Wikijs MCP 服务器 | PlugMCP
跳到正文 Wiki.js MCP Server
Model Context Protocol (MCP) server for Wiki.js integration via GraphQL API.
📖 Description
This project provides an MCP server for interacting with Wiki.js through GraphQL API. MCP (Model Context Protocol) is an open protocol developed by Anthropic that enables AI models to safely interact with external services and tools.
Model Context Protocol (MCP) server for Wiki.js GraphQL API integration. Provides AI models with secure access to Wiki.js functionality through MCP protocol.
把任意工具接入任意智能体。面向非开发者的 MCP 服务器目录;网关承载流量,n8n 承载应用。
产品
目录
开发者
公司
The server provides a unified interface for working with Wiki.js that can be used by various AI agents and tools supporting MCP.
✨ Features
📄 Page Management
Get Wiki.js pages by ID
Get page content by ID
List pages with sorting options
Search pages by query
Create new pages
Update existing pages
Delete pages
🆕 List all pages including unpublished
🆕 Search unpublished pages
🆕 Force delete pages (including unpublished)
🆕 Get page publication status
🆕 Publish unpublished pages
👥 User Management
List users
Search users by query
Create new users
Update user information
🔧 Group Management
🌐 Transports
STDIO : for editor integration (Cursor, VS Code)
HTTP : for web integrations and API access
🚀 Quick Start
Installation git clone https://github.com/heAdz0r/wikijs-mcp-server.git
cd wikijs-mcp-server
This script will automatically:
Install dependencies
Create .env file based on example.env
Build TypeScript code
Configuration
Edit the .env file and specify your Wiki.js settings:
# Port for HTTP MCP server
PORT=3200
# Base URL for Wiki.js (without /graphql)
WIKIJS_BASE_URL=http://localhost:3000
# Wiki.js API token
WIKIJS_TOKEN=your_wikijs_api_token_here
Edit the .cursor/mcp.json file and replace your_wikijs_api_token_here with your real token
How to get Wiki.js API token:
Log into Wiki.js admin panel
Go to "API" section
Create a new API key with necessary permissions
Copy the token to .env AND to .cursor/mcp.json
📦 Running
HTTP server (recommended) # Start main HTTP server with Cursor MCP support
npm start
# or
npm run start:http
# Stop server
npm run stop
TypeScript version
STDIO mode (for direct editor integration)
Development mode
Testing
🔌 Editor Integration
Cursor IDE
⚠️ IMPORTANT: Without .cursor/mcp.json file, Cursor integration will NOT work!
Quick Setup
Automatic configuration setup:
Edit .cursor/mcp.json and specify your real token:
{
"mcpServers": {
"wikijs": {
"transport": "http",
"url": "http://localhost:3200/mcp",
"events": "http://localhost:3200/mcp/events",
"cwd": ".",
"env": {
"WIKIJS_BASE_URL": "http://localhost:3000",
"WIKIJS_TOKEN": "your_real_wiki_js_token_here"
}
}
}
}
Critical Parameters
transport: "http" - mandatory HTTP transport
url: "http://localhost:3200/mcp" - exact URL for JSON-RPC
events: "http://localhost:3200/mcp/events" - URL for Server-Sent Events
WIKIJS_TOKEN - real Wiki.js API token (not placeholder!)
Verification After setup, tools with mcp_wikijs_* prefix should appear in Cursor:
mcp_wikijs_list_pages()
mcp_wikijs_search_pages()
mcp_wikijs_get_page()
And others...
VS Code (with MCP extension) {
"mcp.servers": {
"wikijs": {
"command": "node",
"args": ["lib/mcp_wikijs_stdin.js"],
"cwd": "/path/to/wikijs-mcp"
}
}
}
🛠 Development
Project Structure wikijs-mcp-server/
├── src/ # TypeScript source code
│ ├── server.ts # HTTP server
│ ├── tools.ts # Tool definitions
│ ├── api.ts # Wiki.js API client
│ ├── types.ts # Data types
│ ├── schemas.ts # Zod validation schemas
│ └── README.md # Source code documentation
├── lib/ # JavaScript library files
│ ├── fixed_mcp_http_server.js # Main HTTP server (compiled)
│ ├── mcp_wikijs_stdin.js # STDIN server for editors
│ ├── mcp_client.js # Demo MCP client
│ ├── mcp_wrapper.js # MCP protocol utilities
│ └── README.md # Library documentation
├── scripts/ # Management scripts
│ ├── setup.sh # Initial setup
│ ├── start_http.sh # Start HTTP server
│ ├── stop_server.sh # Stop HTTP server
│ ├── start_typescript.sh # Start TypeScript version
│ ├── setup_cursor_mcp.sh # Cursor setup
│ ├── test.sh # Run tests
│ ├── test_mcp.js # Test HTTP server
│ ├── test_mcp_stdin.js # Test STDIN server
│ └── README.md # Scripts documentation
├── .cursor/ # Cursor MCP configuration
│ └── mcp.json # MCP configuration file (CRITICALLY IMPORTANT!)
