Linglens

Select a term on any web page or PDF and get it explained — in your language, grounded in the document you're reading, using your own API key.

Linglens is a Chrome (Manifest V3) extension for reading technical documentation in a language that isn't your first. You know the words; what you don't know is what a term means here — "runner" in a CI doc, "sink" in a streaming doc, "hydration" in a frontend doc. Highlight it, click the icon, and Linglens explains it in the context of that page, written in the language you read best.

Open source · no servers · no telemetry. Bring your own key: it's stored locally and never leaves your browser except in the request you trigger to the provider you choose. Nothing goes to us — there is no "us" backend. (One caveat you own, not us: a provider's free tier may train on your prompts — Google Gemini's does, its paid tier doesn't. Use a paid key or a local model if that matters. See PRIVACY.md.)

Why another explainer?

Full-page translators translate words but don't explain a term's meaning in context, and they mangle code. Most AI explainers are English-only and run on someone else's key, someone else's cost, and someone else's server. Linglens is different on three axes:

• Your language. Explanations come back in the language you configure (English, 中文, 日本語, and more) — not just English.
• Your key, your control. Bring your own API key — OpenAI, Anthropic, Google Gemini, DeepSeek, or any OpenAI-compatible endpoint — including a local model that never leaves your machine. Real provider-reported token counts are shown per explanation, so you always know exactly what each answer used.
• A conversation, not a one-shot. Ask a follow-up right in the popup — "give an example", "how is this different from X?" — and keep the thread.

Works in PDFs. PDFs open in a built-in reader; the explanation is grounded in the section you selected from — here "dropout" in Regularization, not the data-cleaning "dropout" a paragraph above.

Set up in seconds. Paste a free Gemini key (or connect OpenRouter, or bring any provider's key) and it's checked automatically.

Features

• Click-to-explain: a selection shows a small icon; only clicking it calls the LLM, so an accidental highlight never costs a token.
• Streaming answers rendered in a floating dark-glass popup that reads on any page (light or dark).
• Concise ↔ detailed length toggle; per-explanation token readout.
• Follow-up thread with running token-usage totals.
• Local, searchable history of the terms you've explained (no API call to re-read), with a configurable cap.
• Works in PDFs: PDFs open in a built-in reader (pdf.js) where you select and explain text just like on a web page — grounded in the whole PDF, including Chinese/Japanese/Korean documents. Switch back to Chrome's viewer per PDF or turn the reader off in settings.
• Works in Gmail's PDF attachment previews and inside embedded frames.
• Guided setup: a free Gemini key in ~30 seconds, one-click OpenRouter, or any provider's key — checked automatically when you paste it.
• English and Chinese interface, following your browser's language.
• Configure provider, model, key, language, and length from a toolbar popup.

Install

Chrome Web Store

Install from the Chrome Web Store.

From source

git clone https://github.com/ZGhey/linglens
cd linglens
npm install
npm run build        # outputs to dist/

Then in Chrome: open chrome://extensions, enable Developer mode, click Load unpacked, and select the dist/ folder. Pin the toolbar icon.

Quick start

1. Click the pinned Linglens icon and choose Set up.
2. Easiest: Free Gemini key — open Google AI Studio, sign in with your Google account, click Create API key, and paste the key back into Linglens. It is checked and saved automatically (no credit card).
3. Already have a key? Choose Enter it manually and pick your provider (or point the Custom (OpenAI-compatible) provider at a local model — no key needed for most local servers, enter any placeholder).
4. Choose your explanation language.
5. On any page or PDF, select a term → click the icon that appears.

Your key is stored locally and sent only to the provider you chose. See PRIVACY.md.

Development

npm run dev          # Vite dev server with HMR
npm run typecheck    # tsc --noEmit
npm test             # unit tests (Vitest)
npx vitest run tests/providers/stream.test.ts   # a single test file
npm run test:e2e     # builds, then Playwright drives the loaded extension

Architecture notes for contributors live in CLAUDE.md.

Contributing

Issues and pull requests welcome. Dev setup is above; the architecture overview lives in CLAUDE.md. CI (typecheck + unit + e2e) runs on every PR.

License

MIT.