Claude-GLM Wrapper (ccx)

📢 Community Fork Notice

This is an actively maintained community fork of the original claude-glm-wrapper. Now rewritten in Bun for 10x speed and single-binary simplicity.

Install via: bunx claude-glm-wrapper or download the binary.

Use Z.AI's GLM models, Minimax, OpenAI, and more with Claude Code.

One Binary. Zero Friction.

Why ccx?

🚀 10x Faster: Native binary (written in Bun/TypeScript), starts instantly. 📦 Single Binary: No more "wrapper hell". One ccx executable handles everything. 🛡️ Safe & Clean: No more eval in shell scripts. Configs stored safely in ~/.config. 🔀 Multi-Provider: Switch between GLM, Minimax, OpenAI, Gemini, and Claude instantly. 🩺 Self-Healing: Includes a doctor command to diagnose and fix configuration issues automatically.

Quick Start

Installation

If you have Bun:

bunx claude-glm-wrapper setup

Manual Download (Mac/Linux/Windows):

1. Download the latest release from GitHub Releases.
2. Run ./ccx setup (or .\ccx.exe setup on Windows).

Setup Wizard

Run ccx setup to:

1. Configure your API keys (Z.AI, Minimax, OpenAI, etc.).
2. Install shell aliases (cc, ccg, ccm...) automatically.
3. Verify your Claude Code installation.

Usage

Commands

| Command | Action | |---------|--------| | ccx | Run Claude Code with your default model (e.g. GLM-4.7) | | ccx setup | Run the interactive setup wizard | | ccx config | Open configuration file | | ccx doctor | Run self-diagnostics to check API keys, paths, and dependencies | | ccx update | Update to the latest version | | ccx --model=gpt-4o | Run with a specific model override |

Aliases (Optional)

If you enabled aliases during setup:

| Alias | Equivalent Command | |-------|-------------------| | cc | claude (Standard Anthropic) | | ccg | ccx --model=glm-4.7 | | ccg45 | ccx --model=glm-4.5 | | ccf | ccx --model=glm-4.5-air | | ccm | ccx --model=MiniMax-M2.1 |

Multi-Provider Proxy

ccx automatically acts as a local proxy. You can use ANY supported provider by specifying the prefix:

ccx --model=openai:gpt-4o
ccx --model=gemini:gemini-1.5-pro
ccx --model=minimax:MiniMax-M2.1

Or switch in-session using Claude's slash command: /model openai:gpt-4o

Auto-Discovery (Zero Config)

ccx is smart! It automatically detects API keys in your environment variables, so you might not even need to run setup.

Supported variables:

• ZAI_API_KEY or GLM_API_KEY
• MINIMAX_API_KEY
• OPENAI_API_KEY
• ANTHROPIC_API_KEY
• GEMINI_API_KEY
• OPENROUTER_API_KEY

Configuration

Config is stored in ~/.config/claude-glm/config.json. You can edit it with ccx config.

{
  "zaiApiKey": "sk-...",
  "minimaxApiKey": "...",
  "providers": {
    "openai": { "apiKey": "sk-..." }
  },
  "defaults": {
    "model": "glm-4.7"
  }
}

Troubleshooting

"ccx: command not found" Ensure the binary is in your PATH, or that you've run the setup which helps install aliases.

"Error: 'claude' command not found" Run ccx doctor or ccx setup — it will offer to install Claude Code for you automatically!

License

MIT