xiaohongshu-mcp

Xiaohongshu MCP Skill

Based on xpzouying/xiaohongshu-mcp built

🎯 Core Features

This Skill provides a complete Xiaohongshu automation solution:

Feature	Status	Description
🔐 Login Management	✅ Tested	Supports sending screenshots to Feishu
🔍 Search Content	✅ Tested	Keyword search, filtering
📄 Get Details	✅ Tested	Includes comment list
📤 Publish Image & Text	✅ Tested	Cover generator integration
👍 Like	✅ Tested	Like a single post
💬 Post Comment	✅ Tested	Main comment
↩️ Reply to Comment	✅ Tested	Sub-comment reply
⭐ Favorite	✅ Tested	Favorite/Unfavorite
🔄 Get Recommendations	✅ Tested	Homepage feeds

All 13 MCP tools are fully available!

🚀 Quick Start

1. Log in

# 方式1：一键登录（推荐）
bash xhs_login.sh --notify

# 方式2：本地登录
bash xhs_login.sh

2. Start the MCP server

./xiaohongshu-mcp-darwin-arm64 &

3. Use the features

# 检查登录状态
python3 scripts/xhs_client.py status

# 搜索内容
python3 scripts/xhs_client.py search "AI"

# 发布内容
python3 scripts/xhs_client.py publish "标题" "内容" "图片URL"

📚 Complete Operation Guide

MCP Tool List

Tool	Feature	Use Case
`check_login_status`	Check login status	Confirm account status
`list_feeds`	Get recommendation list	Discover trending content
`search_feeds`	Search content	Keyword search
`get_feed_detail`	Get post details	View comments
`publish_content`	Publish image & text	Create new content
`publish_with_video`	Publish video	Video content
`post_comment_to_feed`	Post comment	Reply to fans
`reply_comment_in_feed`	Reply to comment	Interaction
`like_feed`	Like	Like post
`favorite_feed`	Favorite	Favorite post
`delete_cookies`	Delete cookies	Reset login
`get_login_qrcode`	Get QR code	Log in again
`user_profile`	Get user profile	View profile

💬 Comment Interaction Strategy

Persona maintenance

Persona: rational thinker, not a contrarian

Comment section interaction requirements:

✅ Rational analysis, respect different opinions
✅ Rebuttals backed by data
✅ Natural conversational feel
❌ Attack commenters
❌ Argue unreasonably

Comment rules

Comment type	Like	Reply
Agreeing viewpoint	✅	✅ Continuity
Partial agreement	❌	✅ Add viewpoint
Opposing viewpoint	❌	✅ Respectful expression
Question	✅	✅ Direct answer
Share experience	✅	✅ Resonance

Reply Templates

Agreement type:

"说出了我想说的！[补充细节]"
"对对对，尤其是[具体例子]..."

Partial agreement type:

"有道理，不过我觉得[补充观点]"
"同意一半吧，另外[补充视角]"

Opposite opinion type:

"你的观点挺有意思，不过我觉得[不同看法]"
"可能我表达不清楚，我想说的是[重新解释]"

Question type:

"好问题！我的看法是[直接回答]"
"这个要分情况，[分情况说明]"

Experience sharing type:

"太真实了！[共鸣]"
"你这个经历太有代表性了！[延伸]"

Reply Requirements

Reply to every comment - Show activity
Topic relevance - Stick to the core of the post
Continuity - Not perfunctory
No AI feel - Natural and colloquial
Maintain persona - Rational thinker

🔧 Technical Implementation

MCP HTTP API

All features can be called via HTTP API:

# MCP Endpoint
http://localhost:18060/mcp

# 格式
curl -X POST http://localhost:18060/mcp \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: <SESSION_ID>" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "search_feeds",
      "arguments": {
        "keyword": "AI"
      }
    }
  }'

MCP Session Retrieval

# 初始化
RESPONSE=$(curl -s -i -X POST http://localhost:18060/mcp \
  -H "Content-Type: application/json" \
  -c cookies.txt \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}}}')

# 提取 Session ID
SESSION_ID=$(echo "$RESPONSE" | grep -i "Mcp-Session-Id:" | cut -d' ' -f2)

Example: Complete Publishing Process

#!/bin/bash
MCP_URL="http://localhost:18060/mcp"
COOKIE_FILE="cookies.txt"

# 1. 初始化
RESPONSE=$(curl -s -i -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -c "$COOKIE_FILE" \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"test","version":"1.0"}}}')

SESSION_ID=$(echo "$RESPONSE" | grep -i "Mcp-Session-Id:" | cut -d' ' -f2)

# 2. 发送初始化通知
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{"jsonrpc":"2.0","method":"notifications/initialized","params":{}}' > /dev/null

# 3. 发布内容
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 100,
    "method": "tools/call",
    "params": {
      "name": "publish_content",
      "arguments": {
        "title": "AI正在毁掉这一代年轻人？",
        "content": "🔥 争议话题...\n\n详细内容...",
        "images": ["/tmp/cover.jpg"]
      }
    }
  }'

