home / mcp / documentation mcp server

Documentation MCP Server

Exposes Markdown and OpenAPI sources for AI assistants to search, navigate, and view via MCP.

Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
  "mcpServers": {
    "esola-thomas-your-docs-mcp": {
      "command": "your-docs-mcp",
      "args": [],
      "env": {
        "DOCS_ROOT": "YOUR_PATH_TO_DOCS"
      }
    }
  }
}

You can run a self-contained MCP server that exposes your Markdown and OpenAPI documentation for AI assistants to search, navigate, and query. It provides a browser interface and programmable endpoints so you can integrate documentation access into your workflows.

How to use

Start by running the MCP server to expose your documentation to AI assistants and to a web interface. You will set the root directory for your docs, then launch the server so both the MCP protocol and the web interface are available.

Once running, you can:

To integrate with an MCP client, point the client at the local server process. The standard start command is the executable you run to start the server, and your client will use the same information to request documents, search results, and navigation data.

How to install

Prerequisites: you need Python and Git installed on your system.

# Install from PyPI
pip install your-docs-mcp

# Or install from source
git clone https://github.com/esola-thomas/Markdown-MCP
cd Markdown-MCP
pip install -e .

Basic configuration and startup

Set the documentation root directory and start the MCP server. The server reads your docs from the path you specify.

export DOCS_ROOT=/path/to/your/docs
your-docs-mcp

Configuring client integrations

You can configure clients to use the MCP server by supplying the command to launch the MCP process and the root directory path as an environment variable. The following examples show typical configurations for popular environments.

{
  "mcpServers": {
    "docs": {
      "command": "your-docs-mcp",
      "env": {
        "DOCS_ROOT": "/absolute/path/to/your/docs"
      }
    }
  }
}

Web interface and API

A built-in web interface is available for browser-based access to the same tools your AI assistants use. By default, you can open the web UI at the local address provided by the server.

The web interface exposes a table of contents, full-text search, tag filters, and document viewing. You can also interact with REST API endpoints to health-check, search, navigate, and fetch documents.

Security and reliability notes

The server validates paths, sanitizes queries, and logs access attempts to support auditing and safe usage. Hidden files are excluded, and file changes are detected to keep caches up to date.

If you need to adjust web server behavior, you can enable or disable it and configure host/port using environment variables when you start the server.

Troubleshooting

If the web interface does not appear, verify your DOCS_ROOT points to a valid documentation directory and that the server process is running. Check that the environment variable matches your absolute path.

If you encounter search issues, confirm that your documents are in supported formats (Markdown with YAML frontmatter and OpenAPI specs) and that the parser has loaded them from the root directory.

Available tools

navigate

Navigate the documentation hierarchy to browse nested sections and articles

search

Perform full-text searches across documents with optional metadata filters

getDocument

Retrieve full content of a specific document by URI or path

getTOC

Get the table of contents representing the documentation hierarchy

health

Check the health and status of the MCP server and loaded resources