「用你自己的工具處理你自己的 repo」是最誠實的測試。以下全部輸出都收在 worked/dogfood/。
輸入已是 Markdown,輸出幾乎原封不動(16 KB)。這也展示 convert_uri 直接抓網址的能力:
markitdown "https://raw.githubusercontent.com/microsoft/markitdown/fd239d5d2be43d9b68329730206b9312c7d5a388/README.md" -o readme.md
輸出 16,115 bytes,內容與上游 README 一致(標題、徽章、表格都保留)。
TOML 是純文字 → PlainTextConverter,原文原樣保留(這正好讓你看清 extras 的真實定義):
[project] name = "markitdown" description = 'Utility tool for converting various files to Markdown' requires-python = ">=3.10" license = "MIT" [project.optional-dependencies] all = [ "python-pptx", "mammoth~=1.11.0", "pandas", ..., "SpeechRecognition", "youtube-transcript-api~=1.0.0", ... ] pdf = ["pdfminer.six>=20251230", "pdfplumber>=0.11.9"] docx = ["mammoth~=1.11.0", "lxml"] audio-transcription = ["pydub", "SpeechRecognition"] youtube-transcription = ["youtube-transcript-api"]
markitdown D:\temp\opencode\markitdown\packages\markitdown\pyproject.toml -o pyproject.md
純文字,原樣保留。
markitdown D:\temp\opencode\markitdown\Dockerfile -o dockerfile.md
輸出 717 bytes。
直接把 https://github.com/microsoft/markitdown 丟進去。GitHub 頁有超多導覽列,
而 GitHub 頁「沒有專用 converter」→ 走通用 HtmlConverter,於是一大堆導覽連結也被轉進來(26 KB)。
[Skip to content](#start-of-content)
## Navigation Menu
[Sign in](/login?...)
* Platform
+ AI CODE CREATION
- [GitHub CopilotWrite better code with AI](https://github.com/features/copilot)
+ DEVELOPER WORKFLOWS
- [ActionsAutomate any workflow](https://github.com/features/actions)
...
同樣是通用 HTML 轉換,導覽/警示區塊也進來了(16 KB)。
[Skip to main content](#content) Switch to mobile version Warning Some features may not work without JavaScript. Please try enabling it if you encounter problems. [](/) Search PyPI ...
Python 源碼 → PlainText,原樣保留(32 KB)。等於「用 MarkItDown 備份自己的源碼」。
markitdown D:\temp\opencode\markitdown\packages\markitdown\src\markitdown\_markitdown.py -o core-source.md