[!WARNING] Dieses Projekt wurde von GitHub Copilot geschrieben und wurde nicht vollständig von einem Menschen überprüft. Der Code kann Fehler, Sicherheitsprobleme oder unerwartetes Verhalten enthalten. Die Nutzung erfolgt auf eigenes Risiko. Nicht ohne gründliche Überprüfung in der Produktion verwenden.
A Model Context Protocol (MCP)-Server, der KI-Agenten (Claude, Copilot, Gemini, Cursor usw.) die Fähigkeit gibt, Office-Dokumente zu lesen, zu schreiben und zu bearbeiten – über LibreOffice, mit einem token-effizienten Design, das die LLM-Kontextnutzung minimiert.
Inspiriert von der Architektur von chrome-devtools-mcp.
22 MCP-Tools für das Lesen, Schreiben, Tabellenkalkulationen und Präsentationen
Token-effizientes Design: Gliederungsbasierte Navigation, bereichsbasierter Zugriff, Paginierung
Breite Formatunterstützung: DOCX, DOC, XLSX, XLS, PPTX, PPT, ODT, ODS, ODP, RTF, CSV, TXT, PDF
Legacy-Format-Brücke: .doc, .xls, .ppt werden vor dem Parsen automatisch über LibreOffice konvertiert
Kein LibreOffice für einfaches Lesen erforderlich: Native Parser verarbeiten DOCX, XLSX, PPTX direkt
LibreOffice erforderlich für: Legacy-Formate, PDF-Export, Formatkonvertierung
Related MCP server: docx-mcp-server
Format | Erweiterungen | Lesen | Schreiben | Methode |
Word 2007+ |
| ✅ | ✅ | Native (mammoth read / JSZip OOXML write) |
Word 97-2003 |
| ✅ | ✅ | LibreOffice-Brücke |
Excel 2007+ |
| ✅ | ✅ | Native (ExcelJS) |
Excel 97-2003 |
| ✅ | ✅ | LibreOffice-Brücke |
PowerPoint 2007+ |
| ✅ | ✅ | Native (JSZip OOXML) |
PowerPoint 97-2003 |
| ✅ | ✅ | LibreOffice-Brücke |
OpenDocument Text |
| ✅ | ✅ | LibreOffice-Brücke |
OpenDocument Spreadsheet |
| ✅ | ✅ | LibreOffice-Brücke |
OpenDocument Presentation |
| ✅ | ✅ | LibreOffice-Brücke |
Rich Text Format |
| ✅ | ✅ | LibreOffice-Brücke |
CSV |
| ✅ | ✅ | Native |
| ✅ (Text) | ❌ | LibreOffice CLI | |
Klartext |
| ✅ | ✅ | Native |
Node.js 20+
LibreOffice (optional für einfaches Lesen von DOCX/XLSX/PPTX; erforderlich für .doc/.xls/.ppt und Formatkonvertierung)
Windows: LibreOffice herunterladen
macOS: brew install --cask libreoffice
Linux: sudo apt install libreoffice oder sudo dnf install libreoffice
Mit npx (empfohlen – keine Installation erforderlich):
{
"mcpServers": {
"libreoffice": {
"command": "npx",
"args": ["-y", "@passerbyflutter/libreoffice-mcp-tools"]
}
}
}Globale Installation:
npm install -g @passerbyflutter/libreoffice-mcp-toolsAus dem Quellcode:
git clone https://github.com/passerbyflutter/libreoffice-mcp-tools
cd libreoffice-mcp-tools
npm install
npm run buildFügen Sie dies zur Konfiguration Ihres MCP-Clients hinzu (z. B. Claude Desktop claude_desktop_config.json):
{
"mcpServers": {
"libreoffice": {
"command": "npx",
"args": ["-y", "@passerbyflutter/libreoffice-mcp-tools"],
"env": {
"SOFFICE_PATH": "/path/to/soffice"
}
}
}
}Oder verwenden Sie .mcp.json in Ihrem Projektstammverzeichnis:
{
"mcpServers": {
"libreoffice": {
"command": "npx",
"args": ["-y", "@passerbyflutter/libreoffice-mcp-tools"]
}
}
}node build/bin/libreoffice-mcp.js [options]
--libreoffice-path <path> Path to soffice executable
(default: auto-detected or SOFFICE_PATH env)Tool | Beschreibung |
| Öffnet eine Datei → gibt |
| Gibt Dokument-Handle und temporäre Dateien frei |
| Listet alle geöffneten Dokumente auf |
| Erstellt ein neues leeres Dokument (Writer/Calc/Impress) |
| Speichert unter dem aktuellen oder einem neuen Pfad |
| Exportiert über LibreOffice (PDF, HTML, CSV, etc.) |
| Konvertiert das Dateiformat (DOC→DOCX, XLSX→CSV, etc.) |
Tool | Beschreibung |
| Titel, Autor, Wort-/Seitenzahl, Daten |
