ZAI CLI


> A conversational AI CLI tool powered by Z.ai GLM models with intelligent text editing and tool usage.

Forked from superagent-ai/grok-cli and enhanced with advanced features for the Z.ai GLM ecosystem.

!ZAI CLI Screenshot

✨ What's New in v0.3

$3

ZAI can now spawn specialized agents to handle complex tasks automatically:

- 10 Specialized Agents - Code review, testing, documentation, debugging, security, performance, and more
- Autonomous Execution - GLM decides when to use agents based on your request
- Visual Feedback - Animated indicators show agent status (⚙️ working → ✅ success / ❌ error)
- Isolated Context - Agents work independently without polluting your main conversation

Example: Ask "review my auth module" and GLM automatically launches the code-reviewer agent!

$3

- Shell Auto-completion - Tab completion for bash, zsh, and fish
- Token Budget Management - Set limits with --token-budget to control API costs
- Undo Command - Restore previous file versions with /undo
- Better Error Messages - Context-aware API error explanations
- No-Color Mode - --no-color flag for CI/CD environments
- Debug Mode - --debug for verbose logging

$3

- ↑/↓ - Navigate through command history
- Ctrl+R - Reverse search through command history (fuzzy search)
- Diff Viewer:
- a - Accept changes and apply to file
- r - Reject changes and discard
- d - Toggle between summary and full diff view

🚀 Quick Start

$3

``bash


Install from npm (recommended)

npm install -g @guizmo-ai/zai-cli

Or with Bun

bun add -g @guizmo-ai/zai-cli
`

$3

`bash
zai
`

The interactive wizard will guide you through:
1. Enter your Z.ai API key (get one here)
2. Select default model (GLM-4.6 recommended)
3. Start chatting!

Configuration is saved to ~/.zai/user-settings.json.

✨ What's Different from grok-cli?

$3

- Native Z.ai GLM support - Built specifically for GLM-4.6, GLM-4.5, and GLM-4.5-Air
- 200K context window - Leverage GLM-4.6's massive context for complex projects
- Optimized prompting - Tailored system prompts for GLM model reasoning patterns

$3

- First-run wizard - Guided setup with no manual config needed
- Interactive settings panel - Manage API keys and models with zai config
- Smart defaults - Pre-configured for Z.ai endpoints and best practices

$3

- Thinking mode - Visualize AI reasoning process in real-time (GLM extended thinking)
- Batch editing - Apply changes across multiple files simultaneously
- File watching - Real-time detection of external file changes
- Session persistence - Save and restore conversations with full context
- Typed error system - Contextual error messages with actionable suggestions

$3

- 90+ tests - Comprehensive test suite with Vitest
- Error recovery - Intelligent error handling with retry mechanisms
- Metrics tracking - Built-in performance and usage monitoring

📖 Core Features

$3

Natural language interface with Z.ai GLM models:
- GLM-4.6 (200K context) - Best for complex coding and reasoning
- GLM-4.5 (128K context) - Balanced performance
- GLM-4.5-Air - Fast and lightweight

$3

AI automatically:
- Views, creates, and edits files with interactive diff preview
- Executes shell commands with persistent history
- Selects appropriate tools
- Handles multi-file operations with automatic backups
- Ctrl+R fuzzy search through command history

$3

`bash


Interactive settings panel

zai config

View configuration

zai config --show

Update API key

zai config --set-key YOUR_KEY

Reset to defaults

zai config --reset
`

$3

Interactive Mode:
`bash
zai # Start in current directory
zai -d /path/to/project # Specify working directory
`

Headless Mode (for scripts/CI):
`bash
zai --prompt "analyze package.json and suggest improvements"
zai -p "run tests" --max-tool-rounds 50
`

$3

`bash


Command line

zai --model glm-4.6

Environment variable

export ZAI_MODEL=glm-4.5
zai

User settings (persistent)

zai config # Select "Change Default Model"
`

🔧 Configuration

$3

`bash
export ZAI_API_KEY=your_api_key_here
export ZAI_BASE_URL=https://api.z.ai/api/paas/v4 # Optional
export ZAI_MODEL=glm-4.6 # Optional
`

$3

Create .zai/ZAI.md in your project:

`markdown


Project Guidelines

- Use TypeScript with strict mode
- Follow functional programming patterns
- Add JSDoc comments for public APIs
- Write tests for all new features
`

ZAI will automatically follow these instructions when working in your project.

