Web2Markdown / Features
Web2Markdown features: the Markdown Chrome extension and the editor.
Web2Markdown Capture is the Chrome extension. The Web2Markdown desktop editor is the Windows application. This page lists what each one does, verified against the current builds.
What Web2Markdown does
Web2Markdown is a Chrome capture extension and a Windows desktop editor from GDK Digital LLC. Web2Markdown Capture turns a web page, a highlighted selection, a bounded site tree, or a sitemap into Markdown files on your disk. The Web2Markdown desktop editor opens Markdown and 124 other text file types, edits Markdown with a synchronized preview, and exports finished documents to PDF, DOCX, HTML, or plain text.
Both components work without a product account. Captures are ordinary files with a relative assets folder and a capture.json receipt, so nothing on this page depends on the desktop editor being installed. Describes desktop 2.8.5 and extension 1.7.3. Windows installation checks have passed. Chrome testing and everyday workflow review remain in progress. Release availability is on the download page.
Capture: the Chrome extension
Web2Markdown Capture is a Manifest V3 Chrome extension. It converts readable page content to Markdown locally, keeps the source URL and metadata with the file, and saves where you tell it to.
| Capture mode | What it does |
|---|---|
| Single Page | Extracts the readable content of the current page as Markdown with source metadata. |
| HTML snapshot | Single Page can save passive rendered HTML with selected safe styles and local images. Scripts, forms, frames, media, shadow DOM and external stylesheets are omitted. This is not original server source or a complete offline archive. |
| Capture Element | Select a webpage element and save a portable artifact JSON with selected computed styles and supported embedded images. Review omission warnings before saving; reusable appearance does not include website behavior. |
| Selection | Captures only the passage you highlighted before opening the popup or side panel. |
| Site Tree | Follows same-origin links with a depth limit, a 2 to 50 page cap, optional regular-expression path filters, standard wildcard robots.txt rules, and 600 ms pacing between pages. Sitemap-assisted discovery can add same-site URLs within the same limits. |
| Sitemap | Saves readable sitemap documents, a deterministic URL inventory, and a receipt without opening the listed pages: 1 MB per file, 10 files, two index levels, 2,000 URLs. |
| All open tabs | Batch-captures every open tab as Markdown files. |
| Copy to clipboard | Copies the extracted Markdown without downloading a file. |
| Right-click menu | Clip page, clip selection, copy page, or copy selection from the page context menu. |
Before the file is saved
| Feature | Detail |
|---|---|
| Preview and inline edit | See and edit the Markdown in the popup or the persistent side panel before saving. One shared local draft (24 hours, 50 files, 2 MiB) with revision checks and explicit conflict handling. |
| Wider capture workspace | A 680-pixel toolbar popup and Wide Workspace in a full browser tab keep the intended source page selected. Chrome controls side-panel width; narrow layouts remain responsive. |
| Heading tree and splitting | Preserves the H1 to H6 hierarchy, generates a table of contents, and can split a long page by H2 or H3 into several files. |
| YAML frontmatter | Source URL, title, capture date, author when available, and heading warnings. |
| Junk and noise stripping | Removes ads, navigation, sidebars, comments, cookie and GDPR banners, and floating chat widgets. Strips analytics tracking query parameters (utm_*, fbclid, gclid) and copy buttons/gutters from code blocks before Markdown conversion. |
| LaTeX math formulas | Preserves mathematical equations from KaTeX, MathJax, and Wikipedia fallback images into standard inline and display math syntax. |
| Code syntax preservation | Preserves language classes on fenced code blocks without stripping syntax highlighting identifiers. |
| SPA and docs fallback | Automatically falls back to semantic content tags when client-side single-page applications or documentation sites render outside standard article containers. |
| High-resolution images | Promotes lazy-loaded data-src and srcset attributes over placeholder images so real media is captured. |
| Local image assets | Downloads images into a relative assets folder before the Markdown is published; failed images keep their remote URL. Remote-image mode leaves links to the source site instead. |
| Site Profiles | Match bounded host or URL patterns to a default capture mode, include and exclude selectors, safe folder and filename templates, and extra frontmatter. |
| Capture receipts | Canonical source URLs, stable capture IDs, generated-content boundaries, evidence counts, and a final capture.json in each operation folder. |
| Capture folders | Downloads/Web2Markdown by default, or registered direct folders on any user-accessible drive with 1-click quick switching and permission renewal. If Chrome cannot use the chosen folder, the capture falls back to Downloads and reports the actual destination. |
| Recovery | Bounded Site Tree job history (30 jobs, 7 days, 6 MiB) with manual Resume, Retry failed, and Save partial actions. |
| Page evidence (optional) | page-evidence.json and seo-summary.md with bounded rendered-page metadata, headings, links, images, and JSON-LD inventory. No SEO score, indexing verdict, HTTP audit, or performance result is claimed. |
| Agent bridge (optional) | A trusted local CLI can request an active-tab capture through Chrome native messaging. Opt-in only; no arbitrary URLs and no page content over IPC. |
| Help, About, diagnostics | Searchable local Help, a complete Settings reference with field-level explanations, privacy-safe diagnostics, and a manual update check. |
| Permissions | activeTab, storage, scripting, downloads, contextMenus, tabs, clipboardWrite, notifications, and sidePanel. nativeMessaging is optional and only used when you enable the bridge. |
Edit: the Windows desktop editor
The Web2Markdown desktop editor is an Electron application for Windows. It offers Markdown and HTML previews plus source editing for 125 text file types. Ordinary editing is local; optional AI tools depend on your chosen provider.
