@nan0web/ui-llm 🧠

LLM Protocol layer for NanoWeb. Sessions, Drivers, Models, Streaming.

Architecture

Core Exports

• LLiMoSession — manages a conversation session with an LLM
• LLiMoMessage — message format (role, content, metadata)
• LLiMoResponse — structured response from LLM
• LLiMoDriver — provider-specific driver (Cerebras, HuggingFace, OpenAI, Google)
• LLiMoContact — contact information for LLM endpoint
• Model — model metadata and capabilities
• CliChat — CLI chat abstraction

Philosophy: Chat as the Universal Base

In the One Logic — Many UI strategy, the Chat/LLM interface serves as the primary abstraction for all sequential interactions.

• Voice is simply a Chat where text is transported via audio (STT/TTS)
• Forms are decomposed into a sequence of Chat messages (Input → Reply → Validation → Next)

Component Mapping (CLI → LLM)

| CLI Component | LLM Translation | | :------------------------ | :-------------------------------------------------------- | | Select / Autocomplete | Choice buttons, numbered list, or natural language intent | | Confirm | Yes/No quick replies | | Tree / Navigation | Drill-down messages or breadcrumb-labeled options | | Table | Paginated lists or summarized data cards | | Alert / Toast | System notification messages with level-based styling |

Installation

How to install with npm?

npm install @nan0web/ui-llm

How to install with pnpm?

pnpm add @nan0web/ui-llm

Usage

Session

Session manages the chat sequence.

How to create a LLiMoSession instance?

import { LLiMoSession } from '@nan0web/ui-llm'
const session = new LLiMoSession()
console.info(typeof session)

Java•Script

Uses d.ts files for autocompletion

API

All exported classes should pass basic test to ensure API examples work

Contributing

How to contribute? - check here

License

How to license? - ISC LICENSE file.