LightUp — AI annotations without leaving the page

LightUp lets you highlight any text and get instant AI explanations, summaries, translations, and more—right on the page. No tab switching, no broken flow.

• 🧠 Smart analysis: Explain, summarize, analyze, translate, or ask anything
• 🔒 Privacy-first: Optional local LLM support; session-only memory
• 🎨 Flexible UI: Floating, sidebar, or centered layouts
• ⚡ Fast: Keyboard shortcuts and instant follow-up questions
• 🌐 Multi-provider: OpenAI, Gemini, xAI/Grok, or local models
• 🛠️ Custom prompts: Create your own actions and workflows

---

✅ Key features

• Smart text analysis: Explain, summarize, analyze, translate, or ask anything.
• Multiple layouts: Floating, sidebar, or centered modal UI.
• Follow-up questions: Keep the context and dig deeper.
• Flexible AI backends: OpenAI, Gemini, xAI/Grok, or local LLMs.
• Keyboard shortcuts + theming: Dark/light/system theme and fast actions.
• Privacy-first: Session-only memory; optional local processing.

📥 Install (users)

1. Chrome Web Store: Install LightUp
2. GitHub Releases: Download the latest release
3. Manual install (if you downloaded a ZIP):
   • Open chrome://extensions/
   • Enable Developer mode
   • Click Load unpacked and select the unzipped folder

⚙️ Configure

1. Pin LightUp in your browser toolbar.
2. Right‑click the LightUp icon → Options.
3. Choose your provider:
   • Local LLM (requires a local server URL)
   • OpenAI (API key)
   • Google Gemini (API key + model)
   • xAI/Grok (API key)

⚠️ Using cloud providers sends content to their APIs. Local LLM keeps processing on your machine.

Supported providers

• OpenAI: GPT-4, GPT-4o, GPT-5 (requires API key)
• Google Gemini: Gemini 1.0 Pro, Gemini 1.5 Pro, Gemini 1.5 Flash (requires API key)
• xAI/Grok: Grok, Grok-2 (requires API key starting with xai-)
• Local LLM: Ollama, LM Studio, Text Generation WebUI via http://127.0.0.1:1234 or custom URL
• Basic: Free tier (no API key) — currently uses Grok

Local LLM setup

• Run a local server exposing POST /v1/chat/completions
• Set the server URL in Options → Local LLM
• Example URLs:
  • Ollama: http://127.0.0.1:11434
  • LM Studio: http://127.0.0.1:1234

Supported local models

LightUp supports a wide range of local models, including:

• Llama: Llama 4 (70B/40B), Llama 3.3 (70B/8B), Llama 3.2 (3B/1B), Llama 3.1 (405B/70B/8B), Llama 2 (70B/13B)
• DeepSeek: DeepSeek R1 series, DeepSeek V3, DeepSeek Coder series
• Qwen: Qwen 3 (32B/14B), Qwen 3 Coder
• Mistral: Mistral Large 3, Mixtral (8x22B/8x7B), Codestral 22B, Mistral 7B
• Phi: Phi 4 series (14B/3.8B), Phi 3 series, Phi 4 Reasoning
• Gemma: Gemma 3 (27B/9B/4B), Gemma 2 (27B/9B)
• Other: Neural Chat, OpenThinker, QWQ

Models are categorized by hardware requirements (from 2GB VRAM to 400GB+ VRAM) and use case (general, reasoning, coding, compact).

UI layouts

• Floating: Compact popup near your selection
• Sidebar: Persistent sidebar (access via right edge)
• Centered: Modal overlay with blurred background

Privacy details

• Session-only memory: Conversations are kept in memory only and cleared when you close the popup or navigate away. No history is saved to disk.
• Local processing option: Use a local LLM to keep all text processing on your machine—nothing leaves your device.
• Encrypted storage: API keys are encrypted in browser storage.
• No tracking: LightUp does not track users or send usage data to third parties.
• Cloud providers: If you use OpenAI, Gemini, or xAI/Grok, your selected text is sent to their APIs for processing. We recommend reviewing their privacy policies.
• Open source: All code is public on GitHub for transparency.

See PRIVACY.md for a deeper technical breakdown.

🔧 Usage

1. Select any text on a webpage.
2. LightUp appears with your selected mode.
3. Ask follow‑up questions for deeper context.

Modes

• Explain: Detailed explanations
• Summarize: Quick summaries
• Analyze: Deep analysis
• Translate: Translate to another language
• Free: Ask anything

⌨️ Keyboard shortcuts

• Ctrl+Shift+Z: Explain
• Ctrl+Shift+S: Summarize
• Ctrl+Shift+A: Analyze
• Ctrl+Shift+T: Translate
• Ctrl+Shift+F / Command+Shift+F: Free mode popup
• Ctrl+Shift+X: Toggle LightUp on/off
• Ctrl+Shift+R: Toggle Radically Focus mode
• Ctrl+Shift+D: Toggle theme

🧭 How it works (high level)

Selection → Content Script → Background Service Worker → UnifiedAIService → Provider → UI

Key entry points:

• src/contents/index.tsx (content script UI)
• src/background/index.ts (service worker + routing)
• src/services/llm/UnifiedAIService.ts (provider selection + streaming)

🛠️ Developer quickstart

pnpm install
pnpm dev

Load the dev build from build/chrome-mv3-dev in chrome://extensions/.

📚 Documentation

Start here: docs/README.md

• Architecture
• Development Guide
• Configuration & Providers
• Privacy & Data Handling
• Troubleshooting
• Releasing

🤝 Contributing

See CONTRIBUTING.md.

🔐 Security

See SECURITY.md.