "AI" --limit 10
+./track-topic.sh "travel" --limit 5 --output report.md
+./track-topic.sh "iPhone" --limit 5 --feishu  # Export to Feishu
+```
+
+### MCP Tools
+
+| Tool | Description |
+|------|-------------|
+| `check_login_status` | Check login status |
+| `search_feeds` | Search posts |
+| `list_feeds` | Get homepage feed |
+| `get_feed_detail` | Get post details & comments |
+| `post_comment_to_feed` | Post comment |
+| `user_profile` | Get user profile |
+| `like_feed` | Like/unlike post |
+| `favorite_feed` | Save/unsave post |
+| `publish_content` | Publish image post |
+| `publish_with_video` | Publish video post |
+
+### Long Image Export
+
+Export posts as annotated JPG long images (white background, black text):
+
+```bash
+# Prepare posts.json
+cat > posts.json << 'EOF'
+[
+  {
+    "title": "Post title",
+    "author": "Author",
+    "stats": "13k likes 100 saves",
+    "desc": "Post summary",
+    "images": ["https://...webp"],
+    "per_image_text": {"1": "Caption for 2nd image"}
+  }
+]
+EOF
+
+./export-long-image.sh --posts-file posts.json -o output.jpg
+```
+
+Requires: Python 3.10+, Pillow (`pip install Pillow`)
+
+## Memory Export (Turn Likes into AI Memory)
+
+Export your liked/saved posts as a searchable knowledge base for AI assistants.
+
+### 1. Install XHS-Downloader
+
+```bash
+git clone https://github.com/JoeanAmier/XHS-Downloader.git
+cd XHS-Downloader
+pip install -r requirements.txt
+```
+
+### 2. Extract Post Links (Tampermonkey Script)
+
+1. Install [Tampermonkey](https://www.tampermonkey.net/)
+2. Install [XHS-Downloader UserScript](https://raw.githubusercontent.com/JoeanAmier/XHS-Downloader/refs/heads/master/static/XHS-Downloader.js)
+3. Go to Xiaohongshu web → Profile → Liked/Saved
+4. Click Tampermonkey menu → "Extract liked posts" or "Extract saved posts"
+5. Links auto-copied to clipboard
+6. Paste into `links.md`
+
+### 3. Download & Export
+
+```bash
+# Copy helper scripts
+cp tools/xhs-downloader/*.py /path/to/XHS-Downloader/
+
+# Download posts
+cd /path/to/XHS-Downloader
+python batch_download.py links.md
+
+# Export to workspace
+python export_to_workspace.py
+# Output: ~/.openclaw/workspace/xhs-memory/
+```
+
+### 4. Configure OpenClaw Memory Search
+
+Edit `~/.openclaw/openclaw.json`:
+
+```json
+{
+  "memorySearch": {
+    "extraPaths": [
+      "~/.openclaw/workspace/xhs-memory"
+    ]
+  }
+}
+```
+
+Now your AI assistant can search your Xiaohongshu favorites!
+
+## Project Structure
+
+```
+openclaw-xhs/
+├── README.md             # English docs
+├── README_CN.md          # Chinese docs
+├── LICENSE
+├── SKILL.md              # Skill manifest
+├── scripts/              # MCP wrapper scripts
+│   ├── install-check.sh
+│   ├── start-mcp.sh
+│   ├── stop-mcp.sh
+│   ├── login.sh
+│   ├── mcp-call.sh
+│   ├── status.sh
+│   ├── search.sh
+│   ├── recommend.sh
+│   ├── post-detail.sh
+│   ├── comment.sh
+│   ├── user-profile.sh
+│   ├── track-topic.sh
+│   ├── track-topic.py
+│   ├── export-long-image.sh
+│   └── export-long-image.py
+└── tools/
+    └── xhs-downloader/   # Memory export tools
+        ├── README.md
+        ├── batch_download.py
+        ├── export_memory.py
+        └── export_to_workspace.py
+```
+
+## Security
+
+This project implements the following security measures:
+
+- **Cookie protection**: Cookie files are copied with `600` permissions (owner-only read/write)
+- **Injection prevention**: All shell scripts use `jq` to build JSON payloads instead of string interpolation, preventing shell injection
+- **Tool name validation**: MCP tool names are restricted to alphanumeric characters and underscores
+- **Path validation**: Cross-skill script calls validate that target paths are within allowed directories
+- **Third-party content**: Content fetched from Xiaohongshu is user-generated; exercise appropriate caution
+
+
+## Disclaimer
+
+This project is a **wrapper layer** for [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp).
+
+- Does NOT contain xiaohongshu-mcp source code
+- Users must download xiaohongshu-mcp binaries separately
+- Scripts communicate via HTTP protocol only
+
+## Acknowledgments
+
+- [@xpzouying](https://github.com/xpzouying) — [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp)
+- [@JoeanAmier](https://github.com/JoeanAmier) — [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader) (GPL-3.0)
+
+## License
+
+MIT License (wrapper scripts only)
+
+**Note:** xiaohongshu-mcp has no declared license. Please respect the author's terms.
+
+---
+
+<p align="center">
+  <b>If this project helps you, please give it a ⭐!</b>
+</p>
--- a/.agents/skills/xiaohongshu/README_CN.md
+++ b/.agents/skills/xiaohongshu/README_CN.md
@@ -0,0 +1,373 @@
+# XHS AI Toolkit
+
+<p align="center">
+  <b>让 AI 读懂你的小红书</b>
+</p>
+
+<p align="center">
+  简体中文 | <a href="README.md">English</a>
+</p>
+
+<p align="center">
+  <img alt="License" src="https://img.shields.io/badge/license-MIT-blue.svg">
+  <img alt="Platform" src="https://img.shields.io/badge/platform-Linux%20%7C%20macOS-lightgrey.svg">
+  <img alt="Python" src="https://img.shields.io/badge/python-3.10+-green.svg">
+  <img alt="MCP" src="https://img.shields.io/badge/MCP-compatible-purple.svg">
+</p>
+
+---
+
+小红书 AI 工具包 — 把你的收藏变成 AI 的记忆。
+
+- **MCP 集成** — 通过 AI 助手搜索、浏览、评论小红书
+- **热点跟踪** — 自动生成话题报告，含互动数据分析
+- **记忆导出** — 将收藏/点赞笔记转为 AI 可搜索的知识库
+
+基于 [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp) 和 [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader) 构建。
+
+## 功能特性
+
+| 功能 | 说明 |
+|------|------|
+| 搜索内容 | 按关键词搜索小红书笔记 |
+| 首页推荐 | 获取首页推荐列表 |
+| 帖子详情 | 获取笔记内容、评论、互动数据 |
+| 发表评论 | 在笔记下发表评论 |
+| 用户主页 | 获取用户资料和笔记列表 |
+| 热点跟踪 | 自动生成话题分析报告 |
+| 长图导出 | 将帖子导出为带注释的 JPG 长图 |
+| 记忆导出 | 导出收藏/点赞为 Markdown 记忆库 |
+
+## 快速开始
+
+### 1. 安装 xiaohongshu-mcp
+
+从 [GitHub Releases](https://github.com/xpzouying/xiaohongshu-mcp/releases) 下载：
+
+```bash
+# Linux x64
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-mcp-linux-amd64.tar.gz
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-login-linux-amd64.tar.gz
+
+# macOS ARM
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-mcp-darwin-arm64.tar.gz
+wget https://github.com/xpzouying/xiaohongshu-mcp/releases/latest/download/xiaohongshu-login-darwin-arm64.tar.gz