Files, tabs, and workspaces
| Feature | Detail |
|---|---|
| Multi-tab editing | Open Markdown, JSON, YAML, TOML, XML, CSV, configuration, environment, web, script, source, build, and documentation files together. Likely binary files are rejected before they reach the editor. |
| Two editor groups | Move a tab into a second source editor with one owner per document, independent selection and undo history, and vertical or horizontal orientation. |
| Tab continuity | Pin tabs, reopen recently closed tabs with their unsaved content and scroll state, and use close, close others, close right, close all, copy path, and reveal in folder. |
| Workspaces and sessions | Named Workspaces, session restore of every tab on the next launch, Quick Open (Ctrl+P) for recent files, and a command palette (Ctrl+K). |
| File tree sidebar | Browse the opened folder and open files with a click. |
| Capture Inbox | Watch up to eight capture folders, track new and changed captures as unread, group Markdown captures and sitemap evidence bundles, and review duplicates, receipts, provenance, warnings, and asset health. |
| Sitemap Viewer | Open a sitemap inventory or a captured sitemap file as a searchable URL table and compare it with same-origin links from local Markdown captures without visiting any page. |
| Artifact Library | Import a captured artifact; search name, tags or source; edit text, colors, fonts, spacing or HTML. Save edits separately from the original, duplicate, restore, remove, export self-contained HTML or insert into an HTML document. Local limits: 50 items, 2 MiB per artifact and 20 MiB total storage. |
| HTML visual tools | Sandboxed HTML preview with source reveal, visual element editing, Desktop/Tablet/Mobile width views, component insertion, explicit HTML/Markdown conversion and offline email CSS inlining. |
| AI HTML Assistant (optional) | Bring your own LLM provider or CLI configuration. Review generated changes before applying them. Provider or CLI processing may transmit document content and require a network connection; ordinary capture, editing and the Artifact Library do not require AI. |
Writing Markdown
| Feature | Detail |
|---|---|
| Source and preview | Editor, split, and preview layouts with a resizable divider. Synchronized preview scrolling keeps editor and preview aligned. Preview block editing and task toggles update the underlying Markdown as one undoable change. |
| Outline and navigation | A filterable H1 to H6 outline, previous and next heading commands, and a folder, file, and heading breadcrumb. |
| Formatting | Heading and List split buttons that remember your last style, selection-aware transforms between bullet, numbered, and task lists, links, images, tables, and code blocks. |
| Templates and snippets | Built-in starters for Blank Markdown, Blank Text, SKILL.md, AGENTS.md, CLAUDE.md, a path-scoped Claude rule, and .env.example, plus up to 50 local snippets with 20 explicit variables and JSON import and export. |
| Pasted images | Validated PNG, JPEG, GIF, and WebP data is saved into a sibling assets folder and inserted as a relative Markdown path. |
| Rich-text paste | Clipboard HTML is converted to GitHub-flavored Markdown locally with a preview-first review before insertion. |
| Focus modes | Distraction-free mode, Centered Writing Measure ("Zen Mode"), typewriter scrolling with a configurable caret target, optional active-paragraph focus, and scroll-past-end. |
Search, replace, and batch changes
| Feature | Detail |
|---|---|
| Find and replace | Per-tab find and replace with regular expressions and multiple selections. |
| Search Files | Ctrl+Shift+F searches open tabs and every supported text file in the opened folder. Literal, read-only, cancelable, and bounded. |
| Preview-first folder replacement | Review grouped before-and-after matches, accept whole files or single occurrences, then apply with local snapshots, atomic writes, receipts, and one-command rollback. |
| Batch metadata | Filter Markdown files by path, source domain, capture kind, tag, or property, preview one add, rename, remove, or normalization action, and apply it without rewriting unrelated frontmatter or bodies. |