├── dist/ # Compiled TypeScript code
├── package.json # Project metadata
└── README.md # Main documentation
🚨 CRITICALLY IMPORTANT: .cursor/mcp.json file is required for Cursor integration!
Available Scripts
Setup and Build
npm run setup - Initial project setup
npm run build - Build TypeScript project
npm run setup:cursor - Setup Cursor integration
Running Servers
npm start / npm run start:http - HTTP MCP server (port 3200)
npm run stop - Stop all MCP servers
npm run start:typescript - TypeScript version of server (port 8000)
npm run server:stdio - STDIO version for direct integration
Development and Testing
npm run dev - Development mode with hot reload
npm run demo - Capability demonstration
npm test - Run tests
npm run client - Run demo client
API Endpoints (HTTP mode)
GET /tools - List of available tools
GET /health - Server health check
POST /mcp - MCP JSON-RPC endpoint
Usage Examples // Get list of pages
{
"method": "list_pages",
"params": {
"limit": 10,
"orderBy": "TITLE"
}
}
// Create new page
{
"method": "create_page",
"params": {
"title": "New Page",
"content": "# Title\n\nContent...",
"path": "folder/new-page"
}
}
### Search for pages:
```python
# Search in all content and metadata
result = await mcp_client.call_tool("search_pages", {
"query": "magic system",
"limit": 5
})
Working with Unpublished Pages: # Get all pages including unpublished ones
all_pages = await mcp_client.call_tool("list_all_pages", {
"limit": 100,
"includeUnpublished": True
})
# Search only unpublished pages
unpublished = await mcp_client.call_tool("search_unpublished_pages", {
"query": "draft",
"limit": 10
})
# Check page publication status
status = await mcp_client.call_tool("get_page_status", {
"id": 42
})
# Publish an unpublished page
result = await mcp_client.call_tool("publish_page", {
"id": 42
})
# Force delete page (works with unpublished pages)
result = await mcp_client.call_tool("force_delete_page", {
"id": 42
})
User management: # List all users
users = await mcp_client.call_tool("list_users")
# Search users by query
search_result = await mcp_client.call_tool("search_users", {
"query": "John"
})
# Create new user
new_user = await mcp_client.call_tool("create_user", {
"email": "[email protected] ",
"name": "John Doe",
"passwordRaw": "password123",
"providerKey": "local",
"groups": [1],
"mustChangePassword": false,
"sendWelcomeEmail": true
})
# Update user information
updated_user = await mcp_client.call_tool("update_user", {
"id": 1,
"name": "John Doe Updated"
})
🐛 Troubleshooting
Connection Issues
Ensure Wiki.js is running and accessible
Check WIKIJS_BASE_URL correctness
Verify API token is valid
MCP Issues
Check Node.js version (requires >=18.0.0)
Ensure all dependencies are installed
Check server logs for errors
📚 Documentation
🤝 Contributing
Fork the repository
Create a feature branch (git checkout -b feature/amazing-feature)
Commit your changes (git commit -m 'Add amazing feature')
Push to the branch (git push origin feature/amazing-feature)
Open a Pull Request
📄 License This project is distributed under the MIT License. See LICENSE file for details.
🔗 Useful Links
⭐ Support If this project helped you, please give it a ⭐ on GitHub!
Have questions? Create an Issue or refer to the documentation.
