nodx Lens

Select text or box any page region → AI explain, search, shop, or generate images. Captures stay marked and sync to nodx desktop.

What's new in 1.0 — The customization release • The action wheel is fully yours: every spoke's icon (40-emoji library or upload your own image), name, colour, action, prompt, and submenus are editable — with a live wheel preview while you edit. • A built-in library of 13 search destinations — search engines, shopping sites, video and knowledge platforms — pick from a dropdown; hand-written URL prefixes are now the advanced path. • 5 AI providers with per-provider saved keys, including a free-model option and a local-gateway option that needs no key at all. • Quick model switcher right in the side panel header — change provider and model without opening Settings. • AI answers now render as readable formatted text: bold, lists, inline code, links. • Image generation gets choices: one clean render or the 2×2 style grid, with an editable style prompt. • 1 GB local history budget with oldest-first cleanup; delete any capture box right on the page. • An Android companion app is out too — same action wheel, system-wide on your phone: https://aicon.solutions/nodx/lens/ nodx Lens turns anything you see on a webpage — a phrase, a product photo, a chart — into instant AI action. Select text or box any region of the page, and an action wheel appears: Explain, Search, Shop, Generate, Save. Results stream in place. No tab-switching, no new chat thread to clutter, no copy-pasting into another tool. ═══════════════════════════════════════════ WHO IT'S FOR ═══════════════════════════════════════════ - Anyone reading long AI replies in a chat tool — get a side-channel definition without polluting your main conversation. - Knowledge workers who skim research papers, essays, technical docs, and want fast context on a term they don't know. - Shoppers and collectors — box a product photo and jump straight to a shopping search, or turn it into style-exploration images. - People who collect annotations as they read — every explanation leaves a persistent underline, every capture leaves a marked box you can revisit and act on again. ═══════════════════════════════════════════ CORE FEATURES ═══════════════════════════════════════════ 🎡 One action wheel, two ways in — now fully customizable Select text, or box any region of the page ("📸 Screenshot region" in the side panel). Either way the same four-spoke wheel appears: 🔍 Explain / Search, 💡 Save, 🛒 Shop, 🎨 Generate — that's the stock layout. Every spoke is yours to change: icon (emoji library or your own uploaded image), name, colour, action kind, prompt text, and whether it fans out into a submenu of up to 3 sub-items. A live preview mirrors your edits as you type. 🔍 Select-to-explain Highlight 2–500 characters on any webpage, pick Explain, and a panel slides in below the highlight with a streaming explanation — 50–150 words by default, tuned to be a quick definition with one concrete example. Answers render as formatted text: bold, lists, links. 👁 Vision on any region Box a product photo, a chart, a UI — ask "what is this?" about the exact pixels. Visual search uses a reverse-image search when the region is a real image, and AI recognition + image search otherwise. Shopping lets the AI name the product, then opens your chosen destination with that query. 🧭 A search-destination library built in The "AI identify → open URL" action ships with 13 presets covering the major search engines, shopping sites, and video and knowledge platforms — pick one from a dropdown. Or switch to "Custom URL…" and paste any search URL prefix you like. 🎨 Image generation From selected text or a boxed region, generate one clean image or a 2×2 style grid — product photo, watercolor, 3D render, and line art — powered by your own image-model key. The style prompt is an editable template, so you can define your own four styles. 📦 Boxes are action hubs Every capture leaves a yellow highlight box on the page. Click the box to reopen the wheel and run more actions on the same capture — no re-selecting, no duplicate boxes. The corner badge summarizes what happened there: 💬 3 🔎 🎨 means three Q&A turns, a search, and a generated image. Click the badge to open that capture's card in the side panel. Hover a box to reveal a ✕ that deletes it right there. 🗂 The side panel remembers everything Captures from the current page stay on top, each with its own Q&A thread for follow-up questions. A collapsible global history logs every search, shopping trip, and generated image with a "reopen" link. Text actions are logged too, as quote-card thumbnails. History self-manages within a 1 GB local budget, cleaning oldest entries first. ⚡ Switch models without leaving the page The side panel header has a provider + model quick-switcher, synced two-way with Settings. 