MarkItDown MCP: A useful AI tool to convert files and office documents to Markdown!

Document processing can be said to occupy most of our daily work. Whether it is our company's internal knowledge management or personal learning materials, it is undoubtedly a very tricky and headache-inducing task for us to convert files of different formats into a unified and easy-to-process format.

Well, the tool we are going to recommend today can help you solve this problem, MarkItDown , an open source project supported by Microsoft. It can not only convert multiple file formats into Markdown, but also is specially optimized for the application of Large Language Models (LLM), becoming a wise helper in the document processing tool industry.

The MarkItDown MCP service supports a wide range of file formats, from common PDF, Word, Excel, PowerPoint to images (supporting OCR recognition), audio (supporting transcription), and HTML.

From text-based formats (such as CSV, JSON, XML), to even ZIP files and YouTube links, whether you are doing academic research, business reports or a small project of your own, MarkItDown can help us deal with it easily.

The advantage of the Markdown format can be summed up in two words: " concise". MarkItDown is known for its simplicity and readability. It is very close to plain text, but provides sufficient structural capabilities to represent the key elements of the document.

For mainstream large-scale language models (such as OpenAI's GPT-4o), Markdown can be said to be their "mother tongue", and they are trained on a large amount of Markdown-formatted text.

Therefore, we can better understand and generate Markdown content, which makes the output of MarkItDown not only suitable for our manual reading, but also an ideal input format for LLM.

How to install and use MarkItDown? Here we introduce the configuration method of MarkItDown:

To install MarkItDown, you can use  pip:pip install 'markitdown[all]'. Or you can install from source:

git clone git@github.com:microsoft/markitdown.gitcd markitdownpip install -e packages/markitdown[all]

Using the command line

markitdown path-to-file.pdf > document.md

Or use -o Specify output file:

markitdown path-to-file.pdf -o document.md

You can also pipe content:

cat path-to-file.pdf | markitdown

Optional dependencies,MarkItDown has optional dependencies for activating various file formats. Earlier in the document, we installed all optional dependencies using the [all] option. They can also be installed individually to gain more control. For example:

pip install markitdown[pdf, docx, pptx]

Only dependencies for PDF, DOCX, and PPTX files will be installed.

Currently, the following optional dependencies are available:

• [all]
   Install all optional dependencies;
• [pptx]
   Install dependencies for PowerPoint files;
• [docx]
   Install dependencies for Word files;
• [xlsx]
   Install the dependencies for the Excel file;
• [xls]
   Install dependencies for older Excel files;
• [pdf]
   Install dependencies for PDF files;
• [outlook]
   Install dependencies for Outlook messaging;
• [az-doc-intel]
   Install Azure Document Intelligence dependencies;
• [audio-transcription]
   Install audio transcription dependencies for wav and mp3 files;
• [youtube-transcription]
   Install dependencies for getting YouTube video transcriptions;
  
  

In terms of plug-ins, MarkItDown also supports third-party plug-ins, which are disabled by default. Use the following command to list installed plug-ins:

markitdown --list-plugins

To enable the plugin, use the following command:

markitdown --use-plugins path-to-file.pdf

To find available plugins, we can search for the tag on GitHub #markitdown-plugin ,To develop a plugin, refer to:

packages/markitdown-sample-plugin

Azure Document Intelligence, to use Microsoft Document Intelligence for conversion:

markitdown path-to-file.pdf -o document.md -d -e "<document_intelligence_endpoint>"

Python API

from markitdown import MarkItDownmd = MarkItDown(enable_plugins=False) # Set to True to enable pluginsresult = md.convert("test.xlsx")print(result.text_content)

Document Intelligence transformation in Python:

from markitdown import MarkItDownmd = MarkItDown(docintel_endpoint="<document_intelligence_endpoint>")result = md.convert("test.pdf")print(result.text_content)

To use a large language model for image description, provide llm_client and llm_model:

from markitdown import MarkItDownfrom openai import OpenAIclient = OpenAI()md = MarkItDown(llm_client=client, llm_model="gpt-4o")result = md.convert("example.jpg")print(result.text_content)

Docker

docker build -t markitdown:latest .docker run --rm -i markitdown:latest < ~/your-file.pdf > output.md

The above is all about MarkItDown configuration. If you encounter any problems during the learning process, please leave a message in the comment area for discussion!