Feature request: browser extension, save the page or the selection as Markdown

[virtuarsys]

Today the extension saves a note with the link and a screenshot. The note therefore contains almost no text: full-text search can’t find it, and automated processing (summaries, tagging, indexing with an LLM through the API or MCP) has nothing to work on. Pages behind a login or a bot protection can’t be re-fetched later either.

Proposal: add an action next to “Save link” and “Capture”, for example “Save as Markdown”, that converts the whole page, or only the current selection, to Markdown in the browser and stores it in the note body together with the URL and the date.

Reference: MarkDownload (by deathau, https://github.com/deathau/markdownload) and its Manifest V3 fork MarkSnip (https://github.com/DhruvParikh1/markdownload-extension-updated) already offer this, for the whole page or for the selection only. I mention them only as a reference for the expected behavior. Both rely on two open-source libraries that could be used directly: Readability.js by Mozilla for main content extraction (Apache 2.0) and Turndown for HTML to Markdown conversion (MIT).

Benefits: notes become searchable by content, the content is preserved even if the page disappears, it works for pages that need authentication, and automation tools using the Poznote API or MCP can process the text directly.

[timothepoznanski]

That's a very good idea thanks! Working on it.

[pythoninja]

Please, add the "Save selected text" feature, would very useful.

[timothepoznanski]

I'll see what I can do.

[timothepoznanski]

@virtuarsys It's done in the extension (version 2.3.0): a third button in the popup saves the main text of the page as Markdown, using Readability.js and Turndown as you suggested, below the page address and the date it was saved.

When some text is selected, the button becomes "Save selection" and saves only that part, @pythoninja. A new note created this way is a Markdown note, and when you add the page to an existing note the text is written in that note's format. Since the text is read in your browser, it also works for pages behind a login, and Poznote's search, the API and the MCP server can work on it. Images remain links to the original site for now.