CloudBase AI CLI Usage Documentation

Introduction

CloudBase AI CLI is a unified command-line tool that integrates multiple mainstream AI programming tools, supporting both built-in and custom models.

enables you to use AI programming assistants such as CodeBuddy Code, Claude Code, OpenAI Codex, aider, and Qwen Code with a simple command.

It also features built-in CloudBase MCP, supporting the complete workflow from development to deployment and enabling operation in any environment.

Core Advantages

• 🔧 Unified Management - Manage multiple AI programming CLI tools with a single command, without switching between multiple tools
• 🤖 Multi-Model Support - Supports built-in and custom large models, including Kimi K2, Zhipu GLM-4.5, etc.
• 🚀 One-Click Development & Deployment - Complete process from code generation to cloud deployment, supporting Web applications, Mini Programs, and backend services
• 🌍 Ubiquitous - Runs in any environment, including Mini Program Developer Tools, VS Code, GitHub Actions, etc.

Application Scenarios

Quick Start

1. Install

Mac/Linux/Windows WSL

Recommended: one-click installation script

Installation can be completed in just seconds (compared to npm installation which typically takes minutes), delivering a faster and more convenient experience:

curl https://static.cloudbase.net/cli/install/install.sh  -fsS | bash

Windows PowerShell

irm https://static.cloudbase.net/cli/install/install.ps1 | iex

npm/yarn/pnpm

If you need to install using npm/yarn/pnpm:

npm i -g @cloudbase/cli

or use Yarn 1.x:

yarn global add @cloudbase/cli

or Pnpm:

pnpm add -g @cloudbase/cli

If you encounter issues accessing the official npm source, typically during installation with errors such as "ETIMEOUT" or "npm ERR! network request to https://registry.npmjs.org/@cloudbase%2fcli failed", temporarily specify a domestic npm mirror source:

npm i -g @cloudbase/cli --registry=http://mirrors.cloud.tencent.com/npm/

yarn global add @cloudbase/cli --registry=http://mirrors.cloud.tencent.com/npm/

2. Initialize AI Configuration

Run CloudBase AI CLI in the terminal:

# Start Configuration Wizard
tcb ai

On first startup (when no AI config is detected), the setup wizard will guide you to:

• Choose a default AI tool
• Complete that tool’s configuration (for example, CloudBase one-click login or custom API settings)
• Optionally configure MCP / Skills

After setup, you can start using AI tools for assisted development. Later, you can run tcb ai --setup to reconfigure the current tool and update the default tool, MCP / Skills, and config scope (--scope project|global|custom) as needed.

Use --setup for one-click MCP and Skill setup

tcb ai --setup can complete MCP and Skill installation in one setup flow:

# Interactive setup (recommended for first-time use)
tcb ai --setup

# Non-interactive one-liner (example: CodeBuddy + project scope)
tcb ai --setup -a codebuddy --scope project -y

Common options (for tcb ai --setup when configuring AI tool, MCP, and Skills):

• --setup: start the setup wizard (configure AI tool, MCP, and Skills)
• -a, --agent <agent>: specify the AI tool to configure; allowed values are claude, qwen, codex, aider, codebuddy, cursor
• --scope <scope>: set the MCP/Skills config target scope; allowed values are project, global, custom
• --dir <dir>: set the MCP/Skills config directory; required when --scope custom is used
• -y, --yes: non-interactive mode with default choices

aider does not natively support MCP. When using --setup, MCP steps are skipped automatically and only Skills are installed.

3. Commence Experience

tcb ai

# Or Preset Prompts
tcb ai -- "help me create an order management system"

# Or use non-interactive mode
tcb ai -- -p "Hello"

# List all available commands
tcb ai --help

Detailed Development Guide

Refer to the detailed tutorial: Using CloudBase AI CLI to develop a mini program for recycling idle items in the neighborhood

Try Now

Run Anywhere, from Code Generation to Cloud Deployment!