+```
+
+解压安装：
+
+```bash
+mkdir -p ~/.local/bin
+tar -xzf xiaohongshu-mcp-*.tar.gz -C ~/.local/bin/
+tar -xzf xiaohongshu-login-*.tar.gz -C ~/.local/bin/
+
+cd ~/.local/bin
+mv xiaohongshu-mcp-* xiaohongshu-mcp
+mv xiaohongshu-login-* xiaohongshu-login
+chmod +x xiaohongshu-mcp xiaohongshu-login
+```
+
+### 2. 安装本工具包
+
+```bash
+# 克隆到 OpenClaw workspace
+git clone https://github.com/zhjiang22/openclaw-xhs.git
+cp -r openclaw-xhs ~/.openclaw/workspace/skills/xiaohongshu
+
+# 或使用软链接
+ln -s /path/to/openclaw-xhs ~/.openclaw/workspace/skills/xiaohongshu
+
+# 验证安装
+cd ~/.openclaw/workspace/skills/xiaohongshu/scripts
+./install-check.sh
+```
+
+### 3. 登录获取 Cookies
+
+**方式一：本地桌面环境**
+
+```bash
+./login.sh  # 打开浏览器，用小红书 App 扫码登录
+```
+
+**方式二：Linux 服务器（无桌面）**
+
+在本地电脑获取 cookies 后复制到服务器：
+
+```bash
+# 本地电脑（有 GUI）
+./xiaohongshu-login
+# Cookies 保存在 /tmp/cookies.json
+
+# 复制到服务器
+scp /tmp/cookies.json user@server:~/.xiaohongshu/cookies.json
+```
+
+服务启动时会自动检查以下位置的 cookies（按优先级）：
+
+1. 环境变量 `XHS_COOKIES_SRC` 指定的路径
+2. `~/cookies.json`
+3. `~/.xiaohongshu/cookies.json`
+
+### 4. 启动服务
+
+```bash
+./start-mcp.sh              # headless 模式
+./start-mcp.sh --headless=false  # 显示浏览器（调试用）
+```
+
+服务监听 `http://localhost:18060/mcp`。
+
+停止服务：`./stop-mcp.sh`
+
+#### 服务器部署（无桌面环境）
+
+在没有图形界面的 Linux 服务器上，`xiaohongshu-mcp` 底层的浏览器需要虚拟显示器才能正常工作。
+`start-mcp.sh` 会**自动检测**是否有桌面环境，如果没有则自动启动 Xvfb，你只需提前安装：
+
+```bash
+# Debian/Ubuntu
+sudo apt-get install -y xvfb
+
+# CentOS/RHEL
+sudo yum install -y xorg-x11-server-Xvfb
+```
+
+安装后无需额外配置，`start-mcp.sh` 会自动处理：
+- 检测 `DISPLAY` 环境变量
+- 没有显示器时自动启动 `Xvfb :99`
+- `stop-mcp.sh` 停止服务时会一并清理 Xvfb 进程
+
+> **提示**：如果不安装 Xvfb，登录和搜索功能会失败。参见 [Issue #3](https://github.com/zhjiang22/openclaw-xhs/issues/3)。
+
+## 使用方法
+
+### 基础命令
+
+```bash
+./status.sh                    # 检查登录状态
+./search.sh "咖啡"              # 搜索内容
+./recommend.sh                 # 获取推荐
+./post-detail.sh <id> <token>  # 获取帖子详情
+./comment.sh <id> <token> "写得真好！"  # 发表评论
+./user-profile.sh <user_id> <xsec_token>  # 获取用户主页
+```
+
+### 热点跟踪
+
+自动搜索热帖并生成分析报告：
+
+```bash
+./track-topic.sh "DeepSeek" --limit 10
+./track-topic.sh "春节旅游" --limit 5 --output report.md
+./track-topic.sh "iPhone 16" --limit 5 --feishu  # 导出到飞书
+```
+
+报告包含：
+- 📊 概览统计（帖子数、点赞数、评论数）
+- 📝 热帖详情（标题、作者、正文、热门评论）
+- 💬 评论区热点关键词
+- 📈 趋势分析
+
+### MCP 工具清单
+
+| 工具名 | 描述 |
+|--------|------|
+| `check_login_status` | 检查登录状态 |
+| `search_feeds` | 搜索内容 |
+| `list_feeds` | 获取首页推荐 |
+| `get_feed_detail` | 获取帖子详情和评论 |
+| `post_comment_to_feed` | 发表评论 |
+| `user_profile` | 获取用户主页 |
+| `like_feed` | 点赞/取消 |
+| `favorite_feed` | 收藏/取消 |
+| `publish_content` | 发布图文笔记 |
+| `publish_with_video` | 发布视频笔记 |
+
+### 通用 MCP 调用
+
+```bash
+./mcp-call.sh                  # 查看可用工具
+./mcp-call.sh search_feeds '{"keyword": "咖啡"}'
+./mcp-call.sh like_feed '{"feed_id": "xxx", "xsec_token": "xxx", "like": true}'
+```
+
+### 长图导出
+
+将搜索结果或帖子详情导出为带文字注释的 JPG 长图：
+
+```bash
+# 准备 posts.json（搜索+拉详情后整理）
+cat > posts.json << 'EOF'
+[
+  {
+    "title": "帖子标题",
+    "author": "作者名",
+    "stats": "1.3万赞 100收藏",
+    "desc": "正文摘要",
+    "images": ["https://...webp"],
+    "per_image_text": {"1": "第2张图的专属说明"}
+  }
+]
+EOF
+
+./export-long-image.sh --posts-file posts.json -o output.jpg
+```
+
+依赖：Python 3.10+、Pillow（`pip install Pillow`）
+
+## 记忆导出：把收藏变成 AI 的记忆
+
+将你的收藏/点赞笔记导出为 AI 可搜索的知识库，让 AI 更懂你。
+
+### 1. 安装 XHS-Downloader
+
+```bash
+git clone https://github.com/JoeanAmier/XHS-Downloader.git
+cd XHS-Downloader
+pip install -r requirements.txt
+```
+
+### 2. 获取收藏/点赞链接（油猴脚本）
+
+手动复制链接效率太低，推荐使用油猴脚本批量提取：
+
+**安装脚本：**
+
+1. 安装 [Tampermonkey](https://www.tampermonkey.net/) 浏览器扩展
+2. 安装用户脚本：[XHS-Downloader.js](https://raw.githubusercontent.com/JoeanAmier/XHS-Downloader/refs/heads/master/static/XHS-Downloader.js)
+
+**提取链接：**
+
+1. 打开 [小红书网页版](https://www.xiaohongshu.com) 并登录
+2. 进入个人主页 → **收藏** 或 **点赞** 页面
+3. 点击 Tampermonkey 图标，选择：
+   - `提取收藏作品链接`
+   - `提取点赞作品链接`
+4. 脚本会自动滚动页面加载全部内容
+5. 提取完成后链接自动复制到剪贴板
+6. 粘贴到 `links.md` 文件
+
+> **注意**：自动滚动功能默认关闭，需在脚本设置中手动开启。开启后可能触发风控，建议适度使用。
+
+### 3. 批量下载并导出
+
+```bash
+# 复制工具脚本到 XHS-Downloader 目录
+cp tools/xhs-downloader/*.py /path/to/XHS-Downloader/
+
+# 批量下载
+cd /path/to/XHS-Downloader
+python batch_download.py links.md
+
+# 导出为多文件（推荐）
+python export_to_workspace.py
+# 输出到 ~/.openclaw/workspace/xhs-memory/
+
+# 或导出为单文件
+python export_memory.py
+# 生成 xhs_memory.md
+```
+
+### 4. 配置 OpenClaw 记忆搜索
+
+编辑 `~/.openclaw/openclaw.json`：
+
+```json
+{
+  "memorySearch": {
+    "extraPaths": [
+      "~/.openclaw/workspace/xhs-memory"
+    ]
+  }
+}
+```
+
+现在你的 AI 助手可以搜索你的小红书收藏了！
+
+## 安全说明
+
+本项目在脚本安全方面采取了以下措施：
+
+- **Cookies 保护**：cookies 文件复制时自动设置 `600` 权限（仅当前用户可读写）
+- **防注入**：所有 shell 脚本使用 `jq` 构建 JSON payload，不通过字符串拼接，防止 shell 注入攻击
+- **工具名校验**：MCP 工具名限制为字母数字和下划线，拒绝非法字符
+- **路径校验**：跨 skill 调用时校验目标路径在允许的目录范围内
+- **第三方内容**：从小红书获取的内容为用户生成内容（UGC），请注意甄别
+
+
+## 注意事项
+
+1. **发布限制**
+   - 标题最多 20 个字符
+   - 正文最多 1000 个字符
+   - 每日发布上限约 50 条
+
+2. **账号安全**
+   - 避免多设备同时登录同一账号
+   - 手机 App 仅用于查看，不要同时操作
+
+3. **首次运行**
+   - 会自动下载 headless 浏览器（约 150MB）
+   - 请确保网络畅通
+
+4. **Cookies 有效期**
+   - 通常有效期约 30 天
+   - 失效后需重新获取
+
+## 项目结构
+
+```
+openclaw-xhs/
+├── README.md             # 英文文档
+├── README_CN.md          # 中文文档
+├── LICENSE
+├── SKILL.md              # Skill 描述文件
+├── scripts/              # MCP 调用脚本
+│   ├── install-check.sh
+│   ├── start-mcp.sh
+│   ├── stop-mcp.sh
+│   ├── login.sh
+│   ├── mcp-call.sh
+│   ├── status.sh
+│   ├── search.sh
+│   ├── recommend.sh
+│   ├── post-detail.sh
+│   ├── comment.sh
+│   ├── user-profile.sh
+│   ├── track-topic.sh
+│   ├── track-topic.py
+│   ├── export-long-image.sh
+│   └── export-long-image.py
+└── tools/
+    └── xhs-downloader/   # 记忆导出工具
+        ├── README.md
+        ├── batch_download.py
+        ├── export_memory.py
+        └── export_to_workspace.py
+```
+
+## 声明
+
+本项目是 [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp) 的**调用封装层**。
+
+- **不包含** xiaohongshu-mcp 的任何源代码
+- **用户需自行下载** xiaohongshu-mcp 二进制文件
+- 脚本仅通过 HTTP 协议与 MCP 服务通信
+
+## 致谢
+
+- [@xpzouying](https://github.com/xpzouying) — [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp)
