MCP Document Server

A simple server implementing the Model Context Protocol (MCP) for document search and retrieval.

Features

- Document storage and retrieval using MCP
- Semantic search capabilities
- Compatible with Cursor and other MCP clients
- Pluggable document service architecture:
- Local document storage with JSON
- Confluence integration for searching workspace documents
- Support for HTML

Setup

1. Install dependencies:

npm install

2. Build the TypeScript code:

npm run build

3. Start the server:

npm run stdio

Using with Cursor

This server is designed to be used with Cursor. You can search for documents using the search tool with a keyword parameter.

Example:

search({
  keyword: "data"
})

Creating Documents

You can create new documents directly from Cursor:

create_document({
  title: "New Confluence Page",
  content: "This is a page created in Confluence",
  serviceId: "confluence"
})

Document Services

The architecture is based on pluggable document services. Each service implements a common interface for searching and retrieving documents.

Local Storage Service

The local storage service stores documents on the local filesystem:
- Documents are stored in data/documents/ directory (created automatically when needed)
- Metadata is stored in data/documents.json (created automatically when needed)
- Supports both Markdown and HTML formats for document content

Confluence Service

The Confluence service searches for documents in your Confluence workspace using the Confluence REST API with CQL (Confluence Query Language) for powerful searches. To enable this feature, set the following environment variables:

```bash