中文版 | English
A lightweight browser extension for capturing web page notes with LLM-powered analysis. Hover over any paragraph, press Alt+Click, and get AI-generated insights displayed in a non-intrusive card — without polluting the original page styles.
- Inline Annotations — Select any paragraph on a webpage and get instant AI analysis displayed directly below it
- Streaming Responses — Watch AI responses type out in real-time with SSE streaming support
- Multi-Provider Support — Works with OpenAI, DeepSeek, SiliconFlow, TogetherAI, Ollama, and any OpenAI-compatible API
- Per-Site Toggle — Enable or disable the extension on specific websites
- Think Content Filter — Toggle visibility of AI reasoning content (e.g.,
<think>blocks from models like DeepSeek) - Markdown Export — Export all notes from the current page as a formatted Markdown file
- Hover over any paragraph on a webpage
- Press
Altto see a highlight feedback - Alt+Click to capture the text and trigger AI analysis
- View the AI response in a collapsible card rendered below the paragraph
- Use
Escto dismiss the card - Click the extension icon to export all notes from the current page
- Node.js 18+
- npm 9+
# Install dependencies
npm install --legacy-peer-deps
# Build for Chrome
npm run build:chrome
# Build for Firefox
npm run build:firefoxChrome / Chromium-based browsers:
- Open
chrome://extensions/ - Enable Developer mode
- Click Load unpacked
- Select the
dist_chromedirectory
Firefox:
- Open
about:debugging#/runtime/this-firefox - Click Load Temporary Add-on
- Select
dist_firefox/manifest.json
- Click the InlineNote icon in your browser toolbar
- Click the Settings (gear) icon to open the Options page
- Add your API keys for supported providers
- Optionally customize prompt templates
| Provider | Default Model | API Format |
|---|---|---|
| OpenAI | gpt-3.5-turbo |
OpenAI |
| DeepSeek | deepseek-chat |
OpenAI-compatible |
| SiliconFlow | deepseek-chat |
OpenAI-compatible |
| TogetherAI | meta-llama/Llama-3-70b-chat-hf |
OpenAI-compatible |
| Ollama | llama3 |
OpenAI-compatible |
| Custom | User-defined | OpenAI-compatible |
| Shortcut | Action |
|---|---|
Alt (hold) |
Preview highlight on hovered element |
Alt+Click |
Capture text and create note |
Esc |
Dismiss active note card |
src/
├── pages/
│ ├── content/ # Content Script - DOM interaction, Shadow DOM rendering
│ ├── background/ # Service Worker - LLM API communication, storage
│ ├── popup/ # Popup UI - quick actions, notes overview
│ └── options/ # Options page - API keys, prompt templates
Communication Flow:
Content Script ──message──> Background Worker ──fetch──> LLM API
│ │
│ (streaming SSE)
│ │
Content Script <──message─── Background Worker <──────────┘
│ │
└── Shadow DOM card update ◄───┘
| Category | Technology |
|---|---|
| Build Tool | Vite 6 + @crxjs/vite-plugin |
| Framework | React 19 |
| Language | TypeScript 5 (strict mode) |
| Styling | Tailwind CSS 4 |
| Browser | Chrome / Firefox (Manifest V3) |
| Testing | Vitest + jsdom |
# Run tests
npm test # Watch mode
npm run test:run # Single run
npm run test:ui # UI mode
# Linting & Formatting
npm run lint # ESLint
npm run format # Prettier
# Type checking
npm run typecheckMIT