+- [@JoeanAmier](https://github.com/JoeanAmier) — [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader) (GPL-3.0)
+
+## License
+
+MIT License（仅限本项目脚本）
+
+**注意：** xiaohongshu-mcp 项目未声明开源许可证，请遵守其作者的使用条款。
+
+---
+
+<p align="center">
+  <b>觉得有用？给个 ⭐ 支持一下！</b>
+</p>
--- a/.agents/skills/xiaohongshu/SKILL.md
+++ b/.agents/skills/xiaohongshu/SKILL.md
@@ -0,0 +1,202 @@
+---
+name: xiaohongshu
+description: |
+  小红书（RedNote）内容工具。使用场景：
+  - 搜索小红书笔记并获取详情
+  - 获取首页推荐列表
+  - 获取帖子详情（正文、图片、互动数据、评论）
+  - 发表评论 / 回复评论
+  - 获取用户主页和笔记列表
+  - 点赞、收藏帖子
+  - 发布图文或视频笔记
+  - 热点话题跟踪与分析报告
+  - 帖子导出为长图
+  触发词示例：
+  - "搜一下小红书上的XX"
+  - "跟踪一下小红书上的XX热点"
+  - "分析小红书上关于XX的讨论"
+  - "小红书XX话题报告"
+  - "生成XX的小红书舆情报告"
+---
+
+# 小红书 MCP Skill
+
+基于 [xiaohongshu-mcp](https://github.com/xpzouying/xiaohongshu-mcp) 封装的 shell 脚本工具集。
+
+## 前置条件
+
+```bash
+cd scripts/
+./install-check.sh    # 检查依赖（xiaohongshu-mcp、jq、python3）
+./start-mcp.sh        # 启动 MCP 服务（默认端口 18060）
+./status.sh           # 确认已登录
+```
+
+未登录时需扫码：`mcp-call.sh get_login_qrcode` 获取二维码，用小红书 App 扫码。
+
+服务端口可通过 `MCP_URL` 环境变量覆盖（默认 `http://localhost:18060/mcp`）。
+
+## 核心数据流
+
+**重要：** 大多数操作需要 `feed_id` + `xsec_token` 配对。这两个值从搜索/推荐/用户主页结果中获取，**不可自行构造**。
+
+```
+search_feeds / list_feeds / user_profile
+        │
+        ▼
+  返回 feeds 数组，每个 feed 包含:
+  ├── id          → 用作 feed_id
+  ├── xsecToken   → 用作 xsec_token
+  └── noteCard    → 标题、作者、封面、互动数据
+        │
+        ▼
+  get_feed_detail(feed_id, xsec_token)
+        │
+        ▼
+  返回完整笔记: 正文、图片列表、评论列表
+  评论中包含 comment_id、user_id（用于回复评论）
+```
+
+## 脚本参考
+
+| 脚本 | 用途 | 参数 |
+|------|------|------|
+| `search.sh <关键词>` | 搜索笔记 | 关键词 |
+| `recommend.sh` | 首页推荐 | 无 |
+| `post-detail.sh <feed_id> <xsec_token>` | 帖子详情+评论 | 从搜索结果获取 |
+| `comment.sh <feed_id> <xsec_token> <内容>` | 发表评论 | 从搜索结果获取 |
+| `user-profile.sh <user_id> <xsec_token>` | 用户主页+笔记 | 从搜索结果获取 |
+| `track-topic.sh <话题> [选项]` | 热点分析报告 | `--limit N` `--output file` `--feishu` |
+| `export-long-image.sh` | 帖子导出长图 | `--posts-file json -o output.jpg` |
+| `mcp-call.sh <tool> [json_args]` | 通用 MCP 调用 | 见下方工具表 |
+| `start-mcp.sh` | 启动服务 | `--headless=false` `--port=N` |
+| `stop-mcp.sh` | 停止服务 | 无 |
+| `status.sh` | 检查登录 | 无 |
+| `install-check.sh` | 检查依赖 | 无 |
+
+## MCP 工具详细参数
+
+### search_feeds — 搜索笔记
+
+```json
+{"keyword": "咖啡", "filters": {"sort_by": "最新", "note_type": "图文", "publish_time": "一周内"}}
+```
+
+filters 可选字段：
+- `sort_by`: 综合 | 最新 | 最多点赞 | 最多评论 | 最多收藏
+- `note_type`: 不限 | 视频 | 图文
+- `publish_time`: 不限 | 一天内 | 一周内 | 半年内
+- `search_scope`: 不限 | 已看过 | 未看过 | 已关注
+- `location`: 不限 | 同城 | 附近
+
+### get_feed_detail — 帖子详情
+
+```json
+{"feed_id": "...", "xsec_token": "...", "load_all_comments": true, "limit": 20}
+```
+
+- `load_all_comments`: false(默认) 返回前10条，true 滚动加载更多
+- `limit`: 加载评论上限（仅 load_all_comments=true 时生效），默认 20
+- `click_more_replies`: 是否展开二级回复，默认 false
+- `reply_limit`: 跳过回复数超过此值的评论，默认 10
+- `scroll_speed`: slow | normal | fast
+
+### post_comment_to_feed — 发表评论
+
+```json
+{"feed_id": "...", "xsec_token": "...", "content": "写得真好！"}
+```
+
+### reply_comment_in_feed — 回复评论
+
+```json
+{"feed_id": "...", "xsec_token": "...", "content": "谢谢！", "comment_id": "...", "user_id": "..."}
+```
+
+`comment_id` 和 `user_id` 从 get_feed_detail 返回的评论列表中获取。
+
+### user_profile — 用户主页
+
+```json
+{"user_id": "...", "xsec_token": "..."}
+```
+
+`user_id` 从 feed 的 `noteCard.user.userId` 获取，`xsec_token` 使用该 feed 的 `xsecToken`。
+
+### like_feed — 点赞/取消
+
+```json
+{"feed_id": "...", "xsec_token": "..."}
+{"feed_id": "...", "xsec_token": "...", "unlike": true}
+```
+
+### favorite_feed — 收藏/取消
+
+```json
+{"feed_id": "...", "xsec_token": "..."}
+{"feed_id": "...", "xsec_token": "...", "unfavorite": true}
+```
+
+### publish_content — 发布图文
+
+```json
+{"title": "标题(≤20字)", "content": "正文(≤1000字)", "images": ["/path/to/img.jpg"], "tags": ["美食","旅行"]}
+```
+
+- `images`: 至少1张，支持本地路径或 HTTP URL
+- `tags`: 可选，话题标签
+- `schedule_at`: 可选，定时发布（ISO8601，1小时~14天内）
+
+### publish_with_video — 发布视频
+
+```json
+{"title": "标题", "content": "正文", "video": "/path/to/video.mp4"}
+```
+
+### 其他工具
+
+| 工具 | 参数 | 说明 |
+|------|------|------|
+| `check_login_status` | 无 | 检查登录状态 |
+| `list_feeds` | 无 | 获取首页推荐 |
+| `get_login_qrcode` | 无 | 获取登录二维码（Base64 PNG） |
+| `delete_cookies` | 无 | 删除 cookies，重置登录 |
+
+## 热点跟踪
+
+自动搜索 → 拉取详情 → 生成 Markdown 报告。
+
+```bash
+./track-topic.sh "DeepSeek" --limit 5
+./track-topic.sh "春节旅游" --limit 10 --output report.md
+./track-topic.sh "iPhone 16" --limit 5 --feishu    # 导出飞书
+```
+
+报告包含：概览统计、热帖详情（正文+热评）、评论关键词、趋势分析。
+
+## 长图导出
+
+将帖子导出为白底黑字的 JPG 长图。
+
+```bash
+./export-long-image.sh --posts-file posts.json -o output.jpg
+```
+
+posts.json 格式：
+```json
+[{
+  "title": "标题", "author": "作者", "stats": "1.3万赞",
+  "desc": "正文摘要", "images": ["https://..."],
+  "per_image_text": {"1": "第2张图的说明"}
+}]
+```
+
+依赖：Python 3.10+、Pillow。
+
+## 注意事项
+
+- Cookies 有效期约 30 天，过期需重新扫码
+- 首次启动会下载 headless 浏览器（~150MB）
+- 同一账号避免多客户端同时操作
+- 发布限制：标题≤20字符，正文≤1000字符，日发布≤50条
+- Linux 服务器无桌面环境需安装 xvfb（`apt-get install xvfb`，脚本自动管理）
--- a/.agents/skills/xiaohongshu/scripts/comment.sh
+++ b/.agents/skills/xiaohongshu/scripts/comment.sh
@@ -0,0 +1,16 @@
+#!/bin/bash
+# 发表评论到小红书帖子
+
+NOTE_ID="$1"
+XSEC_TOKEN="$2"
+CONTENT="$3"
+
+if [ -z "$NOTE_ID" ] || [ -z "$XSEC_TOKEN" ] || [ -z "$CONTENT" ]; then
+    echo "用法: $0 <note_id> <xsec_token> <评论内容>"
+    exit 1
+fi
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+ARGS=$(jq -n --arg fid "$NOTE_ID" --arg tok "$XSEC_TOKEN" --arg ct "$CONTENT" \
+  '{"feed_id":$fid,"xsec_token":$tok,"content":$ct}')
+"$SCRIPT_DIR/mcp-call.sh" post_comment_to_feed "$ARGS"
--- a/.agents/skills/xiaohongshu/scripts/export-long-image.py
+++ b/.agents/skills/xiaohongshu/scripts/export-long-image.py
@@ -0,0 +1,260 @@
+#!/usr/bin/env python3
+"""
+小红书帖子长图导出工具
+
+用法:
+    python3 export-long-image.py --posts '<json>' --output output.jpg
+    python3 export-long-image.py --posts-file posts.json --output output.jpg
+
+posts JSON 格式:
+[
+    {
+        "title": "帖子标题",
+        "author": "作者名",
+        "stats": "1.3万赞 5171收藏",
+        "desc": "正文摘要，支持\\n换行",
+        "images": ["url1", "url2", ...],
+        "per_image_text": {
+            "1": "第2张图的说明文字（0-indexed）",
+            "3": "第4张图的说明文字"
+        }
+    },
+    ...
