zhihu-mcp
Provides tools to interact with Zhihu, enabling AI agents to publish articles and videos, search content, get recommendations and feed details, post and reply to comments, manage login sessions, and retrieve user profiles.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@zhihu-mcp搜索知乎上关于'机器学习'的内容"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
zhihu-mcp
知乎 MCP 服务 — 让你的 AI 助手直接操作知乎。
基于 xiaohongshu-mcp 的设计思路,使用 Python + Playwright + FastMCP 实现。
功能总览
# | MCP 工具 | 说明 |
1 |
| 检查知乎登录状态 |
2 |
| 获取登录二维码 |
3 |
| 删除 cookies,重置登录状态 |
4 |
| 发布图文文章 |
5 |
| 发布视频内容 |
6 |
| 搜索知乎内容(结果自动保存为 md 文件) |
7 |
| 获取首页推荐列表 |
8 |
| 获取帖子详情(含正文和评论,自动保存为 md 文件) |
9 |
| 发表评论到帖子 |
10 |
| 获取用户个人主页 |
11 |
| 回复评论 |
Related MCP server: Xiaohongshu (RedBook) MCP Server
快速开始
1. 安装依赖
pip install -r requirements.txt安装 Playwright 浏览器:
python -m playwright install chromium2. 登录知乎
首次使用需要扫码登录,会弹出浏览器窗口:
python login.py登录成功后 cookies 自动保存到 cookies/cookies.json。
3. 启动 MCP 服务
# 默认:端口 18060,显示浏览器
python main.py
# 无头模式(不显示浏览器)
python main.py --headless
# 自定义端口
python main.py --port 8080MCP 服务运行在:http://localhost:18060/mcp
4. 配置代理(可选)
set ZHIHU_PROXY=http://user:pass@proxy:port # Windows
export ZHIHU_PROXY=http://user:pass@proxy:port # Linux/Mac
python main.pyMCP 客户端接入
Claude Code
claude mcp add --transport http zhihu-mcp http://localhost:18060/mcpOpen Code
opencode mcp add
# 名称: zhihu-mcp
# 类型: Remote
# URL: http://localhost:18060/mcpCursor
在项目根目录创建 .cursor/mcp.json:
{
"mcpServers": {
"zhihu-mcp": {
"url": "http://localhost:18060/mcp"
}
}
}VS Code
在项目根目录创建 .vscode/mcp.json:
{
"servers": {
"zhihu-mcp": {
"url": "http://localhost:18060/mcp",
"type": "http"
}
}
}MCP Inspector(调试用)
npx @modelcontextprotocol/inspector
# 连接地址填 http://localhost:18060/mcpOpenCode 技能(Skills)
项目包含两个 OpenCode 技能文件,安装后可让 AI 自动按流程操作知乎:
安装技能
将技能文件复制到 OpenCode 技能目录:
# Windows
copy skills\zhihu-auth\SKILL.md %USERPROFILE%\.agents\skills\zhihu-auth\SKILL.md
copy skills\zhihu-publish\SKILL.md %USERPROFILE%\.agents\skills\zhihu-publish\SKILL.md
# Linux / Mac
mkdir -p ~/.agents/skills/zhihu-auth ~/.agents/skills/zhihu-publish
cp skills/zhihu-auth/SKILL.md ~/.agents/skills/zhihu-auth/SKILL.md
cp skills/zhihu-publish/SKILL.md ~/.agents/skills/zhihu-publish/SKILL.md安装后重启 OpenCode 生效。
技能说明
技能 | 用途 | 触发词 |
| 登录状态检查、扫码登录、退出登录 | "登录知乎"、"检查登录"、"退出登录" |
| 文章发布、视频发布、搜索、详情、评论 | "发布到知乎"、"搜索知乎"、"查看详情" |
文件输出
搜索和获取详情的结果会自动保存到 search_results/ 目录:
search_results/
├── ai岗位_20260328_173948.md # 搜索结果列表
├── 某篇文章标题_20260328_174020.md # 帖子详情(含评论)
└── ...项目结构
zhihu-mcp/
├── browser/
│ ├── __init__.py
│ └── manager.py # Playwright 浏览器管理 + 反检测
├── zhihu/
│ ├── __init__.py
│ ├── actions.py # 知乎浏览器自动化操作
│ ├── cookies.py # Cookie 持久化
│ └── service.py # 服务层(桥接 actions 与 MCP)
├── server/
│ ├── __init__.py
│ └── tools.py # MCP 工具定义与注册
├── skills/
│ ├── zhihu-auth/
│ │ └── SKILL.md # 认证管理技能(登录/退出)
│ └── zhihu-publish/
│ └── SKILL.md # 内容发布技能(发布/搜索/评论)
├── cookies/ # cookies 存储目录
├── search_results/ # 搜索结果输出目录
├── main.py # MCP 服务入口
├── login.py # 登录工具
├── requirements.txt # Python 依赖
├── pyproject.toml # 项目配置
├── README.md
└── .gitignore使用示例
在 AI 助手中直接使用自然语言:
帮我搜索知乎上关于 "春招" 的内容查看第 3 条的详情获取知乎首页推荐帮我发布一篇知乎文章,标题是 "AI 学习笔记",内容是(标题和内容是必要字段,注意:知乎发布内容要求字数大于9个字。) ...对这条帖子发表评论注意事项
登录:知乎同一账号不能在多处网页端同时登录,会互相踢下线
Cookie 过期:如果操作提示未登录,重新运行
python login.py扫码频率控制:建议每天发布不超过 5-10 篇,避免触发风控
发布按钮:知乎编辑器的"发布设置"和"发布"是两个不同的按钮,程序会精确匹配"发布"
URL:知乎文章编辑器地址为
https://zhuanlan.zhihu.com/write,不是https://www.zhihu.com/write
风险提示
该项目仅供学习目的,禁止用于违法行为
自动化操作可能触发知乎的风控机制,建议合理控制频率
请妥善保管
cookies/cookies.json文件,不要泄露给他人
This server cannot be installed
Maintenance
Resources
Unclaimed servers have limited discoverability.
Looking for Admin?
If you are the server author, to access and configure the admin panel.
Latest Blog Posts
- Your AI Chatbot Just Exposed Your CEO's Salary to an InternBy Om-Shree-0709 on .Agent IdentityMCP SecurityOAuth Delegation
- Why MCP Servers Need Execution Sandboxing (And Why Your Current Stack Isn't Enough)By Om-Shree-0709 on .Agentic AiPrompt InjectionWebAssembly
MCP directory API
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/Douyh123/zhihu-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server