Pioneer CLI

A TypeScript-based CLI agent for long-running AI processes with Qwen code-style UX.

Features

• 🎨 Beautiful ASCII art banner
• 💬 Interactive chat interface with real AI
• 🔗 Connects to pioneer-server for LLM inference
• 📍 Context-aware (shows current directory)
• 🔄 Conversation history management
• ⚡ Built with TypeScript

Quick Start

1. Install dependencies

cd apps/pioneer-cli
npm install

2. Configure environment

cp .env.example .env

Edit .env:

PIONEER_URL=http://127.0.0.1:9101

3. Make sure pioneer-server is running

See pioneer-server README for setup.

4. Start the CLI

npm start

Or with hot reload:

npm run dev

Usage

Interactive Mode

Simply run npm start and start typing your requests. The AI will respond via pioneer-server.

Commands

• /help - Show available commands
• /status - Display agent status (mode, context, conversation count, server)
• /clear - Clear the screen
• /reset - Clear conversation history
• /quit or /exit - Exit the CLI

Example Session

❯ help me refactor this authentication code
◆ Pioneer: I'd be happy to help you refactor your authentication code. Let me analyze...

❯ /status
📊 Pioneer Agent Status:
  Mode: Light Mode
  Context: /Users/you/project
  Conversations: 1
  Server: http://127.0.0.1:9101
  Model: gpt-4-turbo-preview
  MCP Server: Not implemented
  Sub-agents: 0 spawned

Architecture

pioneer-cli (TypeScript)
  ├── src/
  │   ├── index.ts      # Main CLI interface
  │   ├── agent.ts      # Agent that talks to pioneer-server
  │   ├── config.ts     # Configuration and .env handling
  │   └── colors.ts     # Terminal color constants
  └── dist/             # Compiled JavaScript

Built with:

• TypeScript for type safety
• Node.js native readline for input handling
• dotenv for environment configuration
• fetch API for HTTP requests to pioneer-server
• ANSI color codes for terminal styling
• Typewriter effect for AI responses

Configuration

Environment Variables

• PIONEER_URL - URL of pioneer-server (default: http://127.0.0.1:9101)
• MODEL - Override default model (optional)

Hard-coded Defaults

If no .env file is present, the CLI will use:

• PIONEER_URL: http://127.0.0.1:9101
• MODEL: gpt-4-turbo-preview

This allows users who just install pioneer-cli to get started quickly with sensible defaults.

Development

Build

npm run build

Run

npm start

Watch mode

npm run dev

Next Steps

• [ ] Implement MCP server integration
• [ ] Add sub-agent spawning
• [ ] Connect to pioneer-desktop
• [ ] Add agent-to-agent communication
• [ ] Stream responses from server
• [ ] Add context file loading (@file syntax)