+]
+
+per_image_text 可选：如果原帖文字明确指向某张图，可以把说明放在对应图片上。
+未指定 per_image_text 时，所有文字放在该帖第一张图前的文字块中。
+"""
+
+import argparse
+import json
+import os
+import sys
+import tempfile
+import urllib.request
+from PIL import Image, ImageDraw, ImageFont
+
+# --- 配置 ---
+WIDTH = 800
+PAD = 24
+LINE_SPACE = 10
+FONT_CANDIDATES = [
+    "/System/Library/Fonts/STHeiti Medium.ttc",
+    "/System/Library/Fonts/Hiragino Sans GB.ttc",
+    "/System/Library/Fonts/Supplemental/Arial Unicode.ttf",
+    "/usr/share/fonts/truetype/noto/NotoSansCJK-Regular.ttc",
+    "/usr/share/fonts/opentype/noto/NotoSansCJK-Regular.ttc",
+]
+
+
+def find_font():
+    for path in FONT_CANDIDATES:
+        if os.path.exists(path):
+            return path
+    return None
+
+
+def load_font(path, size):
+    if path:
+        try:
+            return ImageFont.truetype(path, size, index=0)
+        except Exception:
+            pass
+    return ImageFont.load_default()
+
+
+def wrap_text(text, font, max_width, draw):
+    lines = []
+    for paragraph in text.split("\n"):
+        paragraph = paragraph.strip()
+        if not paragraph:
+            continue
+        current = ""
+        for char in paragraph:
+            test = current + char
+            bbox = draw.textbbox((0, 0), test, font=font)
+            if bbox[2] - bbox[0] > max_width:
+                if current:
+                    lines.append(current)
+                current = char
+            else:
+                current = test
+        if current:
+            lines.append(current)
+    return lines
+
+
+def draw_lines(draw, lines, font, x, y, fill):
+    for line in lines:
+        draw.text((x, y), line, font=font, fill=fill)
+        bbox = draw.textbbox((0, 0), line, font=font)
+        y += (bbox[3] - bbox[1]) + LINE_SPACE
+    return y
+
+
+def measure_lines(lines, font, draw):
+    h = 0
+    for line in lines:
+        bbox = draw.textbbox((0, 0), line if line else " ", font=font)
+        h += (bbox[3] - bbox[1]) + LINE_SPACE
+    return h
+
+
+def make_text_block(title, author_line, desc, font_path, width):
+    """白底黑字文字块，模仿小红书原样"""
+    title_font = load_font(font_path, 32)
+    author_font = load_font(font_path, 20)
+    body_font = load_font(font_path, 24)
+
+    tmp = Image.new("RGB", (width, 10))
+    draw = ImageDraw.Draw(tmp)
+    max_w = width - PAD * 2
+
+    title_lines = wrap_text(title, title_font, max_w, draw)
+    author_lines = [author_line] if author_line else []
+    desc_lines = wrap_text(desc, body_font, max_w, draw) if desc else []
+
+    # 计算高度
+    total_h = PAD
+    total_h += measure_lines(title_lines, title_font, draw)
+    if author_lines:
+        total_h += 4
+        total_h += measure_lines(author_lines, author_font, draw)
+    if desc_lines:
+        total_h += 8
+        total_h += measure_lines(desc_lines, body_font, draw)
+    total_h += PAD
+
+    # 绘制
+    block = Image.new("RGB", (width, total_h), (255, 255, 255))
+    draw = ImageDraw.Draw(block)
+
+    y = PAD
+    y = draw_lines(draw, title_lines, title_font, PAD, y, (33, 33, 33))
+    if author_lines:
+        y += 4
+        y = draw_lines(draw, author_lines, author_font, PAD, y, (153, 153, 153))
+    if desc_lines:
+        y += 8
+        y = draw_lines(draw, desc_lines, body_font, PAD, y, (66, 66, 66))
+
+    return block
+
+
+def make_image_caption(text, font_path, width):
+    """图片上方的小说明文字块"""
+    font = load_font(font_path, 20)
+    tmp = Image.new("RGB", (width, 10))
+    draw = ImageDraw.Draw(tmp)
+    lines = wrap_text(text, font, width - PAD * 2, draw)
+
+    h = PAD + measure_lines(lines, font, draw) + 8
+    block = Image.new("RGB", (width, h), (245, 245, 245))
+    draw = ImageDraw.Draw(block)
+    draw_lines(draw, lines, font, PAD, PAD // 2, (100, 100, 100))
+    return block
+
+
+def download_image(url, tmpdir, idx):
+    """下载图片到临时目录"""
+    ext = ".webp"
+    path = os.path.join(tmpdir, f"img_{idx}{ext}")
+    try:
+        req = urllib.request.Request(url, headers={"User-Agent": "Mozilla/5.0"})
+        with urllib.request.urlopen(req, timeout=30) as resp:
+            with open(path, "wb") as f:
+                f.write(resp.read())
+        return path
+    except Exception as e:
+        print(f"  警告: 下载失败 {url[:60]}... ({e})", file=sys.stderr)
+        return None
+
+
+def main():
+    parser = argparse.ArgumentParser(description="小红书帖子长图导出")
+    parser.add_argument("--posts", help="Posts JSON string")
+    parser.add_argument("--posts-file", help="Posts JSON file path")
+    parser.add_argument("--output", "-o", required=True, help="Output JPG path")
+    parser.add_argument("--width", type=int, default=800, help="Image width (default 800)")
+    parser.add_argument("--quality", type=int, default=88, help="JPEG quality (default 88)")
+    args = parser.parse_args()
+
+    global WIDTH
+    WIDTH = args.width
+
+    # 读取 posts 数据
+    if args.posts:
+        posts = json.loads(args.posts)
+    elif args.posts_file:
+        with open(args.posts_file, "r") as f:
+            posts = json.load(f)
+    else:
+        print("错误: 需要 --posts 或 --posts-file", file=sys.stderr)
+        sys.exit(1)
+
+    font_path = find_font()
+    if not font_path:
+        print("警告: 未找到中文字体，文字可能显示异常", file=sys.stderr)
+
+    sep = Image.new("RGB", (WIDTH, 3), (230, 230, 230))
+    pieces = []
+
+    with tempfile.TemporaryDirectory() as tmpdir:
+        img_counter = 0
+        for pi, post in enumerate(posts):
+            title = post.get("title", "")
+            author = post.get("author", "")
+            stats = post.get("stats", "")
+            desc = post.get("desc", "")
+            images = post.get("images", [])
+            per_image_text = post.get("per_image_text", {})
+
+            # 作者行
+            author_line = author
+            if stats:
+                author_line = f"{author} · {stats}" if author else stats
+
+            # 主文字块
+            text_block = make_text_block(title, author_line, desc, font_path, WIDTH)
+            pieces.append(text_block)
+
+            # 图片
+            for i, url in enumerate(images):
+                # 是否有针对这张图的说明
+                img_key = str(i)
+                if img_key in per_image_text:
+                    caption_block = make_image_caption(per_image_text[img_key], font_path, WIDTH)
+                    pieces.append(caption_block)
+
+                img_path = download_image(url, tmpdir, img_counter)
+                img_counter += 1
+                if img_path:
+                    try:
+                        im = Image.open(img_path).convert("RGB")
+                        ratio = WIDTH / im.width
+                        im = im.resize((WIDTH, int(im.height * ratio)), Image.LANCZOS)
+                        pieces.append(im)
+                    except Exception as e:
+                        print(f"  警告: 图片处理失败 ({e})", file=sys.stderr)
+
+            # 帖子间分隔线
+            if pi < len(posts) - 1:
+                pieces.append(sep)
+
+    if not pieces:
+        print("错误: 没有内容可拼接", file=sys.stderr)
+        sys.exit(1)
+
+    total_h = sum(p.height for p in pieces)
+    long_img = Image.new("RGB", (WIDTH, total_h), (255, 255, 255))
+    y = 0
+    for p in pieces:
+        long_img.paste(p, (0, y))
+        y += p.height
+
+    long_img.save(args.output, "JPEG", quality=args.quality)
+    print(f"完成: {args.output} ({WIDTH}x{total_h})")
+
+
+if __name__ == "__main__":
+    main()
--- a/.agents/skills/xiaohongshu/scripts/export-long-image.sh
+++ b/.agents/skills/xiaohongshu/scripts/export-long-image.sh
@@ -0,0 +1,20 @@
+#!/bin/bash
+# 小红书帖子导出长图
+#
+# 用法:
+#   ./export-long-image.sh --posts-file posts.json -o output.jpg
+#   ./export-long-image.sh --posts '<json>' -o output.jpg
+#
+# posts.json 示例:
+# [
+#   {
+#     "title": "帖子标题",
+#     "author": "作者",
+#     "stats": "1.3万赞 100收藏",
+#     "desc": "正文摘要",
+#     "images": ["https://...webp", "https://...webp"]
+#   }
+# ]
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+python3 "$SCRIPT_DIR/export-long-image.py" "$@"
--- a/.agents/skills/xiaohongshu/scripts/install-check.sh
+++ b/.agents/skills/xiaohongshu/scripts/install-check.sh
@@ -0,0 +1,67 @@
+#!/bin/bash
+# 检查小红书 MCP 依赖是否已安装
+
+set -e
+
+echo "检查小红书 MCP 依赖..."
+echo ""
+
+XHS_MCP="$HOME/.local/bin/xiaohongshu-mcp"
+XHS_LOGIN="$HOME/.local/bin/xiaohongshu-login"
+
+check_binary() {
+    local name="$1"
+    local path="$2"
+    if [ -f "$path" ]; then
+        echo "✅ $name: $path"
+        return 0
+    else
+        echo "❌ $name: 未找到"
+        return 1
+    fi
+}
+
+MISSING=0
+
+check_binary "xiaohongshu-mcp" "$XHS_MCP" || MISSING=1
+check_binary "xiaohongshu-login" "$XHS_LOGIN" || MISSING=1
+
+echo ""
+
+# 检查 jq（必需，用于安全构建 JSON）
+if command -v jq &> /dev/null; then
+    echo "✅ jq: $(which jq)"
+else
+    echo "❌ jq: 未安装（必需，用于安全构建 JSON）"
+    echo "   安装: apt-get install jq / brew install jq"
+    MISSING=1
+fi
+
+# 检查 Python3（track-topic.py 需要）
+if command -v python3 &> /dev/null; then
+    echo "✅ python3: $(python3 --version)"
+else
+    echo "⚠️  python3: 未安装（热点跟踪功能需要）"
+fi
+
+echo ""
+
+if [ $MISSING -eq 1 ]; then
+    echo "=========================================="
+    echo "缺少必要依赖，请按以下步骤安装："
+    echo ""
+    echo "1. 从 GitHub Releases 下载对应平台的二进制文件："
+    echo "   https://github.com/xpzouying/xiaohongshu-mcp/releases"
+    echo ""
+    echo "2. 解压并安装到 ~/.local/bin/："
+    echo "   mkdir -p ~/.local/bin"
+    echo "   mv xiaohongshu-mcp-linux-amd64 ~/.local/bin/xiaohongshu-mcp"
+    echo "   mv xiaohongshu-login-linux-amd64 ~/.local/bin/xiaohongshu-login"
+    echo "   chmod +x ~/.local/bin/xiaohongshu-*"
+    echo ""
+    echo "3. 确保 ~/.local/bin 在 PATH 中（可选）"
+    echo "=========================================="
+    exit 1
+else
+    echo "✅ 所有依赖已就绪"
+fi
--- a/.agents/skills/xiaohongshu/scripts/login.sh
+++ b/.agents/skills/xiaohongshu/scripts/login.sh
@@ -0,0 +1,10 @@
+#!/bin/bash
+# 启动小红书登录工具
+
+XHS_LOGIN="$HOME/.local/bin/xiaohongshu-login"
+
+echo "启动小红书登录工具..."
