English | 中文
A browser extension based on the LLM Wiki concept that clips web pages to Obsidian as Markdown. Supports Chrome / Edge (Chromium).
Extends Karpathy's LLM Wiki "capture-compile-accumulate" workflow into a complete personal knowledge capture system:
- raw/ — Save original web content (immutable source)
- wiki/ — Compiled, structured knowledge articles
- Template System — Customize output format
- 🎯 Smart content extraction (Mozilla Readability algorithm)
- 📝 HTML → Markdown conversion (Turndown)
- 🏷️ Auto-extract tags and metadata
- 📁 Save directly to Obsidian Vault via File System Access API, or to browser downloads
- 🎨 Customizable templates
- 🛡️ XSS protection with DOMPurify
- 💾 Image localization support
- Open Chrome/Edge and navigate to
chrome://extensions/oredge://extensions/ - Enable "Developer mode" (top right)
- Click "Load unpacked"
- Select the
obsidian-web-clipperfolder
- Click the extension icon, then click ⚙️ Settings
- Click "Select Vault Directory" and choose your Obsidian Vault folder
- Configure default save folder and template
- Click "Save Settings"
Note: If no Vault directory is selected, files will be saved to the browser's default download directory.
- Browse any webpage
- Click the extension icon
- Preview the extracted content
- Choose save location and template
- Add tags (optional)
- Click "Save to Obsidian"
obsidian-web-clipper/
├── manifest.json # Extension manifest
├── background/
│ └── service-worker.js # Service worker (MV3)
├── content-script/
│ └── content-collector.js # Content extraction
├── popup/
│ ├── popup.html # Popup window
│ ├── popup.css
│ └── popup.js
├── options/
│ ├── options.html # Settings page
│ ├── options.css
│ └── options.js
├── templates/ # Note templates
│ ├── raw-template.md
│ ├── article-template.md
│ └── archive-template.md
├── lib/ # Third-party libraries (browser bundle)
├── patches/ # Documentation for lib modifications
├── tests/ # Test files
└── icons/
├── icon-48.png
└── icon-128.png
| Variable | Description |
|---|---|
{{title}} |
Article title |
{{url}} |
Original URL |
{{date}} |
Capture date |
{{author}} |
Author |
{{content}} |
Body content |
{{excerpt}} |
Excerpt/summary |
{{tags}} |
Tag list |
{{siteName}} |
Site name |
- Ensure the vault path has write permissions
- Windows: Run browser as administrator
- Linux/Mac: Check folder permissions
- Some sites have anti-scraping measures and may not extract completely
- Use the fallback copy-to-clipboard option
- Manifest V3 (MV3)
- Service Worker
- Content Script
- File System Access API
- IndexedDB
# No build step required — load as unpacked extension directlynpm testMIT