Best Web Clipper in 2026 — After MarkDownload's Removal and Pocket's Shutdown
Best Web Clipper in 2026 — After MarkDownload's Removal and Pocket's Shutdown
In the span of six months in 2025, two of the most-used save-for-later tools disappeared from the landscape:
- MarkDownload — the de-facto standard browser extension for saving webpages as Markdown — was removed from the Chrome Web Store for "not following best practices" and has been unmaintained for over two years.
- Pocket — Mozilla's read-it-later service, with millions of users — was shut down on July 8, 2025, with all saved content access ending October 2025.
Together, these two events left somewhere between 5 and 10 million users without their preferred tool. If you're one of them, this article is the cheat sheet — what's left, what each remaining option does well, and what each one is bad at.
What changed in 2025
Before recommending replacements, it helps to understand why the landscape compressed.
MarkDownload's decline wasn't a single event. The original maintainer stopped shipping updates in 2023. The extension fell behind the Manifest V3 migration. Issues piled up: code blocks producing walls of <span> tags (#395), tables failing to convert (#298), MathJax dropping (#377), Brave compatibility breaking (#28). In late 2024, the Chrome Web Store removed it for non-compliance. As of 2026, it still works for users who installed it before removal, but new users can't get it and it won't be patched.
Pocket's shutdown was a cleaner narrative — Mozilla announced retirement in May 2025 and ended service in July. The reasons were institutional rather than technical, but the outcome is the same: the service is gone, and users had a 90-day window to export their data.
Both events accelerated a trend that was already underway: AI workflows replaced "save for later" as the dominant use case. People aren't saving articles to read later — they're saving them to feed to Claude, ChatGPT, or NotebookLM, or to clip into Obsidian for a knowledge base. The tools that survive the transition are the ones designed for this new shape.
What's left, honestly assessed
Obsidian Web Clipper (official)
The default recommendation if you live in Obsidian. It's free, it works, the templates system is genuinely powerful — you can configure custom selectors per domain.
Where it's weak:
- Silent data loss (#828) — cancel the "Open Obsidian?" dialog and your clip is just gone
- Reddit threads (#95276) — clipping stops at the first link
- Selection capture on Chrome and Edge (forum #106195) — captures the entire page instead of the highlight
- YouTube transcripts break every time the YouTube UI updates
- Obsidian-only — if you also use Notion, NotebookLM, or send to Claude, you need a second tool
Best for: Obsidian-centric workflows on stable content (blogs, Wikipedia, basic articles).
Web2MD
Disclosure: I work on this one. Built specifically for the post-MarkDownload, post-Pocket landscape — AI workflows first, multiple destinations supported.
What's there:
- 20+ site-specific extractors: Reddit (full thread + comments via API), arXiv (LaTeX preserved), Stack Overflow, Hacker News, YouTube, Twitter/X, GitHub, Notion, Substack, Medium
- Chinese platform extractors: Xiaohongshu, WeChat 公众号, Zhihu, Bilibili (none of the other clippers handle these)
- AI conversation export: ChatGPT, Claude, DeepSeek, Gemini, Perplexity
- Send to AI: one-click into a new Claude/ChatGPT/Gemini tab with the Markdown injected
- Save to Obsidian (with frontmatter), save to Notion (native API), export to NotebookLM
- Code block fidelity: strips Highlight.js / Prism / Shiki residue spans, keeps language tags
- LaTeX fidelity: KaTeX and MathJax converted back to TeX source
- Batch convert up to 50 URLs at once (Pro)
- MCP server for AI agents (Claude Code, Cursor)
Where it's weak:
- Free tier is 3 conversions/day. Heavy users need Pro at $9/month.
- Extension is closed source (MCP server is open source). If purity matters, this is a red flag.
- Browser-only. If you need server-side unattended scraping at scale, see Firecrawl or Crawl4AI.
Best for: AI-first workflows, multi-destination saves, login-walled or anti-bot-protected sources.
