MCP Document Server

What is MCP Document Server?

MCP Document Server is a Model Context Protocol (MCP) server for document search and retrieval. It supports pluggable document services—including local JSON storage and Confluence integration—and is compatible with MCP clients like Cursor. It is designed for users who need to search and manage documents through an MCP interface.

How to use MCP Document Server?

Install dependencies with npm install, build the TypeScript code with npm run build, then start the server with npm run stdio. In Cursor, use the search tool with a keyword parameter (e.g., search({ keyword: "data" })) and the create_document tool to create documents. For Confluence, set environment variables for CONFLUENCE_BASE_URL, CONFLUENCE_USERNAME, CONFLUENCE_API_TOKEN, and CONFLUENCE_SPACE before starting the server.

Key features of MCP Document Server

• Document storage and retrieval via MCP
• Semantic search capabilities
• Compatible with Cursor and other MCP clients
• Pluggable document services: local JSON and Confluence
• Support for HTML document content
• Extensible architecture via the DocumentService interface

Use cases of MCP Document Server

• Searching local documents stored in JSON files
• Searching Confluence workspace documents using CQL queries
• Creating documents directly from MCP clients like Cursor
• Extending the server with new document services (e.g., Google Drive, Notion)

FAQ from MCP Document Server

What protocols or transports does the server use?

The server implements the Model Context Protocol (MCP) and is designed to run over stdio. It does not mention other transports.

What runtime or dependencies are required?

Node.js and npm are required. The server is built with TypeScript and uses npm install and npm run build for setup.

How does the Confluence integration authenticate?

It uses Basic Authentication with your email and API token, sent as a Base64-encoded string in the Authorization header.

Where are local documents stored?

Local documents are stored in the data/documents/ directory, with metadata in data/documents.json. These files are created automatically on first use and are not included in the repository.

Can the server be extended with other document services?

Yes. The server uses a pluggable architecture. You can add new services by implementing the DocumentService interface and registering the service in src/services/index.ts. Examples mentioned include Google Drive, OneDrive/SharePoint, Notion, GitHub Wiki, and custom databases.