What it does NLUX is a monorepo of npm packages for building LLM chat interfaces. The React layer gives you <AiChat /> and useChatAdapter; the vanilla core (@nlux/core) has zero dependencies and no external UI libraries. Adapters cover OpenAI, LangChain LangServe, Hugging Face Inference, and the team’s own nlbridge middleware. A CLI (nlux-cli) generates starter projects for Next.js, React, or plain TypeScript.

The interesting bit The license is MPL 2.0 with custom paragraphs 3.6 and 3.7 tacked on: you can’t use the source to train AI models or feed code-translation tools. It’s a rare legal hedge in a space where most code is LLM fodder.

Key highlights

• React Server Component support and “generative UI” for Next.js
• Stream or batch adapters for any LLM via a flexible interface
• Markdown stream parser and Highlight.js-based syntax highlighter as optional add-ons
• 600+ unit tests per the README badge
• Theming, personas, and layout customization without ejecting from the component

Caveats

• The “zero dependency” claim applies only to @nlux/core; React packages obviously depend on React
• Star count (1,376) suggests early traction, not battle-tested ubiquity
• The “under a minute” setup promise is from the README, not independently verified

Verdict Worth a look if you need a drop-in chat UI with multiple backend adapters and care about bundle weight. Skip if you’re already invested in a different chat component ecosystem or need deep message-thread logic beyond what’s shown.