AI Tools

📦 Installation

Requires Node.js ≥18. Install globally via npm:

# Install globally via npm
npm install -g @openai/codex

# Or install via Homebrew on macOS
brew install --cask codex

# Verify
codex --version

🔑 Authentication

On your first run of the codex command, the terminal will guide you to sign in. You can authenticate via OAuth in your browser using your ChatGPT account (supporting Plus, Pro, Team, or Enterprise), or supply an OpenAI API key in the configuration.

# Start and authenticate
codex

⚡ Core Commands & Flags

🔧 Configuration (config.toml)

Codex CLI reads user defaults from ~/.codex/config.toml. You can customize the active model, safety policies, and register MCP servers:

# Sample ~/.codex/config.toml
model = "gpt-4o"
approval_policy = "on-request"

# Configure Model Context Protocol (MCP) servers
[mcp_servers.my_helper]
enabled = true
command = "node"
args = ["/path/to/server.js"]

Project-level configs can be defined in a local .codex/config.toml to override global user configs (trusted projects only).

🤖 Interactive TUI Commands

Within the active codex TUI session, type / to trigger a popup containing the following commands:

⚠️ Security Warning

Make sure you verify the exact package name before running npm install. Malicious actors have historically published package typos (e.g., codexui-android) attempting to steal authentication tokens from ~/.codex/auth.json. Always verify official package sources.

🔗 Useful Links

OpenAI Codex Dev Docs →
OpenAI Platform →

📦 Installation

Run the following command in your terminal to install agy:

# macOS & Linux Installation
curl -fsSL https://antigravity.google/cli/install.sh | bash

# Windows (PowerShell) Installation
irm https://antigravity.google/cli/install.ps1 | iex

# Verify installation
agy --version

🔑 Authentication

On your first run, type agy and press Enter. It will guide you through the OAuth flow in your browser to sign in to your Google account, then paste the code back in the terminal.

# Start and login
agy

# Logout
# Run /logout inside the interactive session

⚡ Core Commands

📁 Subcommands

🤖 Interactive TUI Commands

Within the interactive agy TUI session, you can run the following shortcut slash commands:

🔌 Use via Antigravity Python SDK

Or programmatically orchestrate agents using the official Python SDK:

pip install google-antigravity

import asyncio
from google.antigravity import Agent, LocalAgentConfig, CapabilitiesConfig

async def main():
    # Read-only by default. Pass CapabilitiesConfig() to allow write/command execution.
    config = LocalAgentConfig(
        system_instructions="You are an expert codebase assistant.",
        capabilities=CapabilitiesConfig()
    )
    
    async with Agent(config) as agent:
        response = await agent.chat("Write a short Python script to list files")
        # Stream response tokens
        async for token in response:
            print(token, end="", flush=True)

if __name__ == "__main__":
    asyncio.run(main())

🔗 Useful Links

Agy Official Docs (Home) →
Antigravity SDK GitHub →

📦 Installation

Requires Node.js ≥22. Install globally via npm:

# Install globally via npm
npm install -g @github/copilot

# Verify
copilot --version

Note: Older packages like @githubnext/github-copilot-cli are deprecated. Please migrate to the new @github/copilot package.

🔑 Authentication

Run the copilot command and use the /login command inside the session to authenticate with your GitHub account:

# Start Copilot CLI
copilot

# Run login inside the session
/login

Requires an active GitHub Copilot subscription.

⚡ Core Modes & Interactive Controls

🤖 Interactive Slash Commands

Within the interactive copilot session, use these slash commands starting with /:

🔌 Use via GitHub Copilot SDK

For programmatic agent orchestration, you can use the official Copilot SDK in Node.js / TypeScript projects:

# Install official SDK
npm install @github/copilot-sdk

This SDK communicates via JSON-RPC with the Copilot CLI or its underlying runtime to orchestrate task planning, tool execution, and codebase edits.

🔗 Useful Links

GitHub Copilot Homepage →
GitHub Copilot CLI Repository →