Free MD Viewer & md2rich: A Privacy-First Markdown Workflow

On August 31, 2026 a Show HN: Markdown Viewer and Editor by user beerglass climbed to 22 points with 11 comments by September 1, 2026. The repo is hattray/markdown-editor under MIT, by Kingsbridge Consultancy LLC; the live app runs at kingsbridge-consultancy.com/md-viewer. It is one self-contained ~3.6 MB index.html with every library and font embedded. You can save the page and run it from file:// with no network connection. The privacy claim is verifiable, not promised: open devtools, watch the network tab, and read the source with view-source.

md2rich, the tool behind this site, shares that privacy posture. Both run entirely in the browser. Both reject uploads. Both refuse accounts. The difference is where they sit in the writer's loop: Free MD Viewer handles the read-and-edit step, opening the .md files that AI tools hand you and saving them back to disk. md2rich handles the publish step, converting the same Markdown into the rich text that social platforms accept on paste. This post is the integration recipe: open the .md in Free MD Viewer, edit it locally, then load the same file into md2rich and paste the rich text into LinkedIn, X Articles, Medium, or Notion.

What Free MD Viewer actually does

The README and the built HTML's own license banner list the bundled libraries: marked 14.1.4 (MIT), DOMPurify 3.2.7 (Apache-2.0 / MPL-2.0), highlight.js 11.11.1 (BSD-3-Clause), KaTeX 0.16.22 (MIT), and Mermaid 11.12.0 (MIT). KaTeX's embedded .woff2 fonts are under SIL OFL 1.1. The app's own code is MIT. Everything is inlined into one HTML file so the page bundle is the entire deliverable.

Features that matter for a writer workflow:

• Open from disk, edit, save back to the original file. Chrome and Edge desktop expose the File System Access API, so dropping a .md on the page hands back a FileSystemFileHandle; Save writes straight to that handle instead of forcing a download. On other browsers Save falls back to a download, which still works.
• Installable PWA that registers file_handlers. Once installed, double-clicking a .md file in Explorer or Finder opens it directly in the viewer. The launch handle is preserved across the install, so Save still writes back to the file you opened.
• GFM Markdown: tables, task lists, strikethrough, fenced code with GitHub-style syntax highlighting. A Mermaid ```mermaid block renders in place; the theme follows dark and light toggles. LaTeX runs through KaTeX for $inline$, $$display$$, \(...\), and \[...\]; the README notes that math is extracted before Markdown parsing so underscores and asterisks inside TeX survive.
• GitHub callouts render as colored panels: > [!NOTE], > [!TIP], > [!IMPORTANT], > [!WARNING], > [!CAUTION]. This is the syntax GitHub uses for READMEs; if your AI tool generates Markdown with these markers it renders correctly here.
• Export to .html, .png, and PDF. All three are generated in-page with no upload. .html is a self-contained document with styles and KaTeX fonts embedded. .png rasterizes the preview through an SVG <foreignObject> at up to 2× device scale. PDF routes through the browser's own print-to-PDF pipeline so text stays selectable, not baked into an image. Per-block downloads hover a Mermaid diagram for an SVG button (real dimensions from the viewBox) or a table for a CSV button (RFC-style quoting plus a UTF-8 BOM so Excel reads accents correctly).
• DOMPurify sanitizes the rendered HTML. Even if a malicious source document tries to inject script, the rendered output is sanitized on the way to the preview. The README and the built license banner document this contract.
• Draft auto-restore writes to localStorage, so refresh or crash loses nothing. Keyboard shortcuts include ⌘S (save), ⌘⇧S (save as), and ⌘O (open). Reader controls let you tune text size and line width (the Aa button), persisted across sessions.

Two features depend on Chromium-desktop-only APIs and degrade gracefully elsewhere. Save-back-to-original-file uses the File System Access API; on other browsers it falls back to a download. Double-clicking a .md file from disk to open it in the viewer uses file_handlers and launchQueue in the manifest; on other browsers that path is simply not offered. Everything else — rendering, editing, exports, per-block downloads — works in any current browser.

What md2rich actually does

md2rich is also a single-page browser app with no backend, no upload, and no account. It loads [email protected] from jsDelivr, runs the same parser family as Free MD Viewer, and pipes the rendered HTML through a sanitizer, then paints the result into a contenteditable preview that mirrors how LinkedIn, X Articles, Medium, and Notion render the rich text you paste.

The publish-side problem is what Free MD Viewer does not solve: most social platforms do not accept raw Markdown pasted into their compose box. LinkedIn strips tables and task lists. X Articles strips anything not in its block grammar. Medium accepts Markdown in its editor mode but converts differently than its WYSIWYG paste expects. Notion loses headings on paste from an external source. Each platform has its own paste quirks, and md2rich encodes them as render modes so that what you see in the preview is what the platform will render after you paste.

