A comprehensive Model Context Protocol (MCP) server that provides tools and resources for interacting with LibreOffice documents. This server enables AI assistants and other MCP clients to create, read, convert, and manipulate LibreOffice documents programmatically.
This repository is organized into logical directories:
src/- Core MCP server implementationtests/- Test suite and validation scriptsexamples/- Demo scripts and usage examplesconfig/- Configuration templates for integrationsscripts/- Utility scripts for setup and managementdocs/- Comprehensive documentation
For detailed information, see docs/REPOSITORY_STRUCTURE.md.
- Native Integration: Embedded MCP server directly in LibreOffice
- Real-time Editing: Live document manipulation with instant visual feedback
- Performance: 10x faster than external server (direct UNO API access)
- Multi-document: Work with all open LibreOffice documents
- Auto-start: Automatically available when LibreOffice starts
- HTTP API: External AI assistant access via localhost:8765
- Create Documents: New Writer, Calc, Impress, and Draw documents
- Read Content: Extract text from any LibreOffice document
- Convert Formats: Convert between 50+ formats (PDF, DOCX, HTML, etc.)
- Edit Documents: Insert, append, or replace text in Writer documents
- Document Info: Get detailed metadata about documents
- Read Spreadsheets: Extract data from Calc spreadsheets and Excel files
- Structured Data: Get data as 2D arrays with row/column information
- Document Search: Find documents containing specific text
- Batch Convert: Convert multiple documents simultaneously
- Merge Documents: Combine multiple documents into one
- Document Analysis: Get detailed statistics (word count, sentences, etc.)
- GUI Integration: Open documents in LibreOffice for live viewing
- Real-time Updates: See changes as AI assistants modify documents
- Change Monitoring: Watch documents for modifications in real-time
- Interactive Sessions: Create live editing sessions with automatic refresh
- Document Discovery: List all LibreOffice documents (
documents://) - Content Access: Access specific document content (
document://{path})
- LibreOffice: 24.2+ (must be accessible via command line)
- Python: 3.12+
- UV Package Manager: For dependency management
For detailed installation instructions for all platforms, run:
./mcp-helper.sh requirements-
Clone the repository:
git clone https://github.com/patrup/mcp-libre/ cd mcp-libre -
Check prerequisites:
./mcp-helper.sh requirements # Show detailed requirements ./mcp-helper.sh check # Verify your system
-
Install dependencies:
uv sync
-
Make helper script executable:
chmod +x mcp-helper.sh
# Run functionality tests
./mcp-helper.sh test
# Run interactive demo
./mcp-helper.sh demo# Standard MCP mode (stdio)
python src/main.py
# Or using UV
uv run python src/main.py
# Show help and options
python src/main.py --help
# Run tests
python src/main.py --test# Start the MCP proxy
./mcp-helper.sh proxy
# Then configure Super Assistant extension:
# Server URL: http://localhost:3006| Tool | Description |
|---|---|
create_document |
Create new LibreOffice documents |
read_document_text |
Extract text from documents |
convert_document |
Convert between formats |
get_document_info |
Get document metadata |
read_spreadsheet_data |
Read spreadsheet data |
insert_text_at_position |
Edit document text |
search_documents |
Search documents by content |
batch_convert_documents |
Batch format conversion |
merge_text_documents |
Merge multiple documents |
get_document_statistics |
Document analysis |
open_document_in_libreoffice |
Open document in GUI for live viewing |
create_live_editing_session |
Start live editing with real-time preview |
watch_document_changes |
Monitor document changes in real-time |
refresh_document_in_libreoffice |
Force document refresh in GUI |
- Prerequisites: Quick reference for system requirements
- Plugin Migration Guide: Migrate from external server to plugin
- Examples: Code examples and usage patterns
- Live Viewing Guide: See changes live in LibreOffice GUI
- Super Assistant Setup: Chrome extension integration
- ChatGPT Browser Guide: Using with ChatGPT and alternatives
- Troubleshooting: Common issues and solutions
- Quick Start: Quick reference guide
- Complete Solution: Comprehensive overview
The most powerful and efficient way to use the MCP server:
# Build and install the LibreOffice extension
cd plugin/
./install.sh install
# Test the extension
./install.sh testBenefits of the Extension:
- 10x Performance: Direct UNO API access (no subprocess overhead)
- Real-time Editing: Live document manipulation in open LibreOffice windows
- Native Integration: Appears in LibreOffice Tools menu
- Multi-document Support: Work with all open documents simultaneously
- Auto-start: Automatically starts with LibreOffice
- Advanced Features: Full access to LibreOffice formatting and capabilities
Usage:
- The extension provides an HTTP API on
http://localhost:8765 - Configure your AI assistant to use this endpoint
- Access controls via Tools > MCP Server in LibreOffice
- Real-time document editing with instant visual feedback
For detailed plugin information, see plugin/README.md.