| Überschriften (Writer) / Tabellennamen (Calc) / Folientitel (Impress) |
| Paginierter Dokumenttext als Markdown |
| Spezifischer Absatz- oder Folienbereich |
| Findet Text mit umgebendem Kontext |
Tool | Beschreibung |
| Einfügen am Anfang/Ende/nach Überschrift |
| Suchen & Ersetzen (erstes oder alle Vorkommen) |
| Absatz an einem bestimmten Index einfügen |
| Überschriften-/Absatzformatvorlage anwenden |
Tool | Beschreibung |
| Tabellennamen mit Zeilen-/Spaltenanzahl |
| Zellbereich als JSON + Markdown-Tabelle |
| Zellwert oder Formel festlegen |
| 2D-Wertebereich festlegen |
| Neues Tabellenblatt hinzufügen |
| Formelausdrücke in einem Bereich abrufen |
Tool | Beschreibung |
| Folientitel mit Index |
| Vollständiger Folieninhalt (Titel, Text, Notizen) |
| Sprechernotizen |
| Neue Folie hinzufügen (erfordert LibreOffice) |
| Folieninhalt aktualisieren |
Für maximale Token-Effizienz folgen Sie diesem Muster:
1. document_open(filePath) → get docId
2. document_get_metadata(docId) → understand size/type
3. document_get_outline(docId) → see structure
4. document_read_range(docId, startIndex=N, endIndex=M) → read specific sectionAnstatt das gesamte Dokument auszugeben, navigieren Sie genau zu dem, was Sie benötigen.
Tabellenkalkulations-Workflow:
1. document_open(path) → docId
2. spreadsheet_list_sheets(docId) → see all sheets
3. spreadsheet_get_range(docId, sheetName="Sales", range="A1:D20") → targeted datasrc/
├── index.ts # createMcpServer() — MCP server factory
├── LibreOfficeAdapter.ts # soffice subprocess manager
├── DocumentContext.ts # Open document registry
├── DocumentSession.ts # Per-document state + format bridge
├── McpResponse.ts # Response builder (text/JSON/markdown)
├── Mutex.ts # Serializes LibreOffice subprocess calls
├── parsers/
│ ├── DocxParser.ts # DOCX read → {paragraphs, outline, metadata} (mammoth)
│ ├── DocxOoxmlEditor.ts # DOCX write → direct JSZip OOXML manipulation (format-preserving)
│ ├── XlsxParser.ts # XLSX read/write via ExcelJS
│ ├── PptxParser.ts # PPTX read → {slides[]} (JSZip XML)
│ └── PptxOoxmlEditor.ts # PPTX write → add/update slides, create PPTX (JSZip OOXML)
├── formatters/
│ ├── MarkdownFormatter.ts
│ ├── JsonFormatter.ts
│ └── TableFormatter.ts # Spreadsheet → Markdown table
└── tools/
├── documents.ts # open/close/list/create
├── reader.ts # metadata/outline/read/search
├── writer.ts # insert/replace/style
├── spreadsheet.ts # get/set cells/ranges/sheets
├── presentation.ts # slides/notes
└── converter.ts # save/export/convert# Create sample fixtures
node tests/create-fixtures.mjs
# Run smoke tests
npm testVariable | Beschreibung |
| Pfad zur ausführbaren LibreOffice-Datei |
| Auf |
MIT
Apply a heading or paragraph style to a specific paragraph by index.
Inputs: docId (required), paragraphIndex (required), style (required).
Close an open document and release its resources (including any temp files from format bridging).
Inputs: docId (required).
Convert a document file to a different format using LibreOffice CLI. Creates a new file; original is not modified. Useful for: DOC→DOCX, DOCX→PDF, XLSX→CSV, etc. LibreOffice must be installed.
Inputs: filePath (required), format (required), outputPath.
Create a new empty document and open it. Returns a docId. Supports creating: writer (text document → .docx), calc (spreadsheet → .xlsx), impress (presentation → .pptx).
Inputs: documentType (required), filePath (required).
Export a document to a different format using LibreOffice. Supports exporting to: PDF, HTML, TXT, DOCX, XLSX, PPTX, ODT, ODS, ODP, CSV. LibreOffice must be installed.
Inputs: docId (required), format (required), outputPath.
