MarkItDown MCP
Microsoft's utility and MCP server that turn PDF, Word, Excel, PowerPoint, HTML and other files into Markdown for models
- 192,752 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Reads local files and fetches URLs with the process's privileges
- Processes document content that may carry prompt injection
Install
- Works in
Manual install
pip install markitdown-mcp
markitdown-mcpRuns over stdio. For HTTP: markitdown-mcp --http --host 127.0.0.1 --port 3001.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Reads local files and fetches URLs with the process's privileges
- Processes document content that may carry prompt injection
This is third-party code. Review the repository files before installing.
What it does
MarkItDown converts documents to Markdown while keeping structure: headings, lists, tables and links. It supports PDF, DOCX, PPTX, XLSX and XLS, HTML, CSV, JSON, XML, EPUB, ZIP archives, images with metadata and OCR, audio with transcription and YouTube links. The repository includes markitdown-mcp, an MCP server with a single convert_to_markdown tool that accepts http, https, file and data URIs. It runs over stdio, Streamable HTTP or SSE, locally or in Docker.
Who it is for. For anyone feeding office documents and PDFs to an agent who wants compact text instead of raw files.
Good fit when
- The agent needs to read a PDF contract or an Excel sheet
- You need to prepare documents for RAG
- You need audio transcription or YouTube subtitles as text
Not a fit when
- You need faithful layout for humans rather than text for a model
- You must process files from untrusted sources without isolation
Example request
Convert report.pdf to Markdown and extract all Q2 figures from its tablesMCP
- Transport
- stdio, http, sse
- Authentication
- not required
Limitations
Requires Python 3.10+, with format support installed via optional extras. OCR through markitdown-ocr needs an OpenAI-compatible client, and Azure modes need a subscription. The tool performs I/O with the process's privileges, so the HTTP server must not be exposed externally.
How to disable. Remove the markitdown server from the MCP config and uninstall with pip uninstall markitdown-mcp markitdown.
FAQ
Can I use it without MCP?
Yes, after pip install 'markitdown[all]' you get the markitdown file.pdf -o document.md command and a Python API.
How do I expose local files in Docker?
Mount a directory to /workdir; files will be available at that path in the container.
Discussion
- former memberQuestion
Кто нибудь уже ставил? Какие ощущения?
- former member
Best answer
да норм
Related
MCP serversAn MCP server that turns PDF, DOCX, XLSX, PPTX, images, audio, web pages and YouTube transcripts into Markdown
An MCP server on top of pandoc: the agent converts Markdown, HTML, DOCX, LaTeX, EPUB, ODT and other formats into one another
MCP PDF from ComPDF: the agent converts PDFs and images to Word, Excel, Markdown and JSON and merges, splits, protects and compares files via a cloud API
A Firecrawl skill and CLI that lets your agent convert Word, PowerPoint, Excel, OpenDocument, RTF, EPUB, CSV and PDF to Markdown on your own machine
MCP Word server for docx editing with tracked changes: comments, footnotes, version comparison and a change log, all visible in Word
An MCP server built on Docling: converts PDFs into structured documents, builds new ones and feeds them to RAG
Comparison with similar tools
Entries that share the main search query, and how they differ.
| Entry | Official | Authentication | Repository stars | Main difference |
|---|---|---|---|---|
| MarkItDown (this entry) | yes | not required | 192,752 | |
| mcp-pandoc | no | not required | 584 | an MCP server on top of pandoc |
| WebArchiveExtractor: unpack webarchive files | no | 346 | a Claude Code skill for macOS that unpacks Safari .webarchive files |
README in short
The README explains the Markdown choice: models understand it well and the markup is token-efficient. It covers pip install with per-format extras, CLI and Python usage, third-party plugins and the OCR plugin, and Azure Document Intelligence and Content Understanding integrations. The markitdown-mcp package README covers stdio, HTTP and Docker runs and a Claude Desktop config. MIT licensed.