Try without installing

npx @vybestack/llxprt-code --provider synthetic --model hf:zai-org/GLM-5 --keyfile ~/.synthetic_key "simplify the README.md" ```

Quick Start

1. Prerequisites: Node.js 20+ installed (not required for Homebrew)
2. Install:

   # macOS (Homebrew)
   brew tap vybestack/homebrew-tap
   brew update
   brew install llxprt-code

   # npm
   npm install -g @vybestack/llxprt-code
   # Or try without installing:
   npx @vybestack/llxprt-code
   

1. Run: llxprt
2. Choose provider: Use /provider to select your preferred LLM service
3. Start coding: Ask questions, generate code, or analyze projects

First session example:

cd your-project/
llxprt
> Explain the architecture of this codebase and suggest improvements
> Create a test file for the user authentication module
> Help me debug this error: [paste error message]

Free & Subscription Options

Get started immediately with powerful LLM options:

```bash

Access via the commands interface

/subagent list /subagent create <name> ```

Each subagent can be configured with:

• Different providers (Gemini vs Anthropic vs Qwen vs Local)
• Different models (Flash vs Sonnet vs GLM 5 vs Custom)
• Different tool access (Restrict or allow specific tools)
• Different settings (Temperature, timeouts, max turns)
• Isolated runtime context (No memory or state crossover)

Subagents are designed for:

• Specialized tasks (Code review, debugging, documentation)
• Different expertise areas (Frontend vs Backend vs DevOps)
• Tool-limited environments (Read-only analysis vs Full development)
• Experimental configurations (Testing new models or settings)

Full Subagent Documentation →

Zed Integration

LLxprt Code integrates with the Zed editor using the Agent Communication Protocol (ACP):

{
  "agent_servers": {
    "llxprt": {
      "command": "/opt/homebrew/bin/llxprt",
      "args": ["--experimental-acp", "--profile-load", "my-profile", "--yolo"]
    }
  }
}

Configure in Zed's settings.json under agent_servers. Use which llxprt to find your binary path.

Features:

• In-editor chat: Direct AI interaction without leaving Zed
• Code selection: Ask about specific code selections
• Project awareness: Full context of your open workspace
• Multiple providers: Configure different agents for Claude, OpenAI, Gemini, etc.

Zed Integration Guide →

Complete Provider Guide →

Interactive vs Non-Interactive Workflows

Interactive Mode (REPL): Perfect for exploration, rapid prototyping, and iterative development:

```bash