$3

Create .zai/settings.json for project-specific configuration:

`json
{
"model": "glm-4.5",
"mcpServers": {
"linear": {
"name": "linear",
"transport": "sse",
"url": "https://mcp.linear.app/sse"
}
}
}
`

🔌 MCP Integration

Extend ZAI with Model Context Protocol servers:

`bash


Add Linear integration

zai mcp add linear --transport sse --url https://mcp.linear.app/sse

Add custom MCP server

zai mcp add my-server --transport stdio --command "bun" --args server.js

List servers

zai mcp list

Test connection

zai mcp test linear
`

⚡ Morph Fast Apply (Optional)

Enable high-speed editing at 4,500+ tokens/sec:

`bash
export MORPH_API_KEY=your_morph_key
`

When enabled:
- edit_file tool for complex refactoring
- Intelligent diff-based editing
- 98% accuracy on code changes

📊 Advanced Features

$3

GLM automatically spawns specialized agents when needed:

Available Agents:
- code-reviewer - Code quality, bugs, security, best practices
- test-writer - Unit and integration tests
- documentation - README, API docs, technical documentation
- refactoring - Code structure and maintainability improvements
- debugging - Systematic bug diagnosis and fixes
- security-audit - Security vulnerability analysis
- performance-optimizer - Performance analysis and optimization
- explore - Quick codebase exploration and understanding
- plan - Detailed implementation planning

Manual Control:
`bash


In chat mode


/agents # List all available agents
/task code-reviewer "review auth module" # Launch specific agent
/tasks # View agent execution history
`

$3

Your command history is automatically saved across sessions:
`bash

History saved to ~/.zai/history.json

Use ↑/↓ to navigate through previous commands

Press Ctrl+R for fuzzy search through history

`

$3

Preview all file changes before they're applied with side-by-side diff view and options to accept, reject, or view full diff.

$3

Every file edit is automatically backed up to ~/.zai/backups/:
`bash
/undo path/to/file.ts # Restore previous version
`

$3

Control API costs with configurable limits:
`bash
zai --token-budget 50000 # Set max tokens
zai --token-warn-at 80 # Warn at 80% usage
`

$3

Enable tab completion for your shell:
`bash

Generate completion script

zai completion bash > ~/.zai-completion.sh
source ~/.zai-completion.sh

Or for zsh

zai completion zsh > ~/.zai-completion.zsh

Or for fish

zai completion fish > ~/.config/fish/completions/zai.fish
`

$3

- Batch Editing - Apply changes across multiple files
- File Watching - Real-time detection of external file changes
- Session Persistence - Auto-save conversations to ~/.zai/sessions/
- Thinking Mode - Visualize AI reasoning with GLM-4.6

🛠️ Development

`bash


Clone the repository

git clone https://github.com/guizmo-ai/zai-glm-cli.git
cd zai-glm-cli

Install dependencies

npm install

Development mode

npm run dev

Build

npm run build

Run tests

npm test

Type check

npm run typecheck
`

📋 Command Reference

$3

`bash
zai [options] [message...]

Options:
-V, --version Version number
-d, --directory

Working directory
-k, --api-key Z.ai API key
-u, --base-url API base URL
-m, --model AI model (glm-4.6, glm-4.5, glm-4.5-air)
-p, --prompt Headless mode - process and exit
--max-tool-rounds Max tool executions (default: 400)
--no-color Disable colored output (for CI/CD)
--debug Enable debug mode with verbose logging
--token-budget Set maximum token budget (e.g., 50000)
--token-warn-at Warn at percentage (default: 80)
-h, --help Show help

Commands:
config [options] Manage settings
mcp [options] Manage MCP servers
completion [shell] Generate shell completion script
`

$3


`bash
/clear # Clear chat history (with confirmation)
/undo # Restore previous version of file
/agents # List all available specialized agents
/task # Manually launch an agent
/tasks # View agent execution history
`

🔄 Migration from grok-cli

1. Install ZAI CLI:
`bash
npm install -g @guizmo-ai/zai-cli
`

2. Copy settings (optional):
`bash
cp -r ~/.grok ~/.zai
`

3. Update configuration:
`bash
zai config
# Enter Z.ai API key
# Select GLM model
`

4. Environment variables:
- GROK_API_KEY → ZAI_API_KEY
- GROK_BASE_URL → ZAI_BASE_URL
- GROK_MODEL → ZAI_MODEL

📦 Architecture

- Agent - Z.ai GLM-powered command processing with extended thinking
- Tools - File operations, bash execution, batch editing
- UI - React Ink interface with onboarding and settings panels
- Settings - Two-tier config (user + project)
- MCP - Protocol support for extensibility

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

🙏 Credits

Based on grok-cli by Superagent AI.

Enhanced for Z.ai GLM models with:
- Interactive onboarding and configuration
- GLM-specific optimizations and thinking mode
- Advanced tooling (batch editing, file watching, sessions)
- Comprehensive error handling and testing

📄 License

MIT License - Copyright (c) 2025 Guizmo AI

See LICENSE file for details.

---

Get started: npm install -g @guizmo-ai/zai-cli && zai`

Need help? Open an issue on GitHub