docmost-mcp
Gives AI agents read/write access to a self-hosted Docmost wiki through its REST API.
README
docmost-mcp
An MCP server that gives AI agents read/write access to a self-hosted Docmost wiki through its regular REST API.
Why this exists
Docmost ships its own MCP endpoint, but it is gated behind a paid licence — the
Settings UI shows "Available with a paid license" for API management, and
/api/mcp returns 404 on the Community Edition. Minting API keys is gated the
same way.
The ordinary REST API, however, is fully open in the Community Edition. This server is a thin wrapper over it: same operations, session authentication instead of an API key.
Requirements
- Python 3.11+
- A Docmost instance you can reach over HTTP(S)
- A dedicated Docmost user account for the agent
Install
git clone https://github.com/<you>/docmost-mcp.git
cd docmost-mcp
python3 -m venv venv
./venv/bin/pip install -r requirements.txt
Configure
Copy the example config and fill it in:
cp config.example.json config.json
chmod 600 config.json
{
"url": "https://docmost.example.com",
"api_key": "",
"email": "agent@example.com",
"password": "..."
}
Two authentication modes are supported:
| Mode | When to use |
|---|---|
api_key |
If you hold a Docmost Enterprise licence. Sent as a Bearer token. |
email + password |
Community Edition. The server logs in and re-authenticates automatically when the session expires. |
If api_key is set it takes precedence; otherwise the credentials are used.
The config path can be overridden with the DOCMOST_MCP_CONFIG environment
variable.
Create a dedicated account
Do not use the workspace owner's credentials. Invite a separate user (Settings → Members → Invite) and grant it access only to the spaces the agent needs. Space access in Docmost is usually inherited from the default Everyone group, so check what that group can reach before assuming the agent is restricted.
Gmail-style plus-addressing (you+agent@gmail.com) works if you would rather
not create a second mailbox.
Verify
selftest.py exercises the whole chain — login, read, write, read back, delete:
./venv/bin/python selftest.py
Connecting an agent
The server speaks MCP over stdio.
Claude Code
claude mcp add docmost -- /path/to/docmost-mcp/venv/bin/python /path/to/docmost-mcp/server.py
Or add it to ~/.claude.json by hand:
{
"mcpServers": {
"docmost": {
"type": "stdio",
"command": "/path/to/docmost-mcp/venv/bin/python",
"args": ["/path/to/docmost-mcp/server.py"],
"env": {}
}
}
}
Restart Claude Code afterwards — the configuration is read at startup.
Claude Desktop
Add the same block to claude_desktop_config.json:
- macOS:
~/Library/Application Support/Claude/claude_desktop_config.json - Windows:
%APPDATA%\Claude\claude_desktop_config.json
Cursor
Add it to .cursor/mcp.json in the project, or to ~/.cursor/mcp.json
globally, using the same mcpServers shape.
Any other MCP client
Launch server.py with the virtualenv's Python and speak JSON-RPC over stdin
and stdout. A minimal handshake:
{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2025-06-18","capabilities":{},"clientInfo":{"name":"probe","version":"1"}}}
{"jsonrpc":"2.0","method":"notifications/initialized"}
{"jsonrpc":"2.0","id":2,"method":"tools/list"}
A convenience wrapper keeps the command short:
#!/bin/sh
exec /path/to/docmost-mcp/venv/bin/python /path/to/docmost-mcp/server.py "$@"
Tools
| Tool | Arguments | Does |
|---|---|---|
list_spaces |
limit |
Lists workspace spaces |
search |
query, space_id?, limit |
Full-text search across pages |
get_page |
page_id |
Fetches a page by id or slugId |
recent_pages |
space_id?, limit |
Recently changed pages |
create_page |
space_id, title, content?, parent_page_id?, fmt |
Creates a page |
update_page |
page_id, title?, content?, fmt |
Updates title and/or body |
move_page |
page_id, parent_page_id? |
Reparents a page |
delete_page |
page_id, permanently |
Trashes (or permanently deletes) a page |
fmt is markdown (default), html or json.
Notes on the Docmost API
Things worth knowing if you extend this server:
- Every endpoint is
POST, including reads. /pages/createand/pages/updaterequire aformatfield (json|markdown|html). Omitting it returns400.- Responses wrap the payload in
{"data": ...}; the client unwraps it. /api/api-keysanswers200with an empty list on the Community Edition — listing is open, only key creation is licence-gated.- Endpoint paths and payload shapes were taken from the Docmost client source
(
apps/client/src/features/*/services/*.ts), not guessed.
Security
config.jsonholds a plaintext password. Keep it at mode600and out of version control — it is in.gitignore.- Give the agent account the narrowest space access that still lets it work.
- The account's actions appear in Docmost under its own name, so page history stays attributable.
Compatibility
Built against Docmost 0.95.0 and Python MCP SDK 2.0. Note that SDK 2.0 renamed
FastMCP to MCPServer and moved it to mcp.server; the pre-2.0 import path
mcp.server.fastmcp will not work.
Licence
MIT — see LICENSE.
This project is not affiliated with Docmost.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.