Why these two tools belong together

Both are privacy-first. Both are zero-install (you can save the page and run it offline). Both refuse accounts and refuse uploads. The split is functional:

The pattern looks like this for a typical post:

1. An AI tool (Claude, ChatGPT, Cursor, OzBrain) hands you a .md file. Save it to disk.
2. Open it in Free MD Viewer. Read the output as a rendered document, not a wall of raw text. Edit the body, fix any platform-specific phrasing, and Save back to the original file.
3. Copy the polished Markdown into md2rich. Pick the target platform's render mode (LinkedIn / X Articles / Medium / Notion). Paste it into the compose box of the platform.
4. Files live on your machine the whole way. Nothing is uploaded.

Where Free MD Viewer falls short of an Obsidian or Typora

Free MD Viewer is excellent for the open-edit-save loop on a single file, but it is intentionally not a vault manager. There is no graph view of linked files, no plugin marketplace, no JSON-exported metadata table for queries. There is no vault-at-folder level cache; the app loads the file you point it at and forgets it on close. For a thousand-note workflow where you want to find every file that mentions a given tag or back-link across a folder, Obsidian's local-only vault model still wins.

Typora was the WYSIWYG-into-Markdown app that everyone used until it switched to a paid license. Free MD Viewer is not that. The Free MD Viewer experience is split-pane editor on top with live preview below; there is no inline WYSIWYG mode where formatting tools look like a word processor and Markdown appears under the hood. If you want that, Typora's paid license, iA Writer, or Zettlr's typewriter mode cover the use case that Free MD Viewer deliberately does not.

The other thing Free MD Viewer does not have is desktop-not-just-browser. Saving the page as a local file is the official self-hosting recipe and it works, but it is not the same as a windowed native app with a dock icon, system-wide keyboard shortcuts, and an OS-native File menu. Native macOS tools like Markdown Buddy and iA Writer give you that. The trade-off is installation friction in exchange for tighter OS integration.

A practical publish example

Suppose the AI tool generates this Markdown about a release:

# Q4 2026 Roadmap

We shipped the **Privacy Dashboard** this quarter. Next up:

- Webhook signing
- Audit log export
- SAML SSO for enterprise

Read the full announcement: <https://example.com/blog/privacy-dashboard>

Open that file in Free MD Viewer. The H2 renders as a heading, the bulleted list as a real list, the bold text as bold inline. You can scan the document as a rendered page, not as a wall of # and - characters. If a section needs a rewrite, edit in place and Save.

Then drop the same Markdown into md2rich. The preview shows exactly the rich text that LinkedIn or X Articles will render when you paste. The pipeline under the hood is the same as Free MD Viewer's: marked parses the Markdown, DOMPurify sanitizes it, the result becomes a contenteditable surface. But the surface is configured per-platform: LinkedIn mode optimizes for inline emphasis and bullet width, X Articles mode optimizes for the smaller character budgets, Medium mode keeps the rich-text metadata structure, Notion mode preserves headings as database-style blocks.

Why this split is the durable answer for privacy-first writers

For most of 2026 the privacy-first Markdown tool conversation has been about which editor or which renderer to choose, as if a single tool had to cover both jobs. The Free MD Viewer plus md2rich combination is the cleaner split: one open-and-edit tool that lives on your local disk and one publish-and-paste tool that lives in your browser tab. The Markdown source stays the only canonical artifact. Both tools refuse to upload it. Neither tool needs an account. Either tool works offline if you save the page locally first.

If you want to try the workflow, here is the order of operations:

1. Open Free MD Viewer in Chrome or Edge desktop and accept the install prompt. Double-clicking a .md file on your machine now opens it directly in the viewer.
2. Open hattray/markdown-editor if you want to inspect the bundled libraries or self-host the page on your own domain. The entire app is one HTML file.
3. Open md2rich in the same or a different window. Paste the polished Markdown and pick a target render mode. Copy the rich text out and paste it into LinkedIn, X Articles, Medium, or Notion.

Files never leave your machine. AI tools produce a .md; Free MD Viewer reads and edits it; md2rich renders it for the social platform. No tool in this pipeline has an upload endpoint, and the README for Free MD Viewer explicitly lists features that would need a server as out of scope by design. The same is true for md2rich's static page: no fetch, no telemetry, no analytics, no account. If your threat model cares about where the bytes go, this pipeline is checkable end-to-end from view-source.

Last verified 2026-09-01. Vendors and pricing subject to change.

FAQ

Want to try the publish step right now? Try md2rich — paste any Markdown, copy the rendered rich text, paste into LinkedIn, X Articles, Medium, or Notion.