LibreOffice MCP Server
Enables AI assistants to create, read, edit, and manipulate LibreOffice documents with support for track changes, comments, search/replace, and real-time document operations through a native extension or HTTP API.
README
LibreOffice MCP Server
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.
🚀 Features
LibreOffice Extension (Plugin) - Recommended!
- 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)
- 9 Consolidated Tools: Reduced from 32 individual tools for better UX
- Track Changes Support: Full revision tracking awareness
- Multi-document: Work with all open LibreOffice documents
- HTTP API: External AI assistant access via localhost:8765
Document Operations
- Create Documents: New Writer, Calc, Impress, and Draw documents
- Read Content: Extract text with visible_content for Track Changes awareness
- Navigate: Paragraph-level navigation, cursor positioning, document outline
- Edit: Insert, format, select, and replace text
- Search: Find/replace with Track Changes awareness (skips tracked deletions)
- Comments: Add and retrieve document annotations
- Track Changes: Enable, disable, list, accept/reject revisions
🔧 9 Consolidated MCP Tools
The MCP interface provides 9 logical tool groups (consolidated from 32 individual tools):
| Tool | Actions | Description |
|---|---|---|
document |
create, info, list, content, status | Document management |
structure |
outline, paragraph, range, count | Document navigation |
cursor |
goto_paragraph, goto_position, position, context | Cursor control |
selection |
paragraph, range, delete, replace | Text selection |
search |
find, replace, replace_all | Search/replace (Track Changes aware) |
track_changes |
status, enable, disable, list, accept, reject, accept_all, reject_all | Revision tracking |
comments |
list, add | Comment management |
save |
save, export | Save/export documents |
text |
insert, format | Text insertion and formatting |
See docs/TOOL_REFERENCE.md for complete documentation.
📋 Requirements
- LibreOffice: 24.2+ (must be accessible via command line)
- Python: 3.12+
- UV Package Manager: For dependency management
🛠 Installation
LibreOffice Extension (Recommended)
# Clone the repository
git clone https://github.com/jwingnut/mcp-libre.git
cd mcp-libre
# Build and install the LibreOffice extension
cd plugin/
./build.sh
unopkg add ../build/libreoffice-mcp-extension-1.0.0.oxt
# Restart LibreOffice
After installation:
- Open LibreOffice Writer
- Go to Tools > MCP Server > Start MCP Server
- The HTTP API is now available at
http://localhost:8765
FastMCP Bridge (for Claude Code)
# Install FastMCP
pip install fastmcp httpx
# Configure Claude Code
claude mcp add libreoffice -- fastmcp run /path/to/libreoffice_mcp_server.py
🎯 Quick Start
Using with Claude Code
Once configured, you can use natural language commands:
"Get document info"
→ document(action="info")
"Find all occurrences of 'hello'"
→ search(action="find", query="hello")
"Enable track changes"
→ track_changes(action="enable")
"Go to paragraph 5 and get context"
→ cursor(action="goto_paragraph", n=5)
→ cursor(action="context")
HTTP API Examples
# Check server status
curl http://localhost:8765/health
# Get document info
curl -X POST http://localhost:8765/tools/get_document_info_live -d '{}'
# Find text
curl -X POST http://localhost:8765/tools/find_text_live \
-H "Content-Type: application/json" \
-d '{"query": "hello"}'
# Enable Track Changes
curl -X POST http://localhost:8765/tools/set_track_changes_live \
-H "Content-Type: application/json" \
-d '{"enabled": true}'
📂 Repository Structure
mcp-libre/
├── plugin/ # LibreOffice extension
│ ├── pythonpath/
│ │ ├── uno_bridge.py # UNO API wrapper
│ │ └── mcp_server.py # HTTP API server
│ ├── build.sh # Build script
│ └── README.md # Plugin documentation
├── libreoffice_mcp_server.py # FastMCP bridge (9 consolidated tools)
├── docs/
│ ├── TOOL_REFERENCE.md # Complete tool documentation
│ └── KNOWN_ISSUES_AND_ROADMAP.md
├── src/ # Legacy external server
├── tests/ # Test suite
└── README.md
📚 Documentation
- Tool Reference: Complete documentation of all 9 tools
- Plugin README: LibreOffice extension details
- Known Issues & Roadmap: Future plans
🆕 Recent Changes
v0.3.0 - Tool Consolidation
- Consolidated 32 tools into 9 logical groups
- Reduced permission prompts for better UX
- Each tool uses
actionparameter for routing
v0.2.0 - Track Changes Awareness
- Added 7 Track Changes management tools
- Search/replace now skips tracked deletions
get_paragraphreturnsvisible_contentfieldfind_textreturnstrack_changes_activefield
v0.1.0 - Initial Release
- 25 individual MCP tools
- HTTP API on localhost:8765
- LibreOffice extension with UNO API integration
🛡 Security
- Local Execution: All operations run locally
- File Permissions: Limited to user's file access
- No Network: No external network dependencies
- Email Privacy: Uses GitHub noreply email for commits
🤝 Contributing
- Fork the repository
- Create a feature branch
- Make your changes
- Submit a pull request
📄 License
This project is licensed under the MIT License - see the LICENSE file for details.
🔗 Links
- Repository: https://github.com/jwingnut/mcp-libre
- MCP Specification: https://spec.modelcontextprotocol.io/
- LibreOffice: https://www.libreoffice.org/
- FastMCP: https://github.com/modelcontextprotocol/python-sdk
LibreOffice MCP Server v0.3.0 - AI-Powered Document Editing
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。