Get metadata for an open document: title, author, page count, word count, format, file size, dates. Token-efficient: use this before reading document content to understand its size and structure.
Inputs: docId (required).
Get the structural outline of a document:
Writer documents: headings hierarchy (H1, H2, H3...)
Calc spreadsheets: list of sheet names with row/column counts
Impress presentations: slide titles with index numbers Token-efficient: always call this before document_read_text to understand structure.
Inputs: docId (required).
Insert a new paragraph at a specific index in a Writer document.
Inputs: docId (required), text (required), index (required), style.
Insert text into a Writer document at a specified position. Position options: 'start' (beginning), 'end' (end of document), or after a specific heading text.
Inputs: docId (required), text (required), position, headingText, +1 more.
List all currently open documents with their docId, file path, format, and size.
Inputs: no named parameters.
Open a document file and return a docId handle for use with other tools. Supports: .docx, .doc, .xlsx, .xls, .pptx, .ppt, .odt, .ods, .odp, .rtf, .csv, .txt, .pdf Legacy binary formats (.doc, .xls, .ppt, .odt, .ods, .odp, .rtf) are automatically converted via LibreOffice before parsing. Returns docId to use with all other document tools.
Inputs: filePath (required).
Read a specific range of a document by paragraph index (for Writer) or slide index (for Impress). More token-efficient than document_read_text when you only need a specific section.
Inputs: docId (required), startIndex (required), endIndex.
Read document content as Markdown text. For large documents, use limit/offset for pagination. Default returns up to 4000 characters. Use document_get_outline first to understand structure.
Inputs: docId (required), limit, offset, maxChars.
Find and replace text in a Writer document. Optionally replace all occurrences.
Inputs: docId (required), find (required), replace (required), replaceAll, +1 more.
Save the current state of a document to its file path (or a new path).
Inputs: docId (required), filePath.
Search for text within a document. Returns matching paragraphs/cells/slides with surrounding context.
Inputs: docId (required), query (required), limit.
Add a new slide to a presentation.
Inputs: docId (required), title (required), body, notes.
Get speaker notes for all slides or a specific slide.
Inputs: docId (required), slideIndex.
Get the full content of a specific slide: title, body text, and speaker notes.
Inputs: docId (required), slideIndex (required).
List all slides in a presentation with their index and title. Token-efficient overview before reading content.
Inputs: docId (required).
Update the title or body text of an existing slide.
Inputs: docId (required), slideIndex (required), title, body.
Add a new sheet to a spreadsheet workbook.
Inputs: docId (required), sheetName (required), headers.
Get all formulas in a spreadsheet range (returns formula expressions, not computed values).
Inputs: docId (required), sheetName, range.
Read a range of cells from a spreadsheet sheet. Returns data as structured JSON and a markdown table. Use range like "A1:D10" for specific cells, or omit for paginated full sheet. Token-efficient: specify a range rather than reading the whole sheet.
Inputs: docId (required), sheetName, range, limit, +2 more.
List all sheets in a spreadsheet with their names, row counts, and column counts.
Inputs: docId (required).
Set the value or formula of a single cell in a spreadsheet. Changes are saved immediately.
Inputs: docId (required), cell (required), value (required), sheetName, +1 more.
Set multiple cell values in a spreadsheet. Provide a 2D array of values matching the range.
Inputs: docId (required), startCell (required), values (required), sheetName.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 27 tools
Most tools have distinct purposes with clear boundaries, such as document operations versus presentation or spreadsheet tools. However, some overlap exists: document_read_text and document_read_range both handle reading content, and document_convert and document_export both involve format conversion, which could cause minor confusion in selection.
Tool names follow a highly consistent verb_noun pattern with clear prefixes (document_, presentation_, spreadsheet_) that group related operations. All names use snake_case uniformly, making them predictable and easy to parse for agents.
With 27 tools, the count is borderline high for a single server, potentially overwhelming for agents. While the tools cover a broad range of LibreOffice operations, the number feels heavy and could be streamlined by consolidating overlapping functions like reading or conversion tools.
The tool set provides comprehensive coverage for document, presentation, and spreadsheet workflows, including creation, reading, updating, conversion, and metadata operations. There are no obvious gaps; agents can perform full CRUD/lifecycle tasks across all supported file types without dead ends.
Convert, create and edit Word, PowerPoint, Excel and PDF documents from any AI agent.
Online office suite for AI agents: edit docs, sheets, slides and PDFs in browser or via MCP.
Document conversion and OCR for AI agents: PDF, Office docs, images to text.
Real Word and Excel for agents: open your .docx/.xlsm, read, propose, apply, get it back intact.