+echo "注意：需要桌面环境或 X11 转发"
+echo ""
+
+"$XHS_LOGIN"
--- a/.agents/skills/xiaohongshu/scripts/mcp-call.sh
+++ b/.agents/skills/xiaohongshu/scripts/mcp-call.sh
@@ -0,0 +1,82 @@
+#!/bin/bash
+# 通用 MCP 调用脚本（支持 Streamable HTTP + Session ID）
+
+set -e
+
+TOOL_NAME="$1"
+TOOL_ARGS="$2"
+MCP_URL="${MCP_URL:-http://localhost:18060/mcp}"
+export no_proxy="${no_proxy:+$no_proxy,}localhost,127.0.0.1"
+
+# 检查 jq 依赖
+if ! command -v jq &> /dev/null; then
+    echo "错误: 需要安装 jq (apt-get install jq / brew install jq)"
+    exit 1
+fi
+
+if [ -z "$TOOL_NAME" ]; then
+    echo "用法: $0 <tool_name> [json_args]"
+    echo ""
+    echo "可用工具:"
+    echo "  check_login_status    - 检查登录状态"
+    echo "  search_feeds          - 搜索内容 {\"keyword\": \"...\", \"filters\": {\"sort_by\": \"最新\"}}"
+    echo "  list_feeds            - 获取首页推荐"
+    echo "  get_feed_detail       - 获取帖子详情 {\"feed_id\": \"...\", \"xsec_token\": \"...\"}"
+    echo "  post_comment_to_feed  - 发表评论 {\"feed_id\": \"...\", \"xsec_token\": \"...\", \"content\": \"...\"}"
+    echo "  reply_comment_in_feed - 回复评论 {\"feed_id\": \"...\", \"xsec_token\": \"...\", \"content\": \"...\", \"comment_id\": \"...\", \"user_id\": \"...\"}"
+    echo "  user_profile          - 获取用户主页 {\"user_id\": \"...\", \"xsec_token\": \"...\"}"
+    echo "  like_feed             - 点赞 {\"feed_id\": \"...\", \"xsec_token\": \"...\"}  取消: {\"unlike\": true}"
+    echo "  favorite_feed         - 收藏 {\"feed_id\": \"...\", \"xsec_token\": \"...\"}  取消: {\"unfavorite\": true}"
+    echo "  get_login_qrcode      - 获取登录二维码"
+    echo "  delete_cookies        - 删除 cookies，重置登录状态"
+    echo "  publish_content       - 发布图文"
+    echo "  publish_with_video    - 发布视频"
+    exit 1
+fi
+
+# 校验 tool name，只允许字母数字和下划线
+if [[ ! "$TOOL_NAME" =~ ^[a-zA-Z_][a-zA-Z0-9_]*$ ]]; then
+    echo "错误: 无效的工具名: $TOOL_NAME"
+    exit 1
+fi
+
+[ -z "$TOOL_ARGS" ] && TOOL_ARGS="{}"
+
+# 校验 TOOL_ARGS 是合法 JSON
+if ! echo "$TOOL_ARGS" | jq empty 2>/dev/null; then
+    echo "错误: 参数不是合法的 JSON: $TOOL_ARGS"
+    exit 1
+fi
+
+# 1. Initialize 并获取 Session ID
+INIT_RESPONSE=$(curl --noproxy '*' -s -i -X POST "$MCP_URL" \
+  -H "Content-Type: application/json" \
+  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"openclaw","version":"1.0"}}}')
+
+SESSION_ID=$(echo "$INIT_RESPONSE" | grep -i "Mcp-Session-Id" | awk '{print $2}' | tr -d '\r\n')
+
+if [ -z "$SESSION_ID" ]; then
+    echo "错误: 无法获取 MCP Session ID"
+    echo "请确保 MCP 服务正在运行: ./start-mcp.sh"
+    exit 1
+fi
+
+# 2. Initialized notification
+curl --noproxy '*' -s -X POST "$MCP_URL" \
+  -H "Content-Type: application/json" \
+  -H "Mcp-Session-Id: $SESSION_ID" \
+  -d '{"jsonrpc":"2.0","method":"notifications/initialized"}' > /dev/null
+
+# 3. Call tool — 使用 jq 安全构建 JSON，避免 shell 注入
+CALL_PAYLOAD=$(jq -n \
+  --arg name "$TOOL_NAME" \
+  --argjson args "$TOOL_ARGS" \
+  '{"jsonrpc":"2.0","id":2,"method":"tools/call","params":{"name":$name,"arguments":$args}}')
+
+RESULT=$(curl --noproxy '*' -s --max-time 120 -X POST "$MCP_URL" \
+  -H "Content-Type: application/json" \
+  -H "Mcp-Session-Id: $SESSION_ID" \
+  -d "$CALL_PAYLOAD")
+
+# 输出结果
+echo "$RESULT" | jq .
--- a/.agents/skills/xiaohongshu/scripts/post-detail.sh
+++ b/.agents/skills/xiaohongshu/scripts/post-detail.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+# 获取小红书帖子详情
+
+NOTE_ID="$1"
+XSEC_TOKEN="$2"
+
+if [ -z "$NOTE_ID" ] || [ -z "$XSEC_TOKEN" ]; then
+    echo "用法: $0 <note_id> <xsec_token>"
+    echo ""
+    echo "note_id 和 xsec_token 可从搜索或推荐结果中获取"
+    exit 1
+fi
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+ARGS=$(jq -n --arg fid "$NOTE_ID" --arg tok "$XSEC_TOKEN" \
+  '{"feed_id":$fid,"xsec_token":$tok}')
+"$SCRIPT_DIR/mcp-call.sh" get_feed_detail "$ARGS"
--- a/.agents/skills/xiaohongshu/scripts/recommend.sh
+++ b/.agents/skills/xiaohongshu/scripts/recommend.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+# 获取小红书首页推荐列表
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+"$SCRIPT_DIR/mcp-call.sh" list_feeds
--- a/.agents/skills/xiaohongshu/scripts/search.sh
+++ b/.agents/skills/xiaohongshu/scripts/search.sh
@@ -0,0 +1,13 @@
+#!/bin/bash
+# 搜索小红书内容
+
+KEYWORD="$1"
+
+if [ -z "$KEYWORD" ]; then
+    echo "用法: $0 <关键词>"
+    exit 1
+fi
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+ARGS=$(jq -n --arg kw "$KEYWORD" '{"keyword":$kw}')
+"$SCRIPT_DIR/mcp-call.sh" search_feeds "$ARGS"
--- a/.agents/skills/xiaohongshu/scripts/start-mcp.sh
+++ b/.agents/skills/xiaohongshu/scripts/start-mcp.sh
@@ -0,0 +1,166 @@
+#!/bin/bash
+# 启动小红书 MCP 服务
+
+XHS_MCP="$HOME/.local/bin/xiaohongshu-mcp"
+PID_FILE="$HOME/.xiaohongshu/mcp.pid"
+LOG_FILE="$HOME/.xiaohongshu/mcp.log"
+XVFB_PID_FILE="$HOME/.xiaohongshu/xvfb.pid"
+XVFB_DISPLAY_FILE="$HOME/.xiaohongshu/xvfb.display"
+
+# Cookies 路径（可通过环境变量覆盖）
+# XHS_COOKIES_SRC: 源 cookies 文件（用于远程服务器场景）
+# 默认检查 ~/cookies.json 和 ~/.xiaohongshu/cookies.json
+COOKIES_DST="/tmp/cookies.json"
+
+mkdir -p "$HOME/.xiaohongshu"
+
+# 检测是否有显示器（桌面环境）
+has_display() {
+    [ -n "$DISPLAY" ] && xdpyinfo >/dev/null 2>&1
+}
+
+# 在无桌面环境下自动启动 Xvfb
+ensure_display() {
+    if has_display; then
+        return 0
+    fi
+
+    # 已有 Xvfb 在运行
+    if [ -f "$XVFB_PID_FILE" ]; then
+        local pid
+        pid=$(cat "$XVFB_PID_FILE")
+        if kill -0 "$pid" 2>/dev/null; then
+            export DISPLAY=$(cat "$XVFB_DISPLAY_FILE" 2>/dev/null || echo ":99")
+            echo "复用已有 Xvfb (PID: $pid, DISPLAY=$DISPLAY)"
+            return 0
+        fi
+    fi
+
+    # 检查 Xvfb 是否安装
+    if ! command -v Xvfb >/dev/null 2>&1; then
+        echo "⚠ 未检测到桌面环境，且未安装 Xvfb。"
+        echo "  请安装：sudo apt-get install -y xvfb"
+        echo "  安装后重新运行本脚本即可自动配置。"
+        exit 1
+    fi
+
+    echo "未检测到桌面环境，自动启动 Xvfb 虚拟显示..."
