Office Word MCP Server

Instructions

Implementation Reference

• word_document_server/tools/extended_document_tools.py:18-39 (handler)

  Primary handler function implementing the tool logic: validates inputs, ensures file extension, loads via helper, serializes to JSON.

  async def get_paragraph_text_from_document(filename: str, paragraph_index: int) -> str:
      """Get text from a specific paragraph in a Word document.
      
      Args:
          filename: Path to the Word document
          paragraph_index: Index of the paragraph to retrieve (0-based)
      """
      filename = ensure_docx_extension(filename)
      
      if not os.path.exists(filename):
          return f"Document {filename} does not exist"
      
  
      if paragraph_index < 0:
          return "Invalid parameter: paragraph_index must be a non-negative integer"
      
      try:
          result = get_paragraph_text(filename, paragraph_index)
          return json.dumps(result, indent=2)
      except Exception as e:
          return f"Failed to get paragraph text: {str(e)}"

• word_document_server/main.py:805-808 (registration)

  Tool registration using FastMCP @mcp.tool() decorator. Defines the tool endpoint with type hints and docstring for schema, delegates to implementation.

  async def get_paragraph_text_from_document(filename: str, paragraph_index: int):
      """Get text from a specific paragraph in a Word document."""
      return await extended_document_tools.get_paragraph_text_from_document(filename, paragraph_index)

• word_document_server/utils/extended_document_utils.py:8-39 (helper)

  Supporting utility that loads the Document object using python-docx, extracts paragraph by index, includes metadata like style and heading status, returns structured dict.

  def get_paragraph_text(doc_path: str, paragraph_index: int) -> Dict[str, Any]:
      """
      Get text from a specific paragraph in a Word document.
      
      Args:
          doc_path: Path to the Word document
          paragraph_index: Index of the paragraph to extract (0-based)
      
      Returns:
          Dictionary with paragraph text and metadata
      """
      import os
      if not os.path.exists(doc_path):
          return {"error": f"Document {doc_path} does not exist"}
      
      try:
          doc = Document(doc_path)
          
          # Check if paragraph index is valid
          if paragraph_index < 0 or paragraph_index >= len(doc.paragraphs):
              return {"error": f"Invalid paragraph index: {paragraph_index}. Document has {len(doc.paragraphs)} paragraphs."}
          
          paragraph = doc.paragraphs[paragraph_index]
          
          return {
              "index": paragraph_index,
              "text": paragraph.text,
              "style": paragraph.style.name if paragraph.style else "Normal",
              "is_heading": paragraph.style.name.startswith("Heading") if paragraph.style else False
          }
      except Exception as e:
          return {"error": f"Failed to get paragraph text: {str(e)}"}