| Document Check | Format-aware findings for JSON, JSON Lines, NDJSON, XML, YAML, TOML, dotenv, CSV, TSV, and Markdown frontmatter and code fences, with previewable single-undo fixes. |
Safety and recovery
| Feature | Detail |
|---|---|
| Conflict-safe atomic saves | External changes are detected before a file is replaced, with a comparison of the editor and disk versions. |
| File history | Prior disk versions are snapshotted before overwrite, and unsaved session content is recovered after a restart. |
| Optional auto-save | Save modified files every 30 seconds when enabled. Explicit save is the default. |
| Lossless text | Existing LF, CRLF, or CR line endings are preserved when saving, and UTF-8 text is not re-encoded. |
| Large-file mode | Core editing, saving, search, recovery, and history stay available while counts, checks, outline, highlighting, and preview pause at documented limits. |
| Interface | Light, dark, and system themes with instant toolbar toggle, Compact, Cozy, and Writing density, live document stats pill, keyboard-first controls with visible focus, and status announcements. |
| Windows integration | Registers .md and .markdown as a Markdown editor file association. The Windows build is a per-user installer with x64, 32-bit, and ARM64 builds in one package and requires Windows 10 or newer. |
Export Markdown to PDF, DOCX, HTML, or plain text
Export is available for Markdown documents. It never replaces the source file.
| Format | How it is produced |
|---|---|
| The sanitized rich renderer, so the export matches the preview. | |
| DOCX | Word-compatible output. Embeds BMP, GIF, JPEG, and PNG images only; remote or unsupported images become local alt-text placeholders, and the export performs no remote fetch. |
| HTML | The same sanitized rich renderer as PDF. |
| Plain text | Semantic token conversion that keeps code, links, images, nested lists, tables, and Unicode readable. |
| Repeatable exports | Successful per-file targets are remembered with exact overwrite confirmation. Bounded local export recipes, portable filename templates, batch progress, cancellation, and failed-file retry keep recurring output predictable. |
Save As is different from export: it changes the filename and editor mode for any of the supported text formats without converting the content.
What Web2Markdown does not do
- No cloud sync, no product account, and no document service. Files live where you save them.
- Page evidence and sitemap views are factual inventories. They are not SEO scores, indexing verdicts, HTTP audits, or performance results.
- Save As does not convert between languages or configuration formats.
- Capturing a page you can see while signed in does not grant permission to redistribute it.
- There is no Mac editor yet, and the Chrome Web Store listing and public source repository are still in preparation.
Frequently asked questions
Does Web2Markdown need an account or a cloud service?
No. Web2Markdown works with ordinary local files. The Chrome extension saves to a folder you choose and the desktop editor opens files from disk. There is no GDK product account and no cloud document service, and diagnostics stay on your computer unless you choose to send them.
Can I convert a web page to Markdown without the desktop app?
Yes. The Chrome extension captures a page, a selection, a site tree, or a sitemap as Markdown files on its own. The desktop editor is optional, and captured Markdown opens in any editor that reads files from disk.
Which export formats does the editor support?
Markdown documents export to PDF, DOCX, HTML, and plain text. PDF and HTML use the sanitized rich renderer. DOCX is Word-compatible and embeds BMP, GIF, JPEG, and PNG images only. Plain text keeps code, links, images, nested lists, tables, and Unicode.
Is Web2Markdown open source?
An open-source release is planned. The public repository is not available yet. The download page shows the current release status for the Windows editor, the Chrome extension, and the planned Mac editor.
Does Web2Markdown work on Mac?
Not yet. The desktop editor is a Windows application today. A Mac release is planned and will not be advertised until it has been built, signed, and verified on macOS.