MCP PDF and DOCX Server

This MCP server provides tools to read PDFs, extract images, convert PDF to DOCX, create DOCX from text, and generate PDFs from DOCX or text or HTML.

Setup

1. Create and activate a virtual environment.

2. Install dependencies:

pip install -e .

Run (stdio)

python -m mcp_pdf_server

VS Code Copilot MCP

If you want Copilot Agents to auto-install and run this server, use uvx in your MCP config:

{
	"servers": {
		"pdf-docx-tools": {
			"type": "stdio",
			"command": "uvx",
			"args": ["--from", "git+https://github.com/iduryodhanrao/MCP.git", "python", "-m", "mcp_pdf_server"]
		}
	}
}

This requires uv/uvx to be installed on the machine.

Notes

• DOCX to PDF conversion uses docx2pdf. On Windows, this requires Microsoft Word to be installed.

• HTML to PDF uses weasyprint. On Windows, additional system libraries may be required.

Tools

• read_pdf_text(path) -> string

• extract_pdf_images(path, output_dir) -> list of image paths

• pdf_to_docx(pdf_path, docx_path) -> docx path

• docx_to_pdf(docx_path, pdf_path) -> pdf path

• text_to_docx(text, docx_path) -> docx path

• text_or_html_to_pdf(content, pdf_path, content_type) -> pdf path