SuperAPI AI Directory
Partner中文AI newsFavorites
Skillstrae-cn

trae-cn

Coding & development

View repository

Updated 10-16 07:23

About this skill

Trae CN AI IDE Integration

Trae is an AI-native integrated development environment (AI IDE) launched by ByteDance, deeply integrated with AI, providing intelligent Q&A, code auto-completion, and Agent-based AI automatic programming capabilities.

Quick Start

Launch Trae IDE

# 使用 Homebrew 启动
brew services start trae-cn

# 或者直接启动应用
open -a "Trae"

Trae SOLO Mode

Trae's SOLO mode can understand developer needs and automatically execute programming tasks:

# 进入项目目录后启动 Solo 模式
cd your-project
trae solo --task "帮我优化这个函数的性能"

Main Features

1. AI Programming Assistant

  • Intelligent Q&A: Ask AI directly in the IDE
  • Code Generation: Describe requirements, automatically generate code
  • Code Explanation: Select code, AI explains the logic
  • Unit Testing: Automatically generate test cases
  • Intelligent Fixing: AI automatically fixes bugs

2. MCP Integration (Model Context Protocol)

MCP is a protocol that enables AI tools to understand and access external knowledge bases or services.

Configure MCP Server

Configure MCP in Trae:

  1. Open Trae IDE
  2. Go to Settings → MCP
  3. Add MCP Server configuration

Built-in MCP Server

Trae has built-in commonly used MCP Servers:

  • File System: File read and write operations
  • Database: Database queries
  • API Calls: HTTP requests
  • Web Automation: Playwright integration

Custom MCP Server

# 安装 MCP Server
npm install -g @modelcontextprotocol/server-filesystem

# 配置到 Trae
trae config add-mcp @modelcontextprotocol/server-filesystem

3. Agent Mode

Use Agent mode to let AI automatically complete complex tasks:

# 启动 Agent
trae agent --task "创建一个 React 组件，包含表单验证功能"

# 带上下文
trae agent --task "分析这个项目的架构" --context ./project

Common Commands

Project Management

# 创建新项目
trae create --template react my-project

# 打开项目
trae open ./my-project

# 克隆并打开项目
trae clone https://github.com/user/repo.git

Code Operations

# 代码生成
trae generate --prompt "创建一个用户登录组件"

# 代码重构
trae refactor --file ./src/App.tsx --task "将类组件改为函数组件"

# 代码审查
trae review --file ./src/utils.ts

AI Conversation

# 启动对话
trae chat

# 带系统提示
trae chat --system "你是一个 Python 专家"

# 引用代码对话
trae chat --file ./script.py "解释这段代码"

MCP Operations

# 列出已配置的 MCP Server
trae mcp list

# 添加 MCP Server
trae mcp add filesystem -- npx @modelcontextprotocol/server-filesystem ./projects

# 移除 MCP Server
trae mcp remove filesystem

Integration with OpenClaw

Using Trae MCP Server

In OpenClaw, you can call Trae through the MCP protocol:

# Python 示例：使用 Trae MCP Server
import requests

# 调用 Trae MCP 文件系统操作
response = requests.post(
    "http://localhost:3000/mcp/file",
    json={
        "action": "read",
        "path": "./src/App.tsx"
    }
)

OpenClaw Trae Skill

Using OpenClaw's trae-cn Skill:

# 在 OpenClaw 中使用
@trae-cn help

# 创建项目
@trae-cn create-project --name my-app --template react

# 运行代码生成
@trae-cn generate --prompt "创建一个登录表单"

Configuration Options

Environment Variables

# API Key 配置
export TRAE_API_KEY="your-api-key"

# 模型选择
export TRAE_MODEL="deepseek-v3"  # 默认
# 或 claude-sonnet, gpt-4

# MCP Server 默认路径
export TRAE_MCP_PATH="./mcp-servers"

Configuration File

The Trae configuration file is located at ~/.trae/config.json:

{
  "model": {
    "provider": "deepseek",
    "model": "deepseek-v3",
    "temperature": 0.7
  },
  "mcp": {
    "enabled": true,
    "servers": ["filesystem", "database"]
  },
  "theme": "dark",
  "shortcuts": {
    "chat": "cmd+shift+l",
    "generate": "cmd+shift+g"
  }
}

Use Cases

Scenario 1: Rapid Prototyping

# 1. 创建项目
trae create --template nextjs my-saas

# 2. 进入项目
cd my-saas

# 3. 启动开发
trae dev

# 4. AI 生成代码
trae generate --prompt "创建一个用户仪表盘页面"

Scenario 2: Code Review and Optimization

# 1. 打开项目
trae open ./my-project

# 2. AI 审查代码
trae review --file ./src/critical.ts

# 3. AI 优化
trae refactor --file ./src/critical.ts --task "性能优化"

Scenario 3: Integrating External Tools

# 1. 配置 MCP Server
trae mcp add weather -- npx @modelcontextprotocol/server-weather

# 2. 在对话中使用
trae chat "北京今天天气怎么样？"

Scenario 4: Team Collaboration

# 1. 克隆项目
trae clone https://github.com/team/project.git

# 2. 设置上下文
trae context set ./project-docs

# 3. AI 理解项目
trae chat "根据项目文档，帮我生成 API 文档"

FAQ

Q: What is the difference between Trae and VS Code? A: Trae is a native AI IDE, deeply integrated with AI, supporting Agent mode and the MCP protocol, with stronger AI capabilities than VS Code.

Q: Which programming languages does Trae support? A: It supports all mainstream languages, including Python, JavaScript/TypeScript, Go, Rust, Java, C++, etc.

Q: How do I configure a custom MCP Server? A: Refer to the official documentation:https://docs.trae.com.cn/ide/model-context-protocol

Q: Is Trae free? A: Trae offers a free version and a professional version. The free version includes basic AI features, and the professional version provides more advanced features.

Q: How do I switch AI models? A: Configure in settings or use environment variables TRAE_MODEL

Reference Links