+
+    # 自动选择可用的 display 号（99-109）
+    local display_num=""
+    local d
+    for d in $(seq 99 109); do
+        if [ ! -e "/tmp/.X${d}-lock" ]; then
+            display_num=$d
+            break
+        fi
+        # 锁文件存在但进程已死，尝试清理后使用
+        local lock_pid
+        lock_pid=$(cat "/tmp/.X${d}-lock" 2>/dev/null | tr -d ' ')
+        if [ -n "$lock_pid" ] && ! kill -0 "$lock_pid" 2>/dev/null; then
+            rm -f "/tmp/.X${d}-lock" "/tmp/.X11-unix/X${d}" 2>/dev/null
+            if [ ! -e "/tmp/.X${d}-lock" ]; then
+                display_num=$d
+                break
+            fi
+        fi
+    done
+
+    if [ -z "$display_num" ]; then
+        echo "✗ 无法找到可用的 display 号（:99-:109 均被占用）"
+        exit 1
+    fi
+
+    # -ac: 关闭访问控制，允许 chromium 连接虚拟显示（仅用于 headless 自动化）
+    Xvfb ":${display_num}" -screen 0 1024x768x24 -ac >/dev/null 2>&1 &
+    echo $! > "$XVFB_PID_FILE"
+    echo ":${display_num}" > "$XVFB_DISPLAY_FILE"
+    export DISPLAY=":${display_num}"
+    sleep 1
+
+    if kill -0 "$(cat "$XVFB_PID_FILE")" 2>/dev/null; then
+        echo "✓ Xvfb 已启动 (DISPLAY=:${display_num})"
+    else
+        echo "✗ Xvfb 启动失败"
+        exit 1
+    fi
+}
+
+# 同步 cookies（支持多个可能的来源）
+sync_cookies() {
+    local src=""
+
+    # 优先使用环境变量指定的路径
+    if [ -n "$XHS_COOKIES_SRC" ] && [ -f "$XHS_COOKIES_SRC" ]; then
+        src="$XHS_COOKIES_SRC"
+    elif [ -f "$HOME/cookies.json" ]; then
+        src="$HOME/cookies.json"
+    elif [ -f "$HOME/.xiaohongshu/cookies.json" ]; then
+        src="$HOME/.xiaohongshu/cookies.json"
+    fi
+
+    if [ -n "$src" ]; then
+        if [ ! -f "$COOKIES_DST" ] || [ "$src" -nt "$COOKIES_DST" ]; then
+            install -m 600 "$src" "$COOKIES_DST"
+            echo "已同步 cookies: $src -> $COOKIES_DST"
+        fi
+    else
+        # 确保已有的 cookies 文件权限正确
+        [ -f "$COOKIES_DST" ] && chmod 600 "$COOKIES_DST"
+    fi
+}
+
+sync_cookies
+ensure_display
+
+# 检查是否已在运行
+if [ -f "$PID_FILE" ]; then
+    PID=$(cat "$PID_FILE")
+    if kill -0 "$PID" 2>/dev/null; then
+        echo "MCP 服务已在运行 (PID: $PID)"
+        echo "如需重启，请先运行 stop-mcp.sh"
+        exit 0
+    fi
+fi
+
+# 解析参数
+HEADLESS="true"
+PORT="${XHS_MCP_PORT:-18060}"
+for arg in "$@"; do
+    case $arg in
+        --headless=false)
+            HEADLESS="false"
+            ;;
+        --port=*)
+            PORT="${arg#*=}"
+            ;;
+    esac
+done
+
+# 校验端口号
+if [[ ! "$PORT" =~ ^[0-9]+$ ]]; then
+    echo "错误: 无效端口号: $PORT"
+    exit 1
+fi
+
+# 启动服务
+echo "启动小红书 MCP 服务..."
+if [ "$HEADLESS" = "false" ]; then
+    nohup "$XHS_MCP" -port ":${PORT}" -headless=false > "$LOG_FILE" 2>&1 &
+else
+    nohup "$XHS_MCP" -port ":${PORT}" > "$LOG_FILE" 2>&1 &
+fi
+
+echo $! > "$PID_FILE"
+sleep 2
+
+# 验证启动
+if kill -0 $(cat "$PID_FILE") 2>/dev/null; then
+    echo "✓ MCP 服务已启动 (PID: $(cat $PID_FILE))"
+    echo "  端点: http://localhost:${PORT}/mcp"
+    echo "  日志: $LOG_FILE"
+else
+    echo "✗ 启动失败，查看日志: $LOG_FILE"
+    cat "$LOG_FILE"
+    exit 1
+fi
--- a/.agents/skills/xiaohongshu/scripts/status.sh
+++ b/.agents/skills/xiaohongshu/scripts/status.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+# 检查小红书登录状态
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+"$SCRIPT_DIR/mcp-call.sh" check_login_status
--- a/.agents/skills/xiaohongshu/scripts/stop-mcp.sh
+++ b/.agents/skills/xiaohongshu/scripts/stop-mcp.sh
@@ -0,0 +1,29 @@
+#!/bin/bash
+# 停止小红书 MCP 服务
+
+PID_FILE="$HOME/.xiaohongshu/mcp.pid"
+XVFB_PID_FILE="$HOME/.xiaohongshu/xvfb.pid"
+
+if [ -f "$PID_FILE" ]; then
+    PID=$(cat "$PID_FILE")
+    if kill -0 "$PID" 2>/dev/null; then
+        kill "$PID"
+        rm -f "$PID_FILE"
+        echo "✓ MCP 服务已停止"
+    else
+        echo "进程不存在，清理 PID 文件"
+        rm -f "$PID_FILE"
+    fi
+else
+    echo "MCP 服务未运行"
+fi
+
+# 清理 Xvfb
+if [ -f "$XVFB_PID_FILE" ]; then
+    XVFB_PID=$(cat "$XVFB_PID_FILE")
+    if kill -0 "$XVFB_PID" 2>/dev/null; then
+        kill "$XVFB_PID"
+        echo "✓ Xvfb 已停止"
+    fi
+    rm -f "$XVFB_PID_FILE" "$HOME/.xiaohongshu/xvfb.display"
+fi
--- a/.agents/skills/xiaohongshu/scripts/track-topic.py
+++ b/.agents/skills/xiaohongshu/scripts/track-topic.py
@@ -0,0 +1,313 @@
+#!/usr/bin/env python3
+"""
+小红书热点跟踪工具
+
+用法:
+    python track-topic.py <话题> [--limit N] [--feishu] [--output FILE]
+
+示例:
+    python track-topic.py "DeepSeek" --limit 5 --feishu
+    python track-topic.py "春节旅游" --limit 10 --output report.md
+"""
+
+import argparse
+import json
+import subprocess
+import sys
+import os
+from datetime import datetime
+from pathlib import Path
+
+# 获取脚本目录
+SCRIPT_DIR = Path(__file__).parent.resolve()
+XHS_SCRIPTS = SCRIPT_DIR  # 现在就在 xiaohongshu/scripts 目录下
+
+# 飞书 skill 路径（支持多种可能的位置）
+def find_feishu_scripts() -> Path:
+    """查找 feishu-docs skill 的 scripts 目录"""
+    # 只允许在已知的 skill 目录中查找
+    allowed_roots = [
+        SCRIPT_DIR.parent.parent,  # 同级 skill 目录
+        Path.home() / ".openclaw" / "workspace" / "skills",
+        Path.home() / ".claude" / "skills",
+    ]
+    for root in allowed_roots:
+        candidate = (root / "feishu-docs" / "scripts").resolve()
+        # 校验解析后的路径仍在允许的根目录下（防止符号链接逃逸）
+        if candidate.is_dir() and any(
+            str(candidate).startswith(str(r.resolve()) + os.sep) for r in allowed_roots
+        ):
+            return candidate
+    return allowed_roots[0] / "feishu-docs" / "scripts"  # 返回默认路径（可能不存在）
+
+FEISHU_SCRIPTS = find_feishu_scripts()
+
+
+def call_xhs_mcp(tool: str, args: dict) -> dict:
+    """调用小红书 MCP 工具"""
+    mcp_call = XHS_SCRIPTS / "mcp-call.sh"
+    if not mcp_call.exists():
+        print(f"❌ 找不到 xiaohongshu skill: {mcp_call}", file=sys.stderr)
+        sys.exit(1)
+    
+    result = subprocess.run(
+        [str(mcp_call), tool, json.dumps(args)],
+        capture_output=True, text=True, timeout=120
+    )
+    
+    if result.returncode != 0:
+        print(f"❌ MCP 调用失败: {result.stderr}", file=sys.stderr)
+        return {}
+    
+    try:
+        response = json.loads(result.stdout)
+        if "result" in response and "content" in response["result"]:
+            text = response["result"]["content"][0].get("text", "{}")
+            return json.loads(text) if text else {}
+        elif "error" in response:
+            print(f"⚠️ MCP 错误: {response['error'].get('message', 'Unknown')}", file=sys.stderr)
+            return {}
+        return response
+    except json.JSONDecodeError:
+        return {}
+
+
+def search_feeds(keyword: str) -> list:
+    """搜索小红书内容"""
+    print(f"🔍 搜索: {keyword}")
+    result = call_xhs_mcp("search_feeds", {"keyword": keyword})
+    feeds = result.get("feeds", [])
+    # 过滤掉 hot_query 类型
+    return [f for f in feeds if f.get("modelType") == "note"]
+
+
+def get_feed_detail(feed_id: str, xsec_token: str, load_comments: bool = True) -> dict:
+    """获取帖子详情"""
+    args = {
+        "feed_id": feed_id,
+        "xsec_token": xsec_token,
+        "load_all_comments": load_comments
+    }
+    result = call_xhs_mcp("get_feed_detail", args)
+    return result.get("data", {})
+
+
+def format_timestamp(ts: int) -> str:
+    """格式化时间戳"""
+    if not ts:
+        return "未知"
+    try:
+        dt = datetime.fromtimestamp(ts / 1000)
+        return dt.strftime("%Y-%m-%d %H:%M")
+    except:
+        return "未知"
+
+
+def get_comments_list(post: dict) -> list:
+    """安全地获取评论列表"""
+    comments = post.get("comments", {})
