Paste an article into a notes app and the headings become body text, code blocks break, tables collapse and images still point to the original site. The page looked fine; the archive did not survive.
The clipboard stores appearance. Markdown stores structure. Without a real conversion between them, the result becomes hard to search or move.
What plain copy and paste usually loses
- Heading levels become body text.
- List indentation, quotations and code language disappear.
- Tables turn into tabs or spaces.
- Images remain remote links and break later.
- Recommendation cards and ads enter the document.
Clipboard copy is fine for temporary reading. A long-term reference library needs the structure to be fixed during export.
Platforms fail in different ways
| Platform | Common issue |
|---|---|
| Hotlink-protected images and recommendation cards | |
| Zhihu | Dynamic sections and folded answers |
| Juejin / SSPAI | Important code blocks and language markers |
| CSDN / Jianshu | Ads and recommendations mixed into the body |
| Xiaohongshu | Image-heavy posts with short captions |
| X and social threads | Dynamic loading and threaded order |
No single clipboard rule handles every site. Platform-aware extraction identifies the article boundary and removes unrelated modules.
Three ways to save the page
A platform-provided Markdown route
Some services expose a plain-text or Markdown endpoint. It works when available, but may cover only part of the page and leave images remote.
A script or command-line tool
Developers can fetch the article, handle cookies and rewrite images. It is flexible, but not a dependable everyday path for most users.
A browser extension
An extension reads the page already open in your browser and converts headings, lists, code, tables and image references into Markdown. Feishu Toolkit has cleanup rules for common platforms and a general extractor for other pages.
Web-to-Markdown is free. When images must be packaged locally, the Pro image ZIP workflow is available.
Images create the biggest hidden risk
Remote image URLs can expire, require a referrer, or depend on the original login. A durable archive should download the images and use relative paths.
Recommended structure:
article.md
assets/
001-cover.jpg
002-chart.pngNaming images in article order makes it easier to reconnect them with the text.
A five-point quality check
- Confirm H1 to H3 levels.
- Copy a code sample and check formatting.
- Compare table columns.
- Move the folder and confirm images still load.
- Search for a word from the middle of the article to prove the whole body was captured.
What Markdown enables
Markdown works well in Git, note-taking tools and AI context. It is compact, searchable and easy to process in batches.
Keep the source title, author, publication date and URL at the top so the archive remains verifiable.
What not to save
Do not export paid, members-only or explicitly restricted material just because the browser can display it. Access control and copy restrictions are different things, and a conversion tool should not bypass real permissions.
Keep the small set of useful articles in a form you can search and move. Capture volume is the wrong metric.