A command-line interface for launching Claude Code with GLM (ChatGLM) settings via BigModel API, using temporary session-based configuration.

• 🚀 Session-Based Launch: Launch Claude with GLM settings temporarily (no persistent config changes)
• 🎯 Model Selection: Choose different GLM models at launch time (glm-4.6, glm-4.5, glm-4.5-air, etc.)
• 📦 Auto-Install: Install Claude Code with built-in npm dependency checking
• 🔄 Auto-Update: Check for and install updates with interactive update command
• ⚙️ Token Management: Securely manage your authentication token

Automatic Installer:

curl -fsSL https://raw.githubusercontent.com/xqsit94/glm/main/install.sh | bash

Alternative - Manual Quick Install:

# Create user bin directory and download GLM CLI
mkdir -p ~/.local/bin
curl -L -o ~/.local/bin/glm "https://github.com/xqsit94/glm/releases/download/v1.1.0/glm-$(uname -s | tr '[:upper:]' '[:lower:]')-$(uname -m | sed 's/x86_64/amd64/')"
chmod +x ~/.local/bin/glm

# Add to PATH (one-time setup)
echo 'export PATH="$HOME/.local/bin:$PATH"' >> ~/.bashrc
source ~/.bashrc

Both methods will:

• Detect your operating system and architecture
• Download the latest binary release
• Install to your user directory
• Set up PATH for easy access

1. Go to the releases page
2. Download the binary for your platform:
   • macOS Intel: glm-darwin-amd64
   • macOS Apple Silicon: glm-darwin-arm64
   • Linux x64: glm-linux-amd64
   • Linux ARM64: glm-linux-arm64
3. Make it executable and move to PATH:

   chmod +x glm-*
   sudo mv glm-* /usr/local/bin/glm

Prerequisites:

• Go 1.24 or later
• Your GLM API token

git clone https://github.com/xqsit94/glm.git
cd glm
go mod tidy
go build -o glm
sudo mv glm /usr/local/bin/

The GLM CLI supports multiple ways to provide your Anthropic API token:

On first run, the CLI will automatically prompt you to set up your token:

glm  # Will prompt for token if not found

glm token set  # Enter your token securely

export ANTHROPIC_AUTH_TOKEN="your_token_here"
glm

Token Priority Order:

1. Environment variable ANTHROPIC_AUTH_TOKEN
2. Config file ~/.glm/config.json
3. Interactive prompt

Launch Claude with the default model (glm-4.6):

glm

Launch Claude with a specific model:

glm --model glm-4.5-air
glm -m glm-4.5-air

How it works:

• Sets temporary environment variables for the Claude session
• No persistent changes to Claude's configuration files
• Settings only apply to the launched Claude session
• To use Claude without GLM, just run claude directly

Install Claude Code via npm (with automatic Node.js detection):

glm install claude

Set your API token:

glm token set

View current token (masked):

glm token show

Clear stored token:

glm token clear

Check for updates:

glm update --check

Update to latest version:

glm update

Update without confirmation:

glm update --force

Get help for any command:

glm --help
glm install --help
glm token --help
glm update --help

These commands still work but are deprecated. Use glm with --model flag instead:

• glm-4.6 (default)
• glm-4.5
• glm-4.5-air
• Any other GLM model supported by BigModel API

The CLI manages the following files:

• ~/.glm/config.json - Your authentication token and preferences

Note: GLM no longer modifies ~/.claude/settings.json. All configuration is passed via temporary environment variables.

1. Launch (glm): Launches Claude Code with temporary environment variables:

   • ANTHROPIC_BASE_URL=https://open.bigmodel.cn/api/anthropic
   • ANTHROPIC_AUTH_TOKEN=<your_token>
   • ANTHROPIC_MODEL=<selected_model>

2. Session-Based: Settings only exist for the launched Claude session. No persistent file modifications.

3. Token Storage: Your authentication token is securely stored in ~/.glm/config.json for convenience.

4. Install: Checks for npm and installs Claude Code globally.

5. Update: Downloads and replaces the GLM binary with the latest version from GitHub.

# Install GLM CLI
curl -fsSL https://raw.githubusercontent.com/xqsit94/glm/main/install.sh | bash

# First time setup
glm install claude        # Install Claude Code
glm token set            # Enter your token securely

# Launch Claude with GLM (default model: glm-4.6)
glm

# Launch with specific model
glm --model glm-4.5-air

# Use Claude without GLM
claude

# Check for updates
glm update --check

# Update to latest version
glm update

If you get a "curl not found" error:

• macOS: Install Xcode Command Line Tools: xcode-select --install
• Linux: Install curl: sudo apt install curl (Ubuntu/Debian) or sudo yum install curl (CentOS/RHEL)

If the installer fails with permission errors:

# Download and run manually with explicit sudo
curl -fsSL https://raw.githubusercontent.com/xqsit94/glm/main/install.sh -o install.sh
chmod +x install.sh
sudo ./install.sh

If no binary is available for your platform:

1. Check the releases page for available binaries
2. Build from source using the manual installation instructions

If you get an npm error when running glm install claude:

1. Install Node.js from https://nodejs.org/
2. Restart your terminal
3. Run glm install claude again

Set up your token using any of these methods:

• glm token set (recommended)
• Set environment variable: export ANTHROPIC_AUTH_TOKEN="your_token"

The session-based configuration means:

• Settings only apply to Claude sessions launched via glm
• If you run claude directly, it uses default settings
• This is intentional - use glm to launch with GLM settings

If glm command is not found after installation:

1. Check if /usr/local/bin or ~/.local/bin is in your PATH: echo $PATH
2. Add to PATH if missing (add to .bashrc, .zshrc, etc.):

   export PATH="$HOME/.local/bin:$PATH"

3. Restart your terminal or run: source ~/.bashrc (or .zshrc)

If glm update fails with permission denied:

sudo glm update

If you're upgrading from version 1.0.x:

You MUST remove the old persistent configuration file to avoid conflicts:

rm -f ~/.claude/settings.json

Why this is required:

• Version 1.0.x created a persistent ~/.claude/settings.json file that made Claude always use GLM settings
• This conflicts with v1.1.0's session-based approach
• Without removing this file: Running claude directly will still use GLM settings (not the default)
• After removing this file:
  • glm → Uses GLM settings (temporary, session-based)
  • claude → Uses default Claude settings (no GLM)

1. Deprecated commands: glm enable and glm disable still work but show deprecation warnings
2. Removed command: glm set has been removed - use glm --model X instead
3. New usage: Just run glm to launch Claude with GLM, or glm --model X to specify a model

MIT License - see LICENSE file for details.

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

For issues and feature requests, please create an issue in the repository.