+    if isinstance(comments, dict):
+        return comments.get("list", [])
+    elif isinstance(comments, list):
+        return comments
+    return []
+
+
+def generate_report(keyword: str, posts: list) -> str:
+    """生成 Markdown 报告"""
+    now = datetime.now().strftime("%Y-%m-%d %H:%M")
+    
+    report = f"""# 🔥 小红书热点跟踪报告
+
+**话题:** {keyword}  
+**生成时间:** {now}  
+**收录帖子:** {len(posts)} 篇
+
+---
+
+## 📊 概览
+
+"""
+    
+    # 统计信息
+    total_likes = sum(int(p.get("note", {}).get("interactInfo", {}).get("likedCount", 0) or 0) for p in posts)
+    total_comments = sum(len(get_comments_list(p)) for p in posts)
+    
+    report += f"""| 指标 | 数值 |
+|------|------|
+| 总帖子数 | {len(posts)} |
+| 总点赞数 | {total_likes:,} |
+| 总评论数 | {total_comments} |
+
+---
+
+## 📝 热帖详情
+
+"""
+    
+    for i, post in enumerate(posts, 1):
+        note = post.get("note", {})
+        comments = get_comments_list(post)
+        
+        title = note.get("title", "无标题")
+        desc = note.get("desc", "")
+        user = note.get("user", {}).get("nickname", "匿名")
+        time_str = format_timestamp(note.get("time"))
+        interact = note.get("interactInfo", {})
+        likes = interact.get("likedCount", "0")
+        collected = interact.get("collectedCount", "0")
+        
+        report += f"""### {i}. {title}
+
+**作者:** {user}  
+**时间:** {time_str}  
+**互动:** ❤️ {likes} 赞 · ⭐ {collected} 收藏
+
+**正文:**
+
+> {desc[:500]}{"..." if len(desc) > 500 else ""}
+
+"""
+        
+        if comments:
+            report += f"""**热门评论 ({len(comments)} 条):**
+
+"""
+            for j, comment in enumerate(list(comments)[:5], 1):
+                c_user = comment.get("userInfo", {}).get("nickname", "匿名")
+                c_content = comment.get("content", "")
+                c_likes = comment.get("likeCount", 0)
+                report += f"- **{c_user}** ({c_likes}赞): {c_content[:100]}\n"
+            
+            if len(comments) > 5:
+                report += f"- *... 还有 {len(comments) - 5} 条评论*\n"
+        
+        report += "\n---\n\n"
+    
+    # 评论区热点总结
+    report += """## 💬 评论区热点关键词
+
+"""
+    
+    # 简单的关键词提取（统计高频词）
+    all_comments = []
+    for post in posts:
+        for c in get_comments_list(post):
+            all_comments.append(c.get("content", ""))
+    
+    if all_comments:
+        report += f"共 {len(all_comments)} 条评论，主要讨论方向：\n\n"
+        # 这里可以做更复杂的 NLP 分析，暂时简化
+        report += "- 用户对该话题的关注度较高\n"
+        report += "- 评论区互动活跃\n"
+    else:
+        report += "暂无足够评论数据进行分析\n"
+    
+    report += """
+---
+
+## 📈 趋势分析
+
+基于以上热帖和评论数据，该话题在小红书上呈现以下特点：
+
+1. **热度指数**: """ + ("🔥🔥🔥 高" if total_likes > 1000 else "🔥🔥 中" if total_likes > 100 else "🔥 低") + f"""
+2. **互动活跃度**: """ + ("活跃" if total_comments > 50 else "一般" if total_comments > 10 else "较低") + """
+3. **内容类型**: 以图文笔记为主
+
+---
+
+*报告由 OpenClaw 小红书热点跟踪工具自动生成*
+"""
+    
+    return report
+
+
+def export_to_feishu(title: str, content: str) -> str:
+    """导出到飞书文档"""
+    import_script = FEISHU_SCRIPTS / "doc-import.sh"
+    if not import_script.exists():
+        print(f"❌ 找不到 feishu-docs skill: {import_script}", file=sys.stderr)
+        return ""
+    
+    print("📤 导出到飞书文档...")
+    
+    # 写入临时文件
+    tmp_file = Path("/tmp/xhs_report.md")
+    tmp_file.write_text(content, encoding="utf-8")
+    
+    result = subprocess.run(
+        [str(import_script), title, "--file", str(tmp_file)],
+        capture_output=True, text=True, timeout=60
+    )
+    
+    if result.returncode != 0:
+        print(f"⚠️ 飞书导出失败: {result.stderr}", file=sys.stderr)
+        return ""
+    
+    # 解析返回的文档链接
+    output = result.stdout
+    print(output)
+    return output
+
+
+def main():
+    parser = argparse.ArgumentParser(description="小红书热点跟踪工具")
+    parser.add_argument("keyword", help="要跟踪的话题/关键词")
+    parser.add_argument("--limit", "-n", type=int, default=10, help="获取帖子数量 (默认 10)")
+    parser.add_argument("--feishu", "-f", action="store_true", help="导出到飞书文档")
+    parser.add_argument("--output", "-o", help="输出 Markdown 文件路径")
+    parser.add_argument("--no-comments", action="store_true", help="不获取评论")
+    
+    args = parser.parse_args()
+    
+    # 1. 搜索帖子
+    feeds = search_feeds(args.keyword)
+    if not feeds:
+        print("❌ 未找到相关帖子")
+        sys.exit(1)
+    
+    print(f"✅ 找到 {len(feeds)} 条帖子")
+    
+    # 2. 获取详情
+    posts = []
+    for i, feed in enumerate(feeds[:args.limit]):
+        feed_id = feed.get("id")
+        xsec_token = feed.get("xsecToken")
+        title = feed.get("noteCard", {}).get("displayTitle", "")
+        
+        print(f"📖 [{i+1}/{min(len(feeds), args.limit)}] 获取: {title[:30]}...")
+        
+        detail = get_feed_detail(feed_id, xsec_token, not args.no_comments)
+        if detail:
+            posts.append(detail)
+    
+    if not posts:
+        print("❌ 未能获取帖子详情")
+        sys.exit(1)
+    
+    print(f"✅ 成功获取 {len(posts)} 篇帖子详情")
+    
+    # 3. 生成报告
+    print("📝 生成报告...")
+    report = generate_report(args.keyword, posts)
+    
+    # 4. 输出
+    if args.output:
+        output_path = Path(args.output)
+        output_path.write_text(report, encoding="utf-8")
+        print(f"✅ 报告已保存: {output_path}")
+    
+    if args.feishu:
+        doc_title = f"小红书热点跟踪: {args.keyword} ({datetime.now().strftime('%m-%d')})"
+        export_to_feishu(doc_title, report)
+    
+    if not args.output and not args.feishu:
+        # 默认输出到 stdout
+        print("\n" + "="*60 + "\n")
+        print(report)
+    
+    return report
+
+
+if __name__ == "__main__":
+    main()
--- a/.agents/skills/xiaohongshu/scripts/track-topic.sh
+++ b/.agents/skills/xiaohongshu/scripts/track-topic.sh
@@ -0,0 +1,5 @@
+#!/bin/bash
+# 小红书热点跟踪工具
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+python3 "$SCRIPT_DIR/track-topic.py" "$@"
--- a/.agents/skills/xiaohongshu/scripts/user-profile.sh
+++ b/.agents/skills/xiaohongshu/scripts/user-profile.sh
@@ -0,0 +1,17 @@
+#!/bin/bash
+# 获取小红书用户主页
+
+USER_ID="$1"
+XSEC_TOKEN="$2"
+
+if [ -z "$USER_ID" ] || [ -z "$XSEC_TOKEN" ]; then
+    echo "用法: $0 <user_id> <xsec_token>"
+    echo ""
+    echo "user_id 和 xsec_token 可从搜索或推荐结果中获取"
+    exit 1
+fi
+
+SCRIPT_DIR="$(cd "$(dirname "$0")" && pwd)"
+ARGS=$(jq -n --arg uid "$USER_ID" --arg tok "$XSEC_TOKEN" \
+  '{"user_id":$uid,"xsec_token":$tok}')
+"$SCRIPT_DIR/mcp-call.sh" user_profile "$ARGS"
--- a/.agents/skills/xiaohongshu/tools/xhs-downloader/README.md
+++ b/.agents/skills/xiaohongshu/tools/xhs-downloader/README.md
@@ -0,0 +1,98 @@
+# XHS-Downloader 辅助工具
+
+配合 [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader) 使用的工具脚本，用于下载小红书收藏/点赞笔记并导出为 OpenClaw 记忆库格式。
+
+## 依赖
+
+需要先安装 XHS-Downloader：
+
+```bash
+git clone https://github.com/JoeanAmier/XHS-Downloader.git
+cd XHS-Downloader
+pip install -r requirements.txt
+```
+
+## 使用流程
+
+### 1. 获取收藏/点赞链接（使用油猴脚本）
+
+手动复制链接效率低，推荐使用 XHS-Downloader 提供的油猴脚本批量提取：
+
+**安装脚本：**
+
+1. 安装 [Tampermonkey](https://www.tampermonkey.net/) 浏览器扩展
+2. 安装用户脚本：[XHS-Downloader.js](https://raw.githubusercontent.com/JoeanAmier/XHS-Downloader/refs/heads/master/static/XHS-Downloader.js)
+
+**提取链接：**
+
+1. 打开 [小红书网页版](https://www.xiaohongshu.com) 并登录
+2. 进入个人主页 → **收藏** 或 **点赞** 页面
+3. 点击 Tampermonkey 图标，选择：
+   - `提取收藏作品链接`
+   - `提取点赞作品链接`
+4. 脚本会自动滚动页面加载全部内容
+5. 提取完成后链接自动复制到剪贴板
+
+**保存链接：**
+
+将剪贴板内容粘贴到 `links.md` 文件：
+
+```
+https://www.xiaohongshu.com/explore/xxx?xsec_token=...