Example: Comment Interaction

#!/bin/bash
MCP_URL="http://localhost:18060/mcp"
COOKIE_FILE="cookies.txt"
SESSION_ID="YOUR_SESSION_ID"

# 1. 获取评论列表
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 1,
    "method": "tools/call",
    "params": {
      "name": "get_feed_detail",
      "arguments": {
        "feed_id": "698c441c000000002801d381",
        "xsec_token": "YOUR_TOKEN",
        "load_all_comments": true
      }
    }
  }'

# 2. 点赞
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 2,
    "method": "tools/call",
    "params": {
      "name": "like_feed",
      "arguments": {
        "feed_id": "698c441c000000002801d381",
        "xsec_token": "YOUR_TOKEN"
      }
    }
  }'

# 3. 发表评论
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 3,
    "method": "tools/call",
    "params": {
      "name": "post_comment_to_feed",
      "arguments": {
        "feed_id": "698c441c000000002801d381",
        "xsec_token": "YOUR_TOKEN",
        "content": "说出了我想说的！补充细节..."
      }
    }
  }'

# 4. 回复评论
curl -s -X POST "$MCP_URL" \
  -H "Content-Type: application/json" \
  -H "Mcp-Session-Id: $SESSION_ID" \
  -d '{
    "jsonrpc": "2.0",
    "id": 4,
    "method": "tools/call",
    "params": {
      "name": "reply_comment_in_feed",
      "arguments": {
        "feed_id": "68786933000000000d01a693",
        "xsec_token": "YOUR_TOKEN",
        "comment_id": "68786afc000000001101ada6",
        "user_id": "6695e7370000000003032a17",
        "content": "说得有道理！补充观点..."
      }
    }
  }'

🛠️ Script Tools

xhs_client.py - Python Client

# 检查状态
python3 scripts/xhs_client.py status

# 搜索
python3 scripts/xhs_client.py search "AI" --sort "最新" --type "图文" --time "一周内"

# 发布
python3 scripts/xhs_client.py publish "标题" "内容" "图片URL" --tags "标签1,标签2"

# 获取详情
python3 scripts/xhs_client.py detail <feed_id> <xsec_token> --comments

xhs_mcp.py - Direct MCP Call

# 列出所有工具
python3 scripts/xhs_mcp.py tools

# 发表评论
python3 scripts/xhs_mcp.py comment <feed_id> <xsec_token> "评论内容"

generate_cover.py - Cover Generator

# 生成封面
python3 generate_cover.py --title "标题" --output /tmp/cover.jpg

# 选项
--font-size 80      # 字体大小
--padding 60         # 内边距
--max-width 600      # 最大宽度

📁 File Structure

xiaohongshu-mcp-skill/
├── SKILL.md              # 本文档
├── README.md             # 中文文档
├── STRATEGY.md          # 运营策略（含评论互动）
├── install.sh            # 安装脚本
├── xhs_login.sh         # 一键登录
├── generate_cover.py     # 封面生成器
├── data/
│   ├── post_history.json   # 发布记录
│   ├── hot_topics.json    # 热点选题
│   └── cookies.json       # 登录 cookies
└── scripts/
    ├── xhs_client.py     # Python 客户端
    ├── xhs_mcp.py       # MCP 直接调用
    ├── xhs_login_sop.py  # 登录 SOP
    └── publish_smart.py   # 智能发布脚本

📊 Test Records

Tested Features ✅

Feature	Status	Test Time	Notes
Publish Image and Text	✅	2026-02-11	2 posts published
Search Content	✅	2026-02-11	22 results
Get Details	✅	2026-02-11	Includes comment list
Post Comment	✅	2026-02-11	6 comments
Like	✅	2026-02-11	API successful
Favorite	✅	2026-02-11	Functioning normally
Reply to Comment	✅	2026-02-11	API successful

Test Posts

"Art academy students are all using AI? I just laughed"
- Feed ID: 698c441c000000002801d381
- Likes: 2, Comments: 6
"Is AI destroying this generation of young people?"
- Feed ID: 698c76f8000000001a024a93
- Likes: 1, Comments: 0

🔗 Related Resources

GitHub: https://github.com/tclawde/xiaohongshu-mcp-skill
MCP Server: xpzouying/xiaohongshu-mcp
OpenClaw: https://github.com/openclaw/openclaw

📝 Changelog

v3.0 (2026-02-11)

✅ Added complete comment interaction strategy
✅ Added MCP HTTP API call examples
✅ Added complete list of 13 tools
✅ Added script tool usage instructions
✅ Added technical implementation details
✅ Added test records

v2.0 (2026-02-11)

✅ Login fix (supports Xiaohongshu page changes)
✅ Feishu notification integration
✅ Python client improvements

v1.0 (2026-02-11)

✅ Initial version
✅ Basic publishing features
✅ Search feature

Maintainer: TClawDE 🦀 Last updated: 2026-02-11