Fetch Markdown

Convert any web page to clean Markdown. Smart site detection, fallback extraction, and code-aware output for production use.

Fetch Markdown turns any web page into clean, structured Markdown — not by blindly stripping HTML, but by understanding what kind of page it's looking at. It detects 25+ site types including code repositories, Q&A forums, social media, wikis, documentation, news, academic content, and more — then applies a tailored extraction strategy for each. When no site-specific extractor matches, a Readability-style content-density scorer finds the main content automatically. Converting pages to Markdown has always been brittle — one site's quirks would break another site's rules. Fetch Markdown solves this with a confidence-scored detection engine that scores 25+ candidates in parallel. Multiple weak signals combine to correctly identify sites it's never seen before, and low-confidence detections fall through gracefully to the generic fallback. The result: reliable conversion from thousands of sites, not just a handful. Key features - Smart site detection — Automatically identifies the page type and applies the best extraction strategy for each category. Supports code hosting, social media, Q&A forums, wikis, documentation, news, academic papers, AI chat interfaces, and more. - Code-aware processing — Punctuation cleanup, whitespace normalization, and heading hierarchy fixes are applied only to non-code segments. Fenced code blocks and inline code spans are split out, left untouched, then reassembled. Your code samples come out exactly as they appeared on the page. - Lazy-load image unwrapping — Many modern sites hide real image URLs behind data-src, data-original, or data-lazy-src attributes. Fetch Markdown unwraps these so the actual image URL ends up in your Markdown. - Heading hierarchy normalization — If a document jumps from H1 to H3 (skipping H2), heading levels are remapped to a contiguous 1..N hierarchy so your Markdown structure makes sense. - Tracking-param stripping — Common tracking parameters (utm_source, utm_medium, fbclid, gclid, etc.) are stripped from link URLs automatically. - YAML front matter — Optionally prepend metadata (title, URL, date, author, description, site type) as YAML front matter. Customize the template with any placeholders you need. - Multiple ways to trigger — Click the toolbar icon, right-click any page/selection/link/image, or use keyboard shortcuts (Ctrl+Shift+M for pages, Ctrl+Shift+S for selections). Customize shortcuts at chrome://extensions/shortcuts. - Rich popup editor — After conversion, view character/word/line/heading stats, preview the rendered Markdown, edit the output, copy to clipboard, or download as a .md file — all from the popup. - Conversion history — The last 50 conversions are saved with quick copy and download per entry. - Right-click integrations — Right-click any page to convert or copy as Markdown. Right-click a selection to convert just that text. Right-click a link to copy as [text](url). Right-click an image to copy as . - Settings & customization — Toggle title headings, YAML front matter, links, images, fenced code blocks, and smart extraction on/off. Add custom include/exclude CSS selectors to fine-tune extraction on specific sites. Customize the YAML front matter template with any placeholders. - Dark mode — Built-in light and dark themes, toggled from the popup. - Privacy-first — No data is sent anywhere. Everything runs locally in your browser. No accounts, no telemetry, no backend. How it works 1. You trigger conversion via popup, context menu, or keyboard shortcut. 2. The site detector scores candidate types in parallel using host, URL, meta tags, and DOM probes. 3. The highest-scoring extraction strategy selects the main content area — site-specific for known types, Readability-style density scoring for everything else. 4. Noise is stripped: nav bars, ads, cookie banners, paywalls, modals, popups, and hidden elements. 5. Lazy-load image attributes are unwrapped, tracking params are stripped. 6. HTML is converted to Markdown with GFM plugins and custom rules for kbd, mark, sub, sup, abbr, details, figure, and more. 7. Code-aware post-processing cleans punctuation and whitespace without touching code blocks. 8. Optional YAML front matter is prepended. 9. Output appears in the popup editor with stats, ready to copy or download. Site types detected Code hosting (GitHub, GitLab, Bitbucket), Q&A forums (Stack Overflow, Discourse), social media (Reddit, Twitter/X, LinkedIn, YouTube), wikis (Wikipedia, Notion, Confluence), publisher content (blogs, news, documentation, product pages, academic papers), communities (forums, Hacker News), webmail, AI chat interfaces (ChatGPT, Claude), and general web pages. What's changed in v4.0 The detection, extraction, and conversion engine was rewritten from scratch for production reliability. Key improvements: confidence-scored parallel site detection (no more brittle if/else chains), code-aware post-processing (code blocks are never mangled), heading hierarchy normalization, lazy-load image unwrapping, tracking-param stripping, extended inline-element rules (kbd, mark, sub, sup, abbr, details, figure), proper YAML 1.2 escaping, and structured error handling with graceful fallbacks. Tech - Built on Turndown (HTML to Markdown) with GFM plugins - Marked for live Markdown preview - Chrome Manifest V3 - chrome.storage.local for settings and history