glm

README

GLM CLI

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

Features

• 🚀 Session-Based Launch: Launch Claude with GLM settings temporarily (no persistent config changes)
• 🎯 Model Selection: Choose different GLM models at launch time (glm-5, glm-4.7, glm-4.6, glm-4.5, glm-4.5-air, etc.)
• 🔀 Flag Passthrough: Pass any claude CLI flags directly through glm (e.g., --allowedTools, --verbose)
• ⚡ YOLO Mode: Skip permission prompts with --yolo flag for faster workflows
• 📦 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

Installation

Quick Install (Recommended)

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/v1.2.0/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

Manual Installation

Option 1: Download Pre-built Binary

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
   

Option 2: Build from Source

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/

Authentication Setup

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

Option 1: Interactive Setup (Recommended)

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

glm  # Will prompt for token if not found

Option 2: Manual Token Setup

glm token set  # Enter your token securely

Option 3: Environment Variable

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

Usage

Launch Claude with GLM (Primary Usage)

Launch Claude with the default model (glm-5):

glm

Launch Claude with a specific model:

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

Launch Claude in YOLO mode (skip permission prompts):

glm --yolo
glm --yolo --model glm-4.5-air

Pass additional flags directly to claude:

glm --allowedTools "Bash,Read,Write"
glm --verbose
glm --yolo --allowedTools "Bash,Read"

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

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

glm install claude

Manage Authentication Token

Set your API token:

glm token set

View current token (masked):

glm token show

Clear stored token:

glm token clear

Update GLM

Check for updates:

glm update --check

Update to latest version:

glm update

Update without confirmation:

glm update --force

Help

Get help for any command:

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

Commands Reference

Command	Description	Example
glm	Launch Claude with GLM (temporary config)	glm --model glm-5
glm --yolo	Launch with permission prompts skipped	glm --yolo
glm --<flag>	Pass any flag through to claude	glm --allowedTools "Bash"
glm install claude	Install Claude Code	glm install claude
glm token set	Set authentication token	glm token set
glm token show	Show current token (masked)	glm token show
glm token clear	Clear stored token	glm token clear
glm update	Update GLM to latest version	glm update
glm update --check	Check for updates only	glm update --check

Deprecated Commands

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

Command	Status	Replacement
glm enable	⚠️ Deprecated	Use glm instead
glm disable	⚠️ Deprecated	Run claude directly
glm set	❌ Removed	Use glm --model X

Available Models

• glm-5 (default)
• glm-4.7
• glm-4.6
• glm-4.5
• glm-4.5-air
• Any other GLM model supported by BigModel API

Configuration Files

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.

How It Works

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.

Example Workflow

# 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-5)
glm

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

# Launch in YOLO mode (skip permission prompts)
glm --yolo

# Pass additional flags to claude
glm --allowedTools "Bash,Read,Write"

# Use Claude without GLM
claude

# Check for updates
glm update --check

# Update to latest version
glm update

Troubleshooting

Installation Issues

curl not found

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)

Permission denied during installation

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

Binary not found for your platform

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

Runtime Issues

npm not found

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

Authentication token not found

Set up your token using any of these methods:

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

Claude still using default settings

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

Command not found after installation

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)

Update fails with permission error

If glm update fails with permission denied:

sudo glm update

Migration from Previous Versions

If you're upgrading from version 1.0.x:

⚠️ IMPORTANT: Remove Old Configuration File

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)

Other Changes:

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

License

MIT License - see LICENSE file for details.

Contributing

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

Support

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