markitdown

name

markitdown

description

Convert files and URLs to Markdown using Microsoft's markitdown. Use for web pages, web PDFs, or local files — supports: PDF, PowerPoint, Word, Excel, Images (EXIF + OCR), Audio (transcription), HTML, CSV/JSON/XML, ZIP files, YouTube URLs, EPubs, and more. Do NOT use for GitHub URLs — convert to raw.githubusercontent.com and fetch directly instead. Runs via uvx with no install needed. You only need this when you need Markdown! For some links, curl/wget is fine!

MarkItDown

Convert files and web pages to Markdown using Microsoft MarkItDown. Runs instantly via uvx — no installation or venv required.

Read a Web Page

uvx markitdown https://example.com

Fetches the URL and extracts readable content as Markdown. Works well on JS-heavy sites where simpler extractors fail.

Convert Local Files

uvx markitdown document.pdf
uvx markitdown presentation.pptx
uvx markitdown spreadsheet.xlsx
uvx markitdown file.docx

Supported Formats

• PDF — uvx --with 'markitdown[pdf]' markitdown file.pdf
• Word (.docx) — uvx --with 'markitdown[docx]' markitdown file.docx
• PowerPoint (.pptx) — uvx --with 'markitdown[pptx]' markitdown file.pptx
• Excel (.xlsx) — uvx --with 'markitdown[xlsx]' markitdown file.xlsx
• Excel (.xls) — uvx --with 'markitdown[xls]' markitdown file.xls
• HTML — supported by default
• Images — EXIF metadata extraction (add OCR via plugin)
• Audio — uvx --with 'markitdown[audio-transcription]' markitdown file.mp3
• YouTube URLs — see YouTube section below
• CSV, JSON, XML — supported by default
• ZIP files — iterates over contents
• EPub — supported by default

Use markitdown[all] to enable all format support:

uvx --with 'markitdown[all]' markitdown file.pdf

Save Output to File

uvx markitdown input.pdf > output.md
uvx markitdown input.pdf -o output.md

Pipe Content

cat file.pdf | uvx markitdown

YouTube Transcripts

markitdown YouTube support can be unreliable. Use the youtube-transcript-api directly instead:

uvx --with 'youtube-transcript-api' python -c "
from youtube_transcript_api import YouTubeTranscriptApi
ytt_api = YouTubeTranscriptApi()
transcript = ytt_api.fetch('VIDEO_ID')
for entry in transcript:
    print(entry.text)
"

Replace VIDEO_ID with the ID from the URL (e.g., 2JjKn7uhKqY from https://www.youtube.com/watch?v=2JjKn7uhKqY).

When to Use

• Reading web pages (especially JS-heavy sites that simpler extractors fail on)
• Converting PDFs, Office documents, or other files to Markdown for analysis
• Extracting text from images, audio, or YouTube videos
• Any task where you need file content as clean Markdown

When NOT to Use

• GitHub URLs — never use markitdown on github.com blob URLs. Convert to a raw URL and fetch directly instead:
  • Change https://github.com/user/repo/blob/branch/file.ts
  • To https://raw.githubusercontent.com/user/repo/branch/file.ts
  • Then use curl or the Read tool on the raw URL
• Plain text/source files reachable via a raw URL — fetch directly, no conversion needed