No API Key required. The new environment comes with 1 million tokens, supporting quick access to mainstream models like DeepSeek and Kimi K2.

How It Works

Detailed Usage Instructions

Launch AI Tool

# AI Tool with Default Configuration
tcb ai

# Specify AI tool
tcb ai -a codebuddy
tcb ai -a claude
tcb ai -a codex
tcb ai -a aider
tcb ai -a qwen

Parameter Passthrough

Pass parameters directly to the target AI tool using the -- delimiter:

# Pass Parameters to CodeBuddy
tcb ai -a codebuddy -- --mcp-server "config.json"

# Pass Parameters to Claude
tcb ai -a claude -- -p hi

# Equivalent to running codebuddy --mcp-server "config.json" or claude -p hi

Specify Template

Directly specify the template during AI development:

# Use AI Command and Download Templates
tcb ai --template rules
tcb ai --template miniprogram
tcb ai --template react

Supported templates

• miniprogram - WeChat Mini Program + CloudBase
• react - Web application - React + CloudBase
• vue - Web application - Vue + CloudBase
• uniapp - Cross-platform application - UniApp + CloudBase
• rules - AI rules and configurations
• none - Do not download any templates

View current configuration

tcb ai --config

Reconfigure

# Reset all configurations
tcb ai --reset

# Run Configuration Wizard
tcb ai --setup

Supported AI Tools

AI Tool	Command Parameter
CodeBuddy Code	-a codebuddy
Claude Code	-a claude
Openai Codex	-a codex
aider	-a aider
Qwen Code	-a qwen

Getting Help

Command Line Help

# View command help
tcb ai --help

Frequently Asked Questions

Having issues? Refer to CloudBase CLI FAQ for detailed help.

Technical Support

• Help Center: https://tcb.cloud.tencent.com/help-center - Online consultation, ticket system, technical documentation
• Official Documentation: https://docs.cloudbase.net - Complete product documentation and usage guides
• GitHub Repository: CloudBase-AI-ToolKit - Open-source code and issue reporting

🔥 WeChat Discussion Group

Encountered any issues or want to exchange experiences? Join our technical community!

Scan the QR code to join the WeChat technical discussion group

Why Choose CLI?

CLI turns AI programming capabilities into programmable infrastructure that can be invoked by scripts, automated processes, and CI/CD systems, eliminating the need to switch between multiple GUI tools and enabling integration into existing development workflows.

Supporting cloud-native development, containerized deployment, and continuous integration/continuous deployment (CI/CD), propelling AI evolution from "code generation" to "full-process collaboration"

Configuration Instructions

Currently, CloudBase AI CLI supports configuration for the following AI tools:

• CodeBuddy Code
• Claude Code
• Openai Codex
• aider
• Qwen Code

All tools support two configuration methods:

• Configure a custom LLM: usually requires configuring your own API KEY and the corresponding Base URL
• Configure LLM in the CloudBase environment: No need to configure the API KEY by yourself, just log in and select the CloudBase environment by following the command-line prompts

AI Tools	Custom Configuration Requirements	CloudBase Environment Configuration	Related Documentation
CodeBuddy Code	- API KEY - Support OAuth authentication - MCP server configuration - Environment variable configuration	✨ One-click login, automatic configuration	Configuration document
Claude Code	- Base URL - API KEY - Requires configuration for Anthropic-compatible models - Supports configuration for Kimi-K2, Zhipu GLM-4.5	✨ One-click login, automatic configuration	Configuration document
OpenAI Codex	- Base URL - API KEY - Model name - Requires configuration for OpenAI-compatible models	✨ One-click login, automatic configuration	Configuration document
aider	- Base URL - API KEY - Model name - Requires configuration for OpenAI-compatible models	✨ One-click login, automatic configuration	Configuration document
Qwen Code	- Base URL - API KEY - Model name - Requires configuration for OpenAI-compatible models	✨ One-click login, automatic configuration	Configuration document