🆕 New Feature: Automatic URLs
Search Stages Search works in 4 stages:
GraphQL API search - fast search through indexed content
Metadata search - search in titles, paths, and page descriptions
HTTP content search - deep search in page content via HTTP
Forced verification - fallback search on known pages
Usage Examples
Content Search {
"method": "search_pages",
"params": {
"query": "ZELEBOBA",
"limit": 5
}
}
[
{
"id": 103,
"path": "test/test-page",
"title": "Test Page",
"description": "Test page to demonstrate Wiki.js API capabilities",
"url": "http://localhost:8080/en/test/test-page"
}
]
Title Search {
"method": "search_pages",
"params": {
"query": "find me",
"limit": 3
}
}
[
{
"id": 108,
"path": "test/test-gemini-mcp",
"title": "Test Gemini MCP Page (find me)",
"url": "http://localhost:8080/en/test/test-gemini-mcp"
}
]
New Search Benefits
✅ Finds pages even with limited API permissions - uses HTTP fallback
✅ Multi-level search - combines multiple strategies
✅ Content search - finds text inside pages
✅ Metadata search - titles, paths, descriptions
✅ Fallback methods - guaranteed results for known pages
✅ Correct URLs - all results contain ready-to-use links
Technical Details
HTML Content Processing The system automatically extracts text from HTML using:
Search in <template slot="contents"> block
HTML tags and entities cleanup
Fallback to full page content
With limited GraphQL API permissions, the system:
Switches to HTTP method for content retrieval
Uses direct requests to HTML pages
Preserves all page metadata
📝 Changelog
Version 1.3.0 - Unpublished Pages Management (Latest)
🆕 New Features:
list_all_pages - Get all pages including unpublished ones
search_unpublished_pages - Search specifically in unpublished pages
force_delete_page - Enhanced deletion that works with unpublished pages
get_page_status - Check publication status of any page
publish_page - Publish unpublished pages programmatically
🔧 Improvements:
Enhanced server API with new routes for unpublished page management
Better error handling for page deletion operations
Comprehensive GraphQL mutation support for advanced page operations
Restructured project : Moved JavaScript files to lib/ directory for better organization
🐛 Bug Fixes:
Fixed issues with accessing unpublished pages through standard APIs
Improved authentication handling for admin-level operations
Version 1.2.0 - International Release
🌍 Internationalization:
Complete English translation of documentation
README.md and QUICK_START.md now available in English
Prepared for international market expansion
Version 1.1.0 - Enhanced Search & User Management
✨ Features:
Smart multi-method page search (GraphQL + content + metadata)
User management tools (create, update, search)
Group management capabilities
Improved content extraction from HTML pages
🛠️ Available Tools
📄 Page Tools
👥 User Tools
🔗 Group Tools WeKnora Open-source LLM knowledge platform: turn raw documents into a queryable RAG, an autonomous reasoning agent, and a self-maintaining Wiki. OpenMetadata The Open Context Layer for Data and AI , OpenMetadata is the open platform for building trusted data context and business semantics for humans, AI assistants, and agents. Quelvio **Your company's brain for AI agents.**
Quelvio continuously understands your organization's knowledge across every connected system — what's authoritative, what's recent, what's persistent, what's t Hyperresearch Convert Claude Code or Codex into the most intelligent Deep Research Agent. Collect, search, and synthesize web research into a persistent, searchable wiki that builds on itself. Hosted API + MCP: hyp Httprunner HttpRunner 是一款开源的 API/UI 测试框架,简单易用,功能强大,具有丰富的插件化机制和高度的可扩展能力。 Solon 🔥 Java enterprise application development framework for full scenario: Restrained, Efficient, Open, Ecologicalll!!! 700% higher concurrency 50% memory savings Startup is 10 times faster. Packing 90% Crosspaste Desktop Cross-device clipboard sync for macOS, Windows & Linux — end-to-end encrypted, LAN-only, no cloud. OCR, CLI and MCP server built in. Llm Wiki Compiler The knowledge compiler. Raw sources in, interlinked wiki out. Inspired by Karpathy's LLM Wiki pattern. Llmwiki Compile documents into a cited knowledge wiki. Retrieve evidence, query, and review via local MCP. GitLab MCP Server GitLab MCP server for projects, merge requests, issues, pipelines, wiki, releases, and more. Llmwiki Open Source Implementation of Karpathy's LLM Wiki. Upload documents, connect your Claude account via MCP, and have it write your wiki ! Arkon Arkon: Enterprise AI Knowledge Hub & MCP Server. Self-hosted knowledge base for teams to manage RAG contexts, access policies, and AI skills. Connect Claude and other LLMs via Model Context Protocol ( Collaboration MCP 服务器 最佳 Collaboration MCP 服务器
暂无工具列表。远程服务器在首次健康检查后会获得工具列表。
如何安装 Wikijs? 在本页复制适用于 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 配置。它使用发布者的安装方式,而不是 PlugMCP 代理。
Wikijs 需要 API 密钥吗? 此记录未列出必需的环境变量。
Wikijs 使用哪种传输方式? Wikijs 通过 stdio 在本地运行。
Wikijs 上次检查是什么时候? 上次检查时间:2026-10-28。 Model Context Protocol (MCP) server for Wiki.js GraphQL API integration. Provides AI models with secure access to Wiki.js functionality through MCP protocol.
工具
0 / 10