@cadivus/mcp-server-lo-doc-reader
v0.1.0
Published
LibreOffice Document Reader MCP Server - converts documents to Markdown via MCP
Maintainers
Readme
mcp-server-lo-doc-reader
The local-first, privacy-focused MCP server for high-fidelity document parsing.
lo-doc-reader is a Model Context Protocol (MCP) server that enables AI assistants (like Claude, Cursor, and Windsurf) to read complex documents by leveraging a local LibreOffice installation.
🌟 Why lo-doc-reader?
While many document parsers struggle with complex layouts or legacy formats, lo-doc-reader uses the battle-tested LibreOffice engine to provide:
- 🔒 Absolute Privacy: Your documents never leave your machine. No cloud APIs, no data collection, and no third-party processing.
- 🏛️ Superior Legacy Support: Seamlessly handles
.doc,.rtf, and older formats that break modern lightweight libraries. - ⚙️ Headless Efficiency: Runs in the background using isolated user profiles. It won't interfere with your open LibreOffice windows or documents.
- 💰 Zero Cost: No per-page fees or subscription limits.
🚀 Quick Start
Prerequisites
- Node.js ≥ 18
- LibreOffice ≥ 24.2 (Download here)
MCP Client Configuration
Add this to your claude_desktop_config.json or equivalent client settings:
{
"mcpServers": {
"lo-doc-reader": {
"command": "npx",
"args": ["-y", "@cadivus/mcp-server-lo-doc-reader"]
}
}
}📂 Supported File Formats
| Extension | Description |
| :--- | :--- |
| .docx / .doc | Microsoft Word (Modern & Legacy) |
| .odt / .fodt | OpenDocument Text / Flat XML |
| .rtf | Rich Text Format |
| .epub | Electronic Publication (E-books) |
| .html / .htm | Web Documents |
🛠️ How It Works
- Detection: Automatically finds the
sofficebinary across macOS, Windows, and Linux. - Sandboxing: Creates a temporary, isolated user profile for every request to ensure thread safety and prevent configuration conflicts.
- Conversion: Invokes a headless instance of LibreOffice to perform a native export to Markdown.
- Cleanup: Instantly purges all temporary files and profiles once the content is delivered to your LLM.
🧰 Available Tools
read_document
Extracts text from a document and returns it as LLM-optimized Markdown.
- Input:
filePath(The absolute path to your file).
list_supported_formats
Displays all extensions currently supported by your local installation.
⚠️ Limitations & Notes
- Formatting: Complexity is limited by LibreOffice's Markdown export filter (tables and images may have simplified rendering).
- Scopes: Spreadsheets (
.xlsx) and Presentations (.pptx) are not currently supported. - Paths: Ensure you provide the absolute path to the file so the server can locate it reliably.
🏷️ Keywords & Use Cases
mcp-server, context-window, RAG, local-llm, document-parsing, privacy-focused-ai, word-to-markdown, offline-ai-tools, claude-desktop-tools.
📜 License
Copyright © 2026. Licensed under the Apache License, Version 2.0.
