Help

Into your own browser's local storage, under four keys: the document, the file name shown in the toolbar, the view you chose and the reading theme. Nothing else, and nothing anywhere else.

Clear in the toolbar removes it, and asks first if there is anything to lose. Clearing your browser's site data removes it too, along with everything else that site data holds.

In a private window, or with storage switched off, nothing is kept and nothing breaks: the viewer notices the write failed and carries on.

No, and the reason is the point of the site. Your document never reaches us, so there is no address where a copy of it lives and nothing for a link to point at.

Send the file itself. Download .md gives it back under the name it came in with, or as document.md if you pasted it rather than opened it.

The converters stop at 25 MB, which is far past any markdown file and about where a browser tab starts to suffer for a PDF or a spreadsheet.

Rendering is not the slow part: a megabyte of markdown renders in well under a second. Drawing thousands of results is, so lists like the checker's findings and the table of contents stop at five hundred and say how many they left out.

Loading the page. After that, rendering, converting, checking and editing all run in the tab, and nothing you open is sent anywhere.

Two pieces arrive later, the first time they are wanted: the syntax highlighter and the diagram library. Open a page with code or a mermaid diagram once while online and they are in the browser's cache after that.

There is no offline mode. This is a website, so a browser that cannot reach it cannot open it. Files you need to read without a network are what the apps are for.

To read or edit a file you already have, the viewer on the home page: drop it in and it renders, with the source beside it.

To end up with a .md without writing markdown, the builder. You type or paste into an ordinary editing surface and take the markdown away.

To find out why something renders wrong, the checker. To tidy a file without changing what it says, the formatter. To move between formats, the converters, one page each.

Yes. Open in viewer appears wherever a tool has produced markdown, and hands the result to the viewer with its name attached, so a converted PDF can be read, edited and downloaded without a trip through your disk in between.

It travels through your own browser's storage, like everything else here. Nothing is uploaded on the way.

Public repositories, and markdown files. It asks GitHub for the file list, reads up to fifty markdown files and says how many it skipped. A private repository would mean holding a token, which is the one thing this site will not do.

GitHub allows sixty unauthenticated requests an hour from one address. The file list costs one of those; the files themselves come from a different host that does not count against it.

In the viewer, undo and redo: Command or Control Z, and either Shift Z or Control Y, because Windows and macOS disagree about which one redo is. It keeps a hundred steps, and typing is grouped, so one undo takes back a phrase rather than a letter.

In the builder, the ones you already use: Command or Control B, I and K for bold, italics and a link. The markdown shortcuts work while you type as well, so a hash and a space makes a heading.

No, and the reason is the same one that makes the privacy claim true: every tool here runs inside the page, so there is no endpoint to call and nothing on a server doing the work.

For one repository, the checker reads it from GitHub in your browser. For anything larger or repeated, the honest answer is that a command-line tool on your own machine is the better instrument, and markdownlint and pandoc are where most people end up.

No, and there is nowhere for it to go: the page that renders your document is not allowed to make a network request at all. That is a capability, not an intention, which is why it is worth stating.

On the website you can check it yourself. Open your browser's network tab, load a document, and read the requests. None of them carries your file, because this site has no upload endpoint to carry it to.

The one exception is the repository checker, which asks GitHub about a public repository you named. That page says so above its input box.

Only when you save, and then it writes back to the same file, in the same folder, where your other tools can see it. Nothing is imported and no copy is kept.

On the website there is nowhere to save back to, so the editor is for reading, changing and taking away: use Download .md to keep your version.

No. It would mean storing a token, and the promise that nothing leaves this device is worth more than the feature.

Public repositories work everywhere, with no account and no sign-in.

That is deliberate, and the app says so above the document: “This is a reading of the page. Styling and scripts are not applied.”

If the page is nearly empty you will see a different line: “This page builds its content with scripts, so there is little to read here. The source view shows the file.” Which is the honest answer for a modern web application, whose HTML file contains almost nothing until a browser runs it.

A PDF has no structure. It has glyphs at coordinates in a font size, and every heading, list and paragraph is inferred from that. It gets most documents right and it will not get a complicated layout right, because the information simply is not in the file.

A Word file is better, because .docx does record what is a heading and what is a list. What it loses is anything markdown has no word for: text boxes, columns, comments and tracked changes.

Run it through the checker, on this website or in the app. It reports the mistakes that render wrong silently rather than loudly: a checkbox with no space between the brackets, a table missing its divider row, a fence that never closes and swallows the rest of the file, a link with a space in the wrong place. Checking a whole folder at once does the same thing across a directory, and looks at the markdown files only: markdown rules applied to a .swift would cry wolf on every line.

If the document is clean and it still looks wrong somewhere else, it is probably not a mistake but a difference between renderers. The compatibility table has every construct against seven platforms.

Print it. Command-P, then Save as PDF from the PDF dropdown at the bottom of the print panel. The document goes onto white paper with the reading chrome stripped.

The same two steps work in a browser on this website, and print the document rather than the page around it.

Print it. Export, then Print, then tap the share icon in the corner of the print sheet and choose Save to Files. There is no Save as PDF button on a phone, which catches people out.

In a browser on this website it is the same gesture: the share icon on the print sheet, then Save to Files.

Because nothing in the file says where a page should end. Markdown has no page concept at all, so the breaks fall where the printer puts them.

The website prints with a stylesheet that puts the document on the page and nothing else, so what you get is the document rather than a screenshot of a website. Where a break lands inside it is the browser's decision.

On this website: out to Word as a real .docx with proper heading styles, and to HTML, PDF, RTF and plain text. In from PDF, Word, Excel, CSV, JSON, HTML and RTF.

The app converts less, on purpose, and differently in one place. Out: markdown, HTML, Word and PDF through the print panel. Its Word export is a .doc rather than a .docx, which is HTML that Word opens, and the honest extension for it. In: PDF, Word and Excel. A CSV, a JSON or an HTML file is not converted in the app at all: it is opened and shown as what it is.

Either way it runs on your own machine.