Save to Notion
If your knowledge base is Notion, the official Save to Notion extension is the path of least resistance. CWS rating is 4.3. Common complaints in negative reviews: forces a separate signup flow, hijacks Shift+Cmd+A with no rebind, can't paste images from clipboard.
Best for: Notion-only workflows, simple article saving.
MarkSnip
A newer entrant in 2024 that filled some of the post-MarkDownload gap. Lightweight, free, no signup. The Markdown output is decent for plain articles. Where it falls short: no AI integration, no batch convert, no specific extractors for Reddit/X/YouTube, and the Obsidian integration has the same silent-loss issue (reports on the GitHub repo). Best for: minimalist users who don't need anything beyond "page → Markdown file."
Raindrop.io
Not a Markdown clipper specifically, but worth mentioning since some Pocket users moved here. Strong bookmark organization, decent reading view, $3.54/month for Pro. Limitation: it's bookmark-first, not Markdown-first. If you want to feed clipped content into AI tools, you're back to copy-paste from Raindrop's reading view, with the same formatting-loss problems.
Readwise Reader
The premium option at $9.99/month. Excellent for reading, highlights, and integrating with the Readwise note-taking ecosystem. The Ghostreader feature embeds AI questions directly. Limitation: locks you into the Readwise ecosystem; doesn't produce portable Markdown by default; not really a Markdown-first tool — it's a reading-first tool with AI.
Best for: heavy reading workflows that don't care about portability.
Jina Reader
Not a clipper but often used as a manual workaround. Prepend r.jina.ai/ to any URL and get back Markdown. Server-side, so all the URL-paste limitations apply: subscribed Substacks, X, LinkedIn, paywalled content, and Chinese platforms all return broken or stub responses. Free tier rate-limits aggressively.
Best for: quick one-off conversions of fully public pages from a developer environment.
Decision tree
If you live in Obsidian and clip mostly stable, public content (blogs, Wikipedia): Obsidian Web Clipper is fine, just be aware of the silent-loss issue and have a backup.
If you save webpages specifically to feed into Claude / ChatGPT / Gemini: Web2MD's Send-to-AI flow is the closest thing to a one-click pipeline.
If your sources include Reddit, X, paywalled Substacks, LinkedIn articles, or Chinese platforms: Web2MD is currently the only browser-extension clipper that handles all of these because it runs inside your already-logged-in browser session.
If you need to clip into Notion specifically: Save to Notion if simple article saving is enough; Web2MD if you also want Markdown export and AI handoff.
If you need server-side unattended scraping for a RAG pipeline: Firecrawl or Crawl4AI, not a browser extension.
If you only need plain article-to-Markdown with no extras: MarkSnip is fine.
What about MarkDownload migration
If you have data in MarkDownload, the migration path is:
- Settings (auto-copy, image handling, frontmatter): reproduce manually in your new tool's settings.
- Custom rules / Readability selectors: paste them into your new tool's site adapter section.
- Saved files: nothing to migrate — MarkDownload was always a save-and-forget extension.
What about Pocket migration
If you exported your Pocket library before October 2025, you have a list of URLs. The fastest way to re-process them is batch-convert through a clipper that supports URL lists — Web2MD's Pro tier does up to 50 URLs at once via the MCP server, which works for moderate libraries. For larger libraries, Firecrawl's batch API is the right tool.
If you missed the export window, your Pocket library is unfortunately gone. There is no recovery path.
The pattern that survives
The web-clipper landscape is consolidating around tools that solve specific workflows rather than "general save-for-later." MarkDownload was a generic Markdown clipper and lost relevance as workflows specialized. Pocket was a generic save-for-later tool and lost relevance as AI workflows replaced "I'll read it later."
The tools that thrive in 2026 do one of three things well:
- Save into a specific knowledge graph (Obsidian, Notion, Readwise)
- Pipe content into AI tools (Web2MD's Send-to-AI, Readwise Ghostreader)
- Power programmatic agent workflows (Web2MD MCP, Firecrawl)
Pick the tool that maps to your workflow rather than the most popular one — that's the lesson the MarkDownload-and-Pocket exit teaches.
Related: