想要主動寫程式的功能

[kkdai]

透過 PRD 與主要的 task 來開發相關功能

[kkdai]

@agent-fleet need_prd

[agent-fleet]

PRD (Product Requirements Document)

---

Product Requirements Document: GitHub PRD Bot

1. Introduction

• 1.1 Purpose: This document outlines the requirements for the GitHub PRD Bot, a tool designed to automate the creation of Product Requirements Documents (PRDs) and break down development tasks from GitHub Issues.
• 1.2 Goals:
  • Automate the generation of PRDs for new GitHub Issues.
  • Automatically break down PRDs into executable sub-tasks.
  • Improve efficiency and clarity in the software development process.
  • Reduce manual effort in defining requirements and tasks.
• 1.3 Target Audience: This document is intended for developers, product managers, QA engineers, and other stakeholders involved in the development and deployment of the GitHub PRD Bot.

2. Background

Currently, writing PRDs and breaking down tasks requires significant manual effort. This process can be time-consuming and prone to inconsistencies. The GitHub PRD Bot aims to address this problem by leveraging Google Gemini AI to automatically generate PRDs from Issue descriptions and decompose them into actionable sub-tasks. This automation will streamline the development workflow, improve communication, and increase overall team efficiency.

3. Goals

• Automated PRD Generation: Automatically generate comprehensive PRDs for GitHub Issues using Issue content and project README.
• Automated Sub-task Generation: Automatically break down generated PRDs into a checklist of actionable development sub-tasks.
• Reduced Manual Effort: Significantly reduce the time and effort required for product managers and developers to define requirements and tasks.
• Improved Documentation Quality: Ensure consistent and high-quality documentation for each feature or bug fix.
• Enhanced Collaboration: Facilitate better communication and collaboration among team members by providing clear and structured requirements.

4. User Stories

• As a Product Manager, I want the bot to automatically generate a PRD for new GitHub issues so that I can quickly review and refine the requirements, saving me time.
• As a Developer, I want the bot to break down the PRD into actionable sub-tasks so that I have a clear understanding of what needs to be done, enabling me to start development faster.
• As a Project Lead, I want the bot to generate consistent PRDs across all issues so that I can ensure a standard level of documentation and improve project organization.
• As a QA Engineer, I want to easily access the PRD and sub-tasks so that I can understand the scope of testing and create relevant test cases.

5. Requirements

• 5.1 Functional Requirements:

  • 5.1.1 PRD Generation:
    • Automatic Trigger: The bot shall automatically generate a PRD when a new Issue is created.
    • Manual Trigger: The bot shall generate a PRD when mentioned in an Issue comment with the command @<bot-name> need_prd.
    • Data Input: The bot shall read the Issue title, body, and the project's README.md file as input.
    • PRD Generation Engine: The bot shall use Google Gemini AI to generate a PRD in English based on the input data. The PRD should follow the specified structure (Background, Goals, User Stories, Requirements, Success Metrics).
    • Language Detection & Translation: The bot shall detect the primary language of the Issue content and translate the generated English PRD into that language.
    • PRD Output: The bot shall post a comment in the Issue containing both the English PRD and the translated PRD.
  • 5.1.2 Sub-task Generation:
    • Manual Trigger: The bot shall generate sub-tasks when mentioned in an Issue comment with the command @<bot-name> need_sub_task.
    • PRD Retrieval: The bot shall search the Issue's comments to find the latest PRD document.
    • Sub-task Generation Engine: The bot shall use Google Gemini AI to break down the PRD into a list of executable development sub-tasks.
    • Sub-task Output: The bot shall post a comment in the Issue containing the sub-task list formatted as a Markdown checklist.
  • 5.1.3 Error Handling:
    • The bot shall gracefully handle errors such as missing API keys, invalid input data, or failure to connect to Google Gemini AI.
    • The bot shall log errors and provide informative messages to users.

• 5.2 Non-Functional Requirements:

  • 5.2.1 Performance:
    • PRD generation should complete within a reasonable timeframe (e.g., less than 60 seconds).
    • Sub-task generation should complete within a reasonable timeframe (e.g., less than 30 seconds).
  • 5.2.2 Scalability: The bot should be able to handle a large number of Issues and requests without performance degradation.
  • 5.2.3 Security:
    • The bot shall securely store and manage API keys and other sensitive data.
    • The bot shall adhere to GitHub's API rate limits and security best practices.
  • 5.2.4 Availability: The bot should be available and responsive most of the time.
  • 5.2.5 Maintainability: The codebase should be well-structured, documented, and easy to maintain.
  • 5.2.6 Reliability: The bot should consistently generate accurate and relevant PRDs and sub-tasks.