Generate configuration automatically:
./generate-config.sh claude
# Creates ~/.config/claude/claude_desktop_config.jsonThen restart Claude Desktop and start using LibreOffice commands:
- "Create a new Writer document and save it as project-report.odt"
- "Convert my document to PDF format"
Generate configuration and start proxy:
./generate-config.sh mcp
npx @srbhptl39/mcp-superassistant-proxy@latest --config ~/Documents/mcp/mcp.config.json
# Server URL: http://localhost:3006from mcp.shared.memory import create_connected_server_and_client_session
from libremcp import mcp
async with client_session(mcp._mcp_server) as client:
result = await client.call_tool("create_document", {
"path": "/tmp/test.odt",
"doc_type": "writer",
"content": "Hello, World!"
})- "Create a new Writer document with a project report"
- "Convert my ODT file to PDF format"
- "Search for documents containing 'budget' in my Documents folder"
- "Get statistics for my essay - how many words?"
from libremcp import create_document, read_document_text, convert_document
# Create a document
doc = create_document("/tmp/report.odt", "writer", "Project Report")
# Read content
content = read_document_text("/tmp/report.odt")
print(f"Words: {content.word_count}")
# Convert to PDF
result = convert_document("/tmp/report.odt", "/tmp/report.pdf", "pdf")- LibreOffice:
.odt,.ods,.odp,.odg - Microsoft Office:
.doc,.docx,.xls,.xlsx,.ppt,.pptx - Text:
.txt,.rtf
- PDF:
.pdf - Microsoft Office:
.docx,.xlsx,.pptx - Web:
.html,.htm - Text:
.txt - LibreOffice:
.odt,.ods,.odp,.odg - Many others: 50+ formats supported by LibreOffice
# Install and test the plugin
cd plugin/
./install.sh install # Build and install extension
./install.sh test # Test functionality
./install.sh status # Check status
./install.sh interactive # Interactive testing mode# Show system requirements and installation guides
./mcp-helper.sh requirements
# Check dependencies and verify setup
./mcp-helper.sh check
# Run built-in functionality tests
./mcp-helper.sh test
# Interactive demo of all capabilities
./mcp-helper.sh demo
# Test specific functionality directly
uv run python libremcp.py --testGenerate personalized configuration files for Claude Desktop and/or Super Assistant:
# Generate both Claude Desktop and Super Assistant configs
./generate-config.sh
# Generate only Claude Desktop config
./generate-config.sh claude
# Generate only Super Assistant config
./generate-config.sh mcp
# Generate Super Assistant config in custom location
./generate-config.sh mcp /path/to/custom/directoryThis automatically creates configurations with your actual project paths:
- Claude Desktop:
~/.config/claude/claude_desktop_config.json - Super Assistant:
~/Documents/mcp/mcp.config.json(or custom location)
export PYTHONPATH="/path/to/mcp-libre"
export LIBREOFFICE_PATH="/usr/bin/libreoffice" # OptionalEdit libremcp.py to modify document discovery locations:
search_paths = [
Path.home() / "Documents",
Path.home() / "Desktop",
Path("/custom/path"),
Path.cwd()
]- Local Execution: All operations run locally
- File Permissions: Limited to user's file access
- No Network: No external network dependencies
- Temporary Files: Automatically cleaned up
# Check LibreOffice installation
libreoffice --version
libreoffice --headless --help
# Test conversion manually
libreoffice --headless --convert-to pdf document.odt- Java warnings are usually non-fatal
- Core functionality works without Java
- Install Java for full LibreOffice features
- Check file and directory permissions
- Ensure LibreOffice can access document paths
- Verify write permissions for output directories
- Fork the repository
- Create a feature branch
- Make your changes
- Add tests if applicable
- Submit a pull request
This project is licensed under the MIT License - see the LICENSE file for details.
The MIT License is a permissive license that allows:
- โ Commercial use
- โ Modification
- โ Distribution
- โ Private use
For other license options, see LICENSE_OPTIONS.md.
- MCP Specification: https://spec.modelcontextprotocol.io/
- LibreOffice: https://www.libreoffice.org/
- FastMCP Framework: https://github.com/modelcontextprotocol/python-sdk
- Issues: Use GitHub issues for bug reports
- Documentation: See the
docs/folder for detailed guides - Examples: Check
EXAMPLES.mdfor usage patterns
LibreOffice MCP Server v0.1.0 - Bridging AI and Document Processing