Trae AI Agent Enhancements

English | 中文

About This Project

Trae AI Agent Enhancements is a Trae AI Agent rules and skills collection for Trae IDE. It provides Rule-based routing, professional Skills, and persistent Memory capabilities — transforming your Agent from "can do anything but unstable" to "fast when simple, rigorous when complex, always learning".

Architecture

User Input → Rules (Routing & Constraints) → Skills (Execution) → Memory (Learning)

Core Features

• T-Shirt Sizing: Automatic task classification (S/M/L) — small tasks are fast, large tasks are thorough
• Closed-Loop Quality: Every skill produces verifiable evidence — no "should be fine" allowed
• Self-Improvement: Lessons learned are stored in Core Memory — never repeat the same mistake
• Chinese Team Ready: Native Chinese triggers, domestic Git platform support, bilingual skill files
• Windows Native: PowerShell commands, port conflict recovery, path conventions

Quick Start

Method 1: npm install (Recommended)

Run this anywhere (installs to your home directory, global for all projects):

npx trae-agent-enhancements

Follow the interactive prompt to select your Trae edition:

You can also specify the edition directly with --edition:

npx trae-agent-enhancements --edition cn       # China edition
npx trae-agent-enhancements --edition intl     # International edition
npx trae-agent-enhancements --help             # View help

Prerequisites (MCP)

Some skills require the following MCP Servers for full functionality:

These MCPs power the everything-search, chrome-devtools, and memory-kernel skills. Without them, related skills will not work fully.

If you are using Trae Solo, you may already have built-in MCP Memory — check before installing manually.

MCP Memory Server config (add to Trae Settings → MCP):

{
  "mcpServers": {
    "memory": {
      "command": "cmd",
      "args": ["/c", "npx", "-y", "@modelcontextprotocol/server-memory"],
      "env": {
        "MEMORY_FILE_PATH": "D:/AppData/Memory/memory.jsonl"
      }
    }
  }
}

Method 2: Manual Install

# Clone the repository
git clone https://github.com/MorningStar0709/trae-agent-enhancements.git

# Choose your edition and copy to the corresponding directory

# China edition (trae.cn)
cp -r trae-agent-enhancements/rules  ~/.trae-cn/rules
cp -r trae-agent-enhancements/skills ~/.trae-cn/skills

# International edition (trae.ai)
cp -r trae-agent-enhancements/rules  ~/.trae/rules
cp -r trae-agent-enhancements/skills ~/.trae/skills

Try These Commands

"帮我排查这个报错" (Help me debug this error)
"先写计划再实现" (Write a plan first, then implement)
"改好了没？验证一下" (Is it fixed? Verify it)
"帮我提交" (Help me commit)
"记住这个处理方式" (Remember this approach)

Documentation

Developer References

Skill Paths

Windows/Trae Adaptation

• PowerShell Commands: Primary syntax for Windows environments
• Port Recovery: netstat → taskkill → verify → retry
• Path Conventions: Forward slashes in globs, absolute paths with backslashes
• Core Memory: 20-entry limit per scope, auto-eviction for stale entries
• MCP Memory: Knowledge graph persistence via MCP Memory Server, path D:/AppData/Memory/memory.jsonl

Acknowledgments

This project references superpowers-zh (AI 编程超能力 · 中文增强版) for its npx installation pattern and project structure design, and superpowers as the original upstream. Thanks to @jnMetaCode and @obra for their excellent work.

Contributing

Contributions are welcome! Please read CONTRIBUTING.md for guidelines.

For Chinese version, see CONTRIBUTING_zh.md.

License

This project is licensed under the MIT License.