6. Success Metrics

• Adoption Rate: Percentage of new GitHub Issues with automatically generated PRDs.
• User Satisfaction: Measured through surveys or feedback mechanisms, assessing users' satisfaction with the quality and usefulness of the generated PRDs and sub-tasks.
• Time Savings: Reduction in the average time spent writing PRDs and breaking down tasks manually.
• Issue Resolution Time: Observe if using the bot leads to faster issue resolution times.
• Documentation Completeness: Assess the completeness and accuracy of PRDs generated by the bot compared to manually created PRDs.
• Error Rate: Monitor and minimize the number of errors encountered by the bot during PRD and sub-task generation.

7. Deployment and Configuration

• 7.1 Environment: The bot should be deployable as a containerized service on platforms like Google Cloud Run, Heroku, or Fly.io.
• 7.2 GitHub App Configuration: The bot requires setup as a GitHub App with appropriate permissions (Issues: Read & write, Contents: Read-only).
• 7.3 Environment Variables: The following environment variables are required for the bot to function correctly:
  • GITHUB_APP_ID: GitHub App ID.
  • GITHUB_APP_NAME: GitHub App Name.
  • GITHUB_WEBHOOK_SECRET: GitHub Webhook Secret.
  • GOOGLE_API_KEY: Google AI API Key.
  • GITHUB_APP_PRIVATE_KEY: Base64 encoded GitHub App Private Key.
• 7.4 Webhook Configuration: The GitHub App Webhook URL should be configured to point to the deployed service's /webhook endpoint.

8. Future Considerations

• Customizable PRD Templates: Allow users to customize the PRD template used by the bot.
• Integration with Project Management Tools: Integrate the bot with project management tools like Jira or Asana to automatically create tasks from the generated sub-tasks.
• Improved AI Model Training: Continuously train the AI model with feedback from users to improve the quality and relevance of the generated PRDs and sub-tasks.
• Support for other Languages: Expand language support for PRD generation and translation beyond English and the Issue's primary language.
• Automated Testing: Implement automated tests to ensure the bot's functionality and reliability.

---

PRD (Traditional Chinese)

Traditional Chinese Translation:

產品需求文件：GitHub PRD 機器人

1. 介紹

• 1.1 目的： 本文件概述 GitHub PRD 機器人的需求，該工具旨在自動化產品需求文件 (PRD) 的創建，並從 GitHub Issues 中分解開發任務。
• 1.2 目標：
  • 自動化為新 GitHub Issues 生成 PRD。
  • 自動將 PRD 分解為可執行的子任務。
  • 提高軟體開發過程的效率和清晰度。
  • 減少定義需求和任務的手動工作。
• 1.3 目標受眾： 本文件適用於開發人員、產品經理、品質保證工程師以及參與 GitHub PRD 機器人開發和部署的其他利益相關者。

2. 背景

目前，編寫 PRD 和分解任務需要大量的手動工作。 這個過程可能非常耗時並且容易出現不一致。 GitHub PRD 機器人旨在通過利用 Google Gemini AI 從 Issue 描述自動生成 PRD，並將其分解為可操作的子任務來解決這個問題。 這種自動化將簡化開發工作流程、改善溝通並提高團隊整體效率。

3. 目標

• 自動化 PRD 生成： 使用 Issue 內容和項目 README 自動為 GitHub Issues 生成全面的 PRD。
• 自動化子任務生成： 自動將生成的 PRD 分解為可操作的開發子任務清單。
• 減少手動工作： 大幅減少產品經理和開發人員定義需求和任務所需的時間和精力。
• 提高文件品質： 確保每個功能或錯誤修復都有一致且高品質的文件。
• 加強協作： 通過提供清晰且結構化的需求，促進團隊成員之間更好的溝通和協作。

4. 使用者故事

• 身為產品經理，我希望機器人能夠自動為新的 GitHub issues 生成 PRD，這樣我就可以快速審閱和完善需求，節省我的時間。
• 身為開發人員，我希望機器人能夠將 PRD 分解為可操作的子任務，這樣我就能清楚地了解需要做什麼，使我能夠更快地開始開發。
• 身為專案負責人，我希望機器人能夠在所有 issues 中生成一致的 PRD，這樣我可以確保文件標準水平，並改善專案組織。
• 身為品質保證工程師，我希望可以輕鬆訪問 PRD 和子任務，這樣我就可以了解測試範圍並創建相關的測試案例。

