JSPM

  • ESM via JSPM
  • ES Module Entrypoint
  • Export Map
  • Keywords
  • License
  • Repository URL
  • TypeScript Types
  • README
  • Created
  • Published
  • Downloads 10
  • Score
    100M100P100Q53542F

Core runtime library for building AI terminal user interfaces with MCP integration

Package Exports

  • @aiter/core

Readme

@aiter/core

Core runtime library for aiter - provides the infrastructure for building AI terminal user interfaces with Model Context Protocol (MCP) integration.

Part of the aiter project.

What is aiter?

A powerful terminal-based AI chat interface built with OpenTUI and React, featuring a modular agent system. Built on the Vercel AI SDK and OpenTUI.

For developers who want to build with AI, not just use it.

Installation

Note: This is a core library. Most users should use @aiter/cli to create applications:

# Create a new aiter application (recommended)
bunx @aiter/cli create app my-chat-app

If you're building a custom application:

bun add @aiter/core

What's Included

  • UI Components: React components for terminal interfaces (ChatContainer, Messages, ChatBox, etc.)
  • Trigger System: Extensible command processing and registry
  • MCP Integration: Model Context Protocol client management and tool loading
  • Chat Management: Session creation, persistence, and resumption
  • Agent Resolution: Dynamic agent loading with file-system based organization
  • Component Context: System for customizing and overriding UI components

Quick Example

import { 
  ChatContainerWrapper, 
  initializeMCP, 
  createAgentResolver 
} from '@aiter/core';

// Initialize agent system
const agentResolver = createAgentResolver();
const agent = await agentResolver.getAgent('my-agent');

// Initialize MCP tools
await initializeMCP(agent.mcpConfig);

// Render chat interface
<ChatContainerWrapper
  chat={chat}
  prompt={prompt}
  agent={agent}
  customComponents={customComponents}
/>

Documentation

See the main aiter repository for full documentation and examples.

License

See LICENSE file.