Meta-Prompting: AutoGen to develop alphabet game by augment

When developing software—especially as codebases grow to thousands or even tens of thousands of lines—developers often encounter hard-to-fix bugs, or fixing one issue may introduce new problems. Even simple projects can require significant time to implement all desired features. Many existing AI programming tools, such as Cursor, are limited by their context window, typically supporting only about 10K tokens, which is roughly 400 to 600 lines of standard Python code. This short context length restricts the AI’s understanding of the entire project, making it difficult to identify key dependencies and architectural information. As a result, these tools can only handle single files or small-scale projects.

To address these pain points, today we introduce a more powerful AI-driven programming tool—Augment—integrated with AutoGen for complex project development.

We will also test Augment’s capabilities by developing a game: an English word game where players shoot down falling word clouds, and the results are saved locally at the end of the game.

Game Demo: Boomy Sky 蹦蹦貓

Augment 介紹

Augment is an AI programming platform designed for complex codebases, available as an IDE plugin for VS Code, JetBrains (including PyCharm, WebStorm, etc.), Vim, and Neovim. Augment excels at context processing and can also integrate with tools like GitHub, Jira, and Notion. It supports up to 200K tokens of context length—equivalent to about 8,000 to 10,000 lines of standard Python code. This enables Augment to better understand complex project codebases, identify file dependencies and call patterns, improve code quality, and reduce the risk of errors in production. Augment can extract detailed context from integration tools such as GitHub and GitLab, and performs exceptionally well in handling complex refactoring tasks, maintaining context, and truly understanding project structure. This results in better code awareness, smarter autocompletion, and even the ability to fulfill cross-file feature requests.
In this article, we’ll show you how to combine Augment with VS Code and the Context 7 assistant tool to automate the development of a 3D English word game.

AutoGen Introduction

AutoGen is a powerful AI programming framework designed to automate code development and optimization. It allows developers to define multiple roles, each with specific tasks and responsibilities. These roles can collaborate to achieve a more efficient code development and optimization workflow. In the following hands-on section, we will use the AutoGen architecture to define different roles and leverage Augment to automate the code development process.

Augment installation and Setup on VSCode

1. In VS Code, click the “Extensions” icon on the left sidebar.
2. Type “Augment” in the search box.
3. Find the Augment extension and click “Install.”
4. Once installed, click the Augment icon in the VS Code sidebar.
5. Click “Register and Sign In” to start using Augment.

Enhancing Augment with Context 7
To enable Augment to access the latest documentation and coding styles during game development, we can integrate it with Context 7. Context 7 is an MCP (Multi-Context Processor) tool that allows Augment to retrieve up-to-date documentation from open-source projects or libraries such as AutoGen.

Here’s how to add and configure Context 7 in Augment:

1. Click the settings icon in Augment.
2. Under the “Tools” section, click the “Add MCP” button.
3. In the pop-up window, set the name to context7.
4. In the “Command” field, enter npx -y @upstash/context7-mcp. Alternatively, you can use the following JSON configuration:

{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": [
        "-y",
        "@upstash/context7-mcp"
      ]
    }
  }
}
```json
{
  "mcpServers": {
    "context7": {
      "command": "npx",
      "args": [
        "-y",
        "@upstash/context7-mcp"
      ]
    }
  }
}

5. Click the blue “Add” button to successfully add Context 7.

Using Augment to Develop an English Word Game
Now, let’s use Augment and Context 7 to develop a 3D English word game.

Create a New Project

1. In the Augment interface, create a new project.
2. Enter the project name and click “Create.”

Enter Development Prompts
Next, we will input detailed prompts to let Augment understand our development requirements:

1. Specs and Task Lists

In the Augment input box, enter the following prompt:

Please create Specs.md and todolist.md according to the following specifications:

場景一開始是基本及輸入暱稱以及確認按鈕擊，背景是標題天空及雲，點擊確認後，進入遊戲，景，最底下有一個文字輸入匡，可以直接按 Enter送出；上面有草皮及砲塔，背景是天空空，以及上方有一個計分數字。遊戲開始後，會有單字雲從天空上持續掉下來，點擊砲塔可以輸入文字，送出後，砲塔會發射一個砲彈，擊中單字雲後，單字雲會消失，並且得分，遊戲目標是消除越來越多雲字雲。如果單字雲沒有被擊中，則會掉落到地面則結束遊，，並且將遊戲紀錄寫到本地 result.txt (只記錄前 50名)，會顯示得分和排名，以及重新開始按鈕。

In Augment, select Agent mode and enable Auto mode. Augment will automatically generate the Specs.md and todolist.md files.

我將根據您提供的遊戲規格建立 Specs.md 和 todolist.md 文件。讓我先分析一下這個單字雲射擊遊戲的需求，然後建立詳細的規格文件和待辦清單。

2. Meta-prompting Development According to Specifications

After reviewing and confirming that the specifications meet your expectations,
enter the following prompt in the Augment input box:

依照 Specs.md and todolist.md ，透過 Context 7 搜尋 tree.js 最新文檔，使用 3.js 開發一款  英語單字遊戲，使用 Context 7 搜尋最新文檔，以及參考 AutoGen 最新的發風格和類別庫編寫這個遊戲，包含以下角色：
- Agent1 負責開發撰寫程式碼，並且留著著註解
- Agent2 負責審核代碼品質，提出修改建議。
- Agent3 根據 Agent1 的程式碼和 Agent2 的建議進行修改及優化。
- Agent4 負責測試遊戲功能，確保遊戲能正常運行，及運行遊戲伺服器。

Next, enable the Auto feature, which will automatically search documentation and generate code.

• In the mode selection, switch to “agent” mode.
• Make sure the “Auto” option is enabled so that Augment will automate the game development process.
• Click send to let Augment start executing the tasks.

Once the code is generated, Augment will automatically check if the game runs correctly and launch the game server.

Game Showcase
After the server starts, you can test the game developed by Augment:
The most impressive part is that the game runs successfully on the first try without any errors. This demonstrates Augment’s powerful code generation capabilities and high first-time success rate.

Conclusion

Through this hands-on experience, we can clearly see the tremendous potential of Augment in complex project development. It not only solves many traditional programming challenges—such as hard-to-fix bugs caused by large codebases and context length limitations—but also achieves true “code awareness” and “automation” through its ultra-long context processing and cross-file dependency understanding.
Whether analyzing open-source projects, leveraging Context 7 to empower AI agents that can write and optimize code, or using 3.js to develop an aerial combat mini-game, Augment demonstrates outstanding performance and a fully automated development experience. The entire process requires almost no manual coding—just provide clear prompts, and Augment can complete complex development tasks, greatly improving development efficiency.
If you are looking for an AI programming tool that can boost productivity and reduce development difficulty, Augment is undoubtedly a powerful option worth trying!

附註：常見提示詞範例

在使用 Vibe coding 進行開發時，提供清晰的提示詞是非常重要的。以下是一些常見的提示詞範例，這些提示詞可以幫助我們更好地與 Agent 互動，獲取所需的程式碼和功能。

分析新代碼項目提示詞範例

以下是一些針對新代碼項目的分析提示詞範例，這些提示詞可以幫助我們更好地理解專案的架構、程式碼品質以及模塊化設計。

1. 請分析這個項目的整體架構和特點：

• 採用的設計模式和架構風格
• 主要技術依依賴項目
• 模塊及組織結構

2. 程式碼品質分析

• 程式碼品質分析
• 代碼可測試性
• 文檔完整性

3. 程式碼組織結構與模組化

• 模組間的耦合分析
• 接口設計的合理性
• 可拓展性和可維護性評估

分析新代碼項目提示詞範例

以下是針對新代碼項目的分析提示詞範例，這些提示詞可以幫助深入理解專案的結構和程式碼品質 ，並為後續的開發和維護提供有價值的參考。

分析項目請求：

1. 核心函式分析

• 列出主要函式入口處
• 辨識關鍵韓式和輔助函式
• 標注每個函式的作用和重要性

2. 函數調用鏈分析

• 繪製函式調用圖
• 分析函式間的耦合關係
• 分析函式調用方向及次數

3. 依賴模組分析

• 列出所有依賴模組的依賴關係
• 分析模組間的依賴關係
• 標注模組間的依賴方向(import/include)

請用以下格式輸出分析結果：

• 文字描述 + ASCII流程圖
• 或者提供 Mermaid/PlantUML 流程圖
• 按重要性排序函數列表