📌 Persistent underlines When an explanation finishes, the selected text gets an underline that stays on the page. Click any underline later to bring the panel back with the same explanation (no extra API call). Right-click to delete. Text-generated images leave an underline marker on their source text as well. 📚 Deepen on demand The panel has a "deepen" button that swaps in a longer 200–400 word explanation — mechanism, example, common misconceptions, or contrast with adjacent ideas. You decide when to spend more tokens. 🌊 Real streaming Words appear as the model generates them. You can tell within the first sentence whether the answer is what you wanted — and close the panel early if it's off-track. 🖥 Works with nodx desktop (optional) If the nodx desktop app is running, captures and saved snippets sync into its inspiration pool over 127.0.0.1 — your machine only, no cloud hop. The desktop's local gateway can even serve as your AI provider: AI calls run through your existing coding-assistant subscription, no API key needed. The extension is fully useful without it. 📱 And on your phone nodx Lens for Android puts the same action wheel over any app via a floating bubble — box a region, explain / search / shop / generate, everything logged. APK and details: https://aicon.solutions/nodx/lens/ ═══════════════════════════════════════════ PRIVACY ═══════════════════════════════════════════ This extension does not have a backend. We do not run a server. ✓ Your API keys, settings, captures, and history live only in local browser storage on your own machine. Each provider's key is stored separately — switching providers never loses a key. ✓ When you trigger an action, the request goes directly from your browser to the AI provider you configured — never through any nodx-controlled server. ✓ Desktop sync targets 127.0.0.1 (your own computer) and can be switched off with one checkbox. ✓ No analytics, no telemetry, no third-party trackers, no ads. Full privacy policy is linked from the listing. ═══════════════════════════════════════════ HOW IT WORKS (2-MINUTE SETUP) ═══════════════════════════════════════════ 1. Install the extension. 2. Settings opens automatically on first install. 3. Pick an AI provider — five are supported, including a free-model tier and a no-key local option. 4. Paste your own API key (from the provider's console — you keep full control of usage and billing). 5. For image generation, add an image-model key (a free tier works). 6. Go to any webpage — highlight something, or open the side panel and box a region. ═══════════════════════════════════════════ WHY BRING-YOUR-OWN-KEY INSTEAD OF A SUBSCRIPTION ═══════════════════════════════════════════ - Privacy: your queries go to your provider account, not a middleman's. - Pay-as-you-go: a typical short explanation costs a fraction of a cent on fast-tier models, and the free-model option costs nothing at all. - Model freedom: swap providers or models any time, even mid-page from the side panel. You're never locked in. - Transparency: you can see your own API console for exactly what was sent and what it cost. ═══════════════════════════════════════════ SUPPORTED PROVIDERS ═══════════════════════════════════════════ Anthropic, OpenAI, and Google models (fast tiers for short explanations, stronger tiers for deep dives and vision), OpenRouter's free model tier, and your own local nodx gateway. Image generation uses Google's image models with your own key. You can change the picked model per provider in Settings any time — or from the side panel's quick switcher. ═══════════════════════════════════════════ PERMISSIONS, IN PLAIN ENGLISH ═══════════════════════════════════════════ - "Read and change all your data on all websites" — sounds scary; in practice, it's because the extension has to be ready to activate when you select text or box a region on whatever page you're reading. It does not read or modify page contents unless you explicitly act. - "Storage" (incl. unlimitedStorage) — to keep your API keys, settings, captures, and history on your local machine. The extension self-enforces a 1 GB cap and cleans oldest history first. - "Side panel" — the extension's main surface: your captures, Q&A threads, and action history. - "Clipboard write" — the Copy button on answers. - "Host permissions" to the four AI provider APIs — so the extension can talk directly to whichever provider you chose. Plus 127.0.0.1:8787 for optional sync to your own nodx desktop app. ═══════════════════════════════════════════ CONTACT / FEEDBACK ═══════════════════════════════════════════ X (Twitter): https://x.com/LaoMo9394 nodx is a small toolkit for thinking with AI as a sparring partner, not a replacement. Lens is the lightweight browser entry point. A networked decision-thinking desktop app and an Android companion ship alongside it.