+https://www.xiaohongshu.com/explore/yyy?xsec_token=...
+```
+
+> **注意**：自动滚动功能默认关闭，需在脚本设置中手动开启。开启后可能触发风控，建议适度使用。
+
+### 2. 批量下载
+
+```bash
+# 在 XHS-Downloader 目录下运行
+python batch_download.py links.md
+```
+
+下载的数据会保存到 `Volume/Download/ExploreData.db`。
+
+### 3. 导出记忆库
+
+**方式 A：导出为单文件**
+
+```bash
+python export_memory.py
+# 生成 xhs_memory.md
+```
+
+**方式 B：导出为多文件（推荐用于 OpenClaw）**
+
+```bash
+python export_to_workspace.py
+# 生成到 ~/.openclaw/workspace/xhs-memory/
+```
+
+### 4. 配置 OpenClaw 记忆搜索
+
+编辑 `~/.openclaw/openclaw.json`，添加：
+
+```json
+{
+  "memorySearch": {
+    "extraPaths": [
+      "~/.openclaw/workspace/xhs-memory"
+    ]
+  }
+}
+```
+
+或者如果使用单文件导出，将 `xhs_memory.md` 放到 workspace 目录下。
+
+## 脚本说明
+
+| 脚本 | 功能 |
+|------|------|
+| `batch_download.py` | 批量下载笔记并记录到数据库 |
+| `export_memory.py` | 导出为单个 Markdown 文件 |
+| `export_to_workspace.py` | 导出为多个独立文件（按日期+标题命名） |
+
+## 致谢
+
+- [XHS-Downloader](https://github.com/JoeanAmier/XHS-Downloader) - GPL-3.0 License
--- a/.agents/skills/xiaohongshu/tools/xhs-downloader/batch_download.py
+++ b/.agents/skills/xiaohongshu/tools/xhs-downloader/batch_download.py
@@ -0,0 +1,48 @@
+#!/usr/bin/env python
+"""
+批量下载小红书笔记
+
+用法:
+    python batch_download.py [links_file]
+
+默认读取当前目录的 links.md 文件
+"""
+import asyncio
+import sys
+from pathlib import Path
+
+try:
+    from source import XHS
+except ImportError:
+    print("错误: 请在 XHS-Downloader 项目目录下运行此脚本")
+    print("或安装依赖: pip install -e /path/to/XHS-Downloader")
+    sys.exit(1)
+
+
+async def main():
+    # 读取链接文件
+    links_file = Path(sys.argv[1]) if len(sys.argv) > 1 else Path("links.md")
+
+    if not links_file.exists():
+        print(f"错误: 链接文件不存在: {links_file}")
+        print("用法: python batch_download.py [links_file]")
+        sys.exit(1)
+
+    links = links_file.read_text().strip()
+    link_count = len([l for l in links.split() if l.startswith("http")])
+
+    print(f"开始下载，共 {link_count} 个链接...")
+
+    async with XHS(
+        work_path="./Volume",
+        folder_name="Download",
+        record_data=True,       # 记录作品数据到数据库
+        download_record=True,   # 跳过已下载
+        author_archive=True,    # 按作者分文件夹
+    ) as xhs:
+        result = await xhs.extract(links, download=True)
+        print(f"完成！处理了 {len(result)} 个作品")
+
+
+if __name__ == "__main__":
+    asyncio.run(main())
--- a/.agents/skills/xiaohongshu/tools/xhs-downloader/export_memory.py
+++ b/.agents/skills/xiaohongshu/tools/xhs-downloader/export_memory.py
@@ -0,0 +1,67 @@
+#!/usr/bin/env python
+"""
+从 XHS-Downloader 数据库导出笔记到单个 Markdown 文件
+
+用法:
+    python export_memory.py [db_path] [output_file]
+
+默认:
+    db_path: Volume/Download/ExploreData.db
+    output_file: xhs_memory.md
+"""
+import sqlite3
+import sys
+from pathlib import Path
+from datetime import datetime
+
+
+def export_memory(db_path: Path = None, output_file: Path = None):
+    db_path = db_path or Path("Volume/Download/ExploreData.db")
+    output_file = output_file or Path("xhs_memory.md")
+
+    if not db_path.exists():
+        print(f"错误: 数据库不存在: {db_path}")
+        return False
+
+    conn = sqlite3.connect(db_path)
+    cursor = conn.cursor()
+
+    # 查询所有作品
+    cursor.execute("""
+        SELECT 作品标题, 发布时间, 作品链接, 作品描述, 作者昵称, 作品标签
+        FROM explore_data
+        ORDER BY 发布时间 DESC
+    """)
+
+    rows = cursor.fetchall()
+    conn.close()
+
+    if not rows:
+        print("数据库为空")
+        return False
+
+    # 生成 Markdown
+    output = f"# 小红书收藏/点赞笔记 Memory\n\n"
+    output += f"> 导出时间: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}\n"
+    output += f"> 共 {len(rows)} 条笔记\n\n---\n\n"
+
+    for i, (title, time, link, desc, author, tags) in enumerate(rows, 1):
+        output += f"## {i}. {title or '无标题'}\n\n"
+        output += f"- **作者**: {author or '未知'}\n"
+        output += f"- **时间**: {time or '未知'}\n"
+        output += f"- **链接**: {link or '无'}\n"
+        if tags:
+            output += f"- **标签**: {tags}\n"
+        output += f"\n### 内容\n\n{desc or '无内容'}\n\n---\n\n"
+
+    # 保存文件
+    output_file.write_text(output, encoding="utf-8")
+    print(f"导出完成: {output_file.absolute()}")
+    print(f"共 {len(rows)} 条笔记")
+    return True
+
+
+if __name__ == "__main__":
+    db_path = Path(sys.argv[1]) if len(sys.argv) > 1 else None
+    output_file = Path(sys.argv[2]) if len(sys.argv) > 2 else None
+    export_memory(db_path, output_file)
--- a/.agents/skills/xiaohongshu/tools/xhs-downloader/export_to_workspace.py
+++ b/.agents/skills/xiaohongshu/tools/xhs-downloader/export_to_workspace.py
@@ -0,0 +1,101 @@
+#!/usr/bin/env python
+"""
+将小红书笔记按单独文件导出到 OpenClaw workspace
+
+用法:
+    python export_to_workspace.py [db_path] [output_dir]
+
+默认:
+    db_path: Volume/Download/ExploreData.db
+    output_dir: ~/.openclaw/workspace/xhs-memory
+
+导出格式类似 gpt-history，每条笔记一个文件，文件名格式: YYYY-MM-标题.md
+"""
+import sqlite3
+import re
+import sys
+from pathlib import Path
+
+
+def sanitize_filename(name: str, max_len: int = 50) -> str:
+    """清理文件名，移除非法字符"""
+    name = re.sub(r'[<>:"/\\|?*\n\r\t]', '', name)
+    name = re.sub(r'\s+', '-', name.strip())
+    name = re.sub(r'-+', '-', name)
+    name = name.strip('-')
+    if len(name) > max_len:
+        name = name[:max_len].rstrip('-')
+    return name or "无标题"
+
+
+def export_to_workspace(db_path: Path = None, output_dir: Path = None):
+    db_path = db_path or Path("Volume/Download/ExploreData.db")
+    output_dir = output_dir or Path.home() / ".openclaw/workspace/xhs-memory"
+    output_dir.mkdir(parents=True, exist_ok=True)
+
+    if not db_path.exists():
+        print(f"错误: 数据库不存在: {db_path}")
+        return False
+
+    conn = sqlite3.connect(db_path)
+    cursor = conn.cursor()
+
+    cursor.execute("""
+        SELECT 作品标题, 发布时间, 作品链接, 作品描述, 作者昵称, 作品标签
+        FROM explore_data
+        ORDER BY 发布时间 DESC
+    """)
+
+    rows = cursor.fetchall()
+    conn.close()
+
+    if not rows:
+        print("数据库为空")
+        return False
+
+    count = 0
+    for title, time, link, desc, author, tags in rows:
+        # 解析时间: 2026-01-25_18:17:43 -> 2026-01
+        if time:
+            date_prefix = time[:7]  # YYYY-MM
+            full_date = time.replace('_', ' ')
+        else:
+            date_prefix = "unknown"
+            full_date = "未知"
+
+        # 生成文件名
+        safe_title = sanitize_filename(title or "无标题")
+        filename = f"{date_prefix}-{safe_title}.md"
+        filepath = output_dir / filename
+
+        # 避免重复文件名
+        counter = 1
+        while filepath.exists():
+            filename = f"{date_prefix}-{safe_title}-{counter}.md"
+            filepath = output_dir / filename
+            counter += 1
+
+        # 生成内容
+        content = f"# {title or '无标题'}\n\n"
+        content += f"**来源**: 小红书收藏/点赞\n\n"
+        content += f"**日期**: {full_date}\n\n"
+        content += f"**作者**: {author or '未知'}\n\n"
+        content += f"**链接**: {link or '无'}\n\n"
+        if tags:
+            content += f"**标签**: {tags}\n\n"
+        content += "---\n\n"
+        content += "## 内容\n\n"
+        content += f"{desc or '无内容'}\n"
+
+        filepath.write_text(content, encoding="utf-8")
+        count += 1
+
+    print(f"导出完成: {output_dir}")
+    print(f"共生成 {count} 个文件")
+    return True
+
+
+if __name__ == "__main__":
+    db_path = Path(sys.argv[1]) if len(sys.argv) > 1 else None
+    output_dir = Path(sys.argv[2]) if len(sys.argv) > 2 else None
+    export_to_workspace(db_path, output_dir)