@localmode/ui · shadcn registry
Local-First AI UI Components
100+ copy-owned React components and 36 installable blocks for browser-native AI, across Chat, Knowledge & RAG, Vision, Audio, Photo, Agents, and Privacy. Local-first by design, cloud-compatible by contract.
npx shadcn@latest add @localmode/ui/conversation/messageAdd any of the 100+ components the same way. See the install guide.
…and they run real models, on-device
The Blocks gallery composes these elements into installable experiences, each running a real model entirely in your browser. No server, no API key.
ChatMulti-provider on-device chat over a real model.OpenRAG ChatGrounded, cited answers streamed over your own documents, fully in the browser.OpenObject DetectorFind and label objects in a photo or webcam frame with boxes and scores.OpenVoice NotesRecord or upload audio and get a searchable on-device transcript.OpenImage SearchFind photos by describing them or by visual similarity, all on-device.OpenPII RedactorDetect and mask names, emails, and IDs with on-device NER.Open
Plus 30 more blocks across 12 categories, including Writing Tools, Text Insights, Agents, Device, Image Studio, and Text.
See all blocksEverything runs on your device
What your browser reports it can do for on-device AI. Detected locally, nothing is sent anywhere.
Browser-Reported Capabilities
Detected locally in your browser. Nothing is sent anywhere.
Browser--
Device--
CPU Cores--
Browser Memory--
Browser Storage--
GPU--
Feature Support (hover for details)
WebGPU: not supportedWebNN: not supportedWASM: not supportedSIMD: not supportedThreads: not supportedIndexedDB: not supportedOPFS: not supportedWorkers: not supportedSAB: not supportedSW: not supportedCOI: not supportedChrome AI: not supported
Frequently asked questions
Does LocalMode UI work offline?
Yes. After the initial model download, inference runs entirely in the browser with no network requests, and data never leaves your device. The site is also an installable PWA, so its shell works offline.
Do I need an API key or a server?
No. Blocks run real models on-device via WebGPU or WebAssembly. There is no server and no API key, and installation is client-only.
How do I install a component?
Configure the @localmode namespace in your components.json, then run npx shadcn@latest add @localmode/ui/<name>. Components are copy-owned and inherit your shadcn/ui theme.
Which browsers are supported?
WebGPU-capable browsers (Chrome and Edge 113+, Safari 26+) get GPU-accelerated inference; other modern browsers fall back to WebAssembly.
Is it free and open source?
Yes. LocalMode UI is MIT-licensed and free to use.