markitdown

将网站和 URL 转换为适合大语言模型的 Markdown 格式。在阅读网页（标题、列表、表格、链接）、YouTube URL（字幕文本）以及 PDF 文件（包括通过 -x 或 -m 提示从标准输入读取的内容）时，可借此节省 Token 消耗。若 `markitdown` 执行失败，则回退至 `curl r.jina.ai/URL`。

SKILL.md

--- frontmatter

name: markitdown
description: Convert websites and URLs to LLM ready Markdown. Use to save tokens when reading websites (headings, lists, tables, links), YouTube URLs (transcripts) and PDFs (including stdin with -x or -m hints). If `markitdown` fails, fall back to `curl r.jina.ai/URL`.

MarkItDown

Installed extras in this environment: markitdown[pdf,youtube-transcription].

•Convert websites and URLs.

Web page to Markdown:

bash

markitdown 'https://example.com' -o page.md

If MarkItDown fails on a URL, fall back to r.jina.ai:

bash

curl -fsSL 'https://r.jina.ai/https://example.com' > page.md

YouTube URL to transcript Markdown:

bash

markitdown 'https://www.youtube.com/watch?v=VIDEO_ID' -o video.md

•Convert PDFs.

Convert a file (stdout):

bash

markitdown path/to/file.pdf

Write to a file:

bash

markitdown path/to/file.pdf -o output.md

If reading from stdin, set hints so MarkItDown picks the PDF converter:

bash

cat file | markitdown -x pdf
cat file | markitdown -m application/pdf