MCP Document Reader

A Model Context Protocol (MCP) server for extracting text content from office documents. Pure TypeScript, zero system dependencies.

Supported Formats

Extension	Format
.docx	Microsoft Word
.xlsx	Microsoft Excel
.pptx	Microsoft PowerPoint
.odt	OpenDocument Text
.ods	OpenDocument Spreadsheet
.odp	OpenDocument Presentation
.pdf	PDF Document
.rtf	Rich Text Format

Installation

OpenCode / Matrixx

Add to your .mcp.json:

{
  "mcpServers": {
    "document_reader": {
      "command": "npx",
      "args": ["-y", "@klpanagi/mcp-document-reader"]
    }
  }
}

Claude Code

Add to your Claude Code MCP settings:

{
  "mcpServers": {
    "document_reader": {
      "command": "npx",
      "args": ["-y", "@klpanagi/mcp-document-reader"]
    }
  }
}

Cursor / Other MCP Clients

Use the stdio transport with npx -y @klpanagi/mcp-document-reader as the command.

Tools

read_document

Extract text content from a document file.

Parameter	Type	Required	Description
file_path	string	Yes	Absolute path to the document
include_metadata	boolean	No	Include format/size header (default: false)

get_document_info

Get metadata about a document without extracting content.

Parameter	Type	Required	Description
file_path	string	Yes	Absolute path to the document

list_supported_formats

List all supported document formats. No parameters.

Configuration

Environment variables for tuning:

Variable	Default	Description
MAX_FILE_SIZE_MB	50	Maximum file size to process
MAX_OUTPUT_CHARS	500000	Truncation limit for extracted text

Development

bun install
bun test
bun run build

License

MIT