Docmost MCP Server
Provides AI assistants with direct access to self-hosted Docmost documentation through tools for listing spaces, searching content, and retrieving pages in Markdown format. It facilitates seamless documentation lookup and content retrieval within MCP-compatible clients.
README
Docmost MCP Server
An MCP server that gives AI assistants direct access to your self-hosted Docmost documentation via its API.
Works with any MCP-compatible client, including:
Features
| Tool | Description |
|---|---|
| list_spaces | List all available documentation spaces with names, slugs, and IDs |
| search_docs | Full-text search across all documentation, with optional space filtering |
| get_page | Retrieve full page content converted from ProseMirror JSON to Markdown |
Prerequisites
- Python 3.10+
- A running Docmost instance with valid user credentials
- An MCP-compatible client (Claude Code, Claude Desktop, Cursor, VS Code, etc.)
Installation
1. Clone the repository
git clone https://github.com/aleksvin8888/local-docmost-mcp.git
cd docmost-mcp
2. Create a virtual environment and install dependencies
python3 -m venv venv
source venv/bin/activate # Linux / macOS
# venv\Scripts\activate # Windows
pip install -r requirements.txt
This installs two dependencies:
3. Configure credentials
Copy the example config and fill in your details:
cp config.example.json config.json
Edit config.json:
{
"base_url": "https://your-docmost-instance.example.com",
"email": "your-email@example.com",
"password": "your-password",
"timeout": 30
}
| Parameter | Description |
|---|---|
base_url |
URL of your Docmost instance (no trailing slash) |
email |
Email address for Docmost authentication |
password |
Password for Docmost authentication |
timeout |
HTTP request timeout in seconds |
Note:
config.jsoncontains sensitive credentials and is excluded from version control via.gitignore.
4. Verify the setup
source venv/bin/activate # if not already active
python docmost_client.py
Expected output:
=== Spaces ===
My Space (my-space) - 019a2a69-...
Another Space (another) - 019a5e21-...
...
=== Search 'example' ===
Example Page Title - abc123def
...
If you see your spaces and search results, the client is working correctly.
Connecting to an MCP client
The server uses stdio transport, which is supported by all major MCP clients. Below are setup instructions for the most popular ones.
Claude Code
Choose one of the following options.
Important: Use absolute paths to both the Python binary inside
venvandmcp_server.py.
Option 1: Global config (recommended)
Add to ~/.claude/settings.json:
{
"mcpServers": {
"docmost": {
"command": "/absolute/path/to/docmost-mcp/venv/bin/python",
"args": ["/absolute/path/to/docmost-mcp/mcp_server.py"]
}
}
}
Option 2: Project-level config
Create .claude/settings.json in your project root:
{
"mcpServers": {
"docmost": {
"command": "/absolute/path/to/docmost-mcp/venv/bin/python",
"args": ["/absolute/path/to/docmost-mcp/mcp_server.py"]
}
}
}
Option 3: CLI command
claude mcp add docmost \
-c "/absolute/path/to/docmost-mcp/venv/bin/python" \
-- /absolute/path/to/docmost-mcp/mcp_server.py
After adding the config, restart Claude Code or start a new session.
Claude Desktop
Add the server to your Claude Desktop config file:
- Linux:
~/.config/claude/claude_desktop_config.json - macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
{
"mcpServers": {
"docmost": {
"command": "/absolute/path/to/docmost-mcp/venv/bin/python",
"args": ["/absolute/path/to/docmost-mcp/mcp_server.py"]
}
}
}
Restart Claude Desktop after saving the config.
Cursor / Windsurf / VS Code / Other clients
Most MCP clients use the same configuration format. Add the server with:
- Command:
/absolute/path/to/docmost-mcp/venv/bin/python - Args:
/absolute/path/to/docmost-mcp/mcp_server.py - Transport: stdio
Refer to your client's documentation for the exact config file location and format.
Usage
Once the MCP server is connected, your AI assistant can use the following tools:
List all spaces
Show me all available documentation spaces
Search documentation
Search the documentation for "user permissions"
Search within a specific space
Find "API endpoints" in the Engineering space
Get a specific page
Show me the full content of page with slug_id "abc123def"
How it works
- The MCP client launches the server as a subprocess and communicates via stdio (stdin/stdout).
- On the first API call, the server authenticates with your Docmost instance using email/password and receives a JWT token from the
Set-Cookieheader. - The token is cached in
token.jsonfor subsequent requests. If a request returns 401, the server automatically re-authenticates and retries. - Page content is stored in Docmost as ProseMirror JSON. The server converts it to clean Markdown (headings, lists, tables, code blocks, images, links, etc.) before returning it to the client.
Project structure
docmost-mcp/
├── mcp_server.py # MCP server (3 tools)
├── docmost_client.py # Docmost API client
├── requirements.txt # Python dependencies (mcp, requests)
├── config.example.json # Configuration template
├── config.json # Your credentials (not in git)
├── token.json # Cached JWT token (auto-generated, not in git)
├── .gitignore
└── README.md
Security
config.jsoncontains your credentials — never commit it to gittoken.jsoncontains a JWT token — never commit it to git- Both files are already listed in
.gitignore
Troubleshooting
"No authToken in response cookies"
- Verify that
emailandpasswordinconfig.jsonare correct - Verify that
base_urlpoints to your Docmost instance
"Connection refused"
- Check that your Docmost instance is running and accessible
- Check your network connection and firewall rules
"401 Unauthorized"
- The token may have expired — delete
token.jsonand it will be recreated automatically on the next request
MCP tools not appearing in the client
- Ensure you used absolute paths to
venv/bin/pythonandmcp_server.py - Check that the virtual environment has all dependencies installed (
pip install -r requirements.txt) - Restart your MCP client after changing the config
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。