5. 需求

• 5.1 功能需求：

  • 5.1.1 PRD 生成：
    • 自動觸發： 當創建新 Issue 時，機器人應自動生成 PRD。
    • 手動觸發： 當在 Issue 評論中提及機器人並使用命令 @<bot-name> need_prd 時，機器人應生成 PRD。
    • 數據輸入： 機器人應讀取 Issue 標題、正文和專案的 README.md 檔案作為輸入。
    • PRD 生成引擎： 機器人應使用 Google Gemini AI 根據輸入數據以英文生成 PRD。 PRD 應遵循指定的結構（背景、目標、使用者故事、需求、成功指標）。
    • 語言檢測與翻譯： 機器人應檢測 Issue 內容的主要語言，並將生成的英文 PRD 翻譯成該語言。
    • PRD 輸出： 機器人應在 Issue 中發布一則評論，其中包含英文 PRD 和翻譯後的 PRD。
  • 5.1.2 子任務生成：
    • 手動觸發： 當在 Issue 評論中提及機器人並使用命令 @<bot-name> need_sub_task 時，機器人應生成子任務。
    • PRD 檢索： 機器人應搜尋 Issue 的評論以找到最新的 PRD 文件。
    • 子任務生成引擎： 機器人應使用 Google Gemini AI 將 PRD 分解為可執行的開發子任務清單。
    • 子任務輸出： 機器人應在 Issue 中發布一則評論，其中包含格式化為 Markdown 清單的子任務清單。
  • 5.1.3 錯誤處理：
    • 機器人應優雅地處理錯誤，例如缺少 API 金鑰、無效的輸入數據或無法連接到 Google Gemini AI。
    • 機器人應記錄錯誤並向使用者提供資訊豐富的訊息。

• 5.2 非功能需求：

  • 5.2.1 效能：
    • PRD 生成應在合理的時間範圍內完成（例如，少於 60 秒）。
    • 子任務生成應在合理的時間範圍內完成（例如，少於 30 秒）。
  • 5.2.2 可擴展性： 機器人應能夠處理大量的 Issues 和請求，而不會降低效能。
  • 5.2.3 安全性：
    • 機器人應安全地儲存和管理 API 金鑰和其他敏感數據。
    • 機器人應遵守 GitHub 的 API 速率限制和安全性最佳實踐。
  • 5.2.4 可用性： 機器人應在大部分時間內可用且響應迅速。
  • 5.2.5 可維護性： 代碼庫應結構良好、有完善的文檔且易於維護。
  • 5.2.6 可靠性： 機器人應始終如一地生成準確且相關的 PRD 和子任務。

6. 成功指標

• 採用率： 具有自動生成 PRD 的新 GitHub Issues 的百分比。
• 使用者滿意度： 通過調查或回饋機制衡量，評估使用者對生成的 PRD 和子任務的品質和有用性的滿意度。
• 節省時間： 手動編寫 PRD 和分解任務所花費的平均時間減少。
• Issue 解決時間： 觀察使用機器人是否會縮短 issue 解決時間。
• 文件完整性： 評估機器人生成的 PRD 與手動創建的 PRD 相比的完整性和準確性。
• 錯誤率： 監控並最大限度地減少機器人在 PRD 和子任務生成期間遇到的錯誤數量。

7. 部署和配置

• 7.1 環境： 機器人應可作為容器化服務部署在 Google Cloud Run、Heroku 或 Fly.io 等平台上。
• 7.2 GitHub 應用程式配置： 機器人需要設置為具有適當權限的 GitHub 應用程式（Issues：讀寫權限，Contents：唯讀權限）。
• 7.3 環境變數： 機器人正常運作需要以下環境變數：
  • GITHUB_APP_ID：GitHub 應用程式 ID。
  • GITHUB_APP_NAME：GitHub 應用程式名稱。
  • GITHUB_WEBHOOK_SECRET：GitHub Webhook 密碼。
  • GOOGLE_API_KEY：Google AI API 金鑰。
  • GITHUB_APP_PRIVATE_KEY：Base64 編碼的 GitHub 應用程式私鑰。
• 7.4 Webhook 配置： GitHub 應用程式 Webhook URL 應配置為指向已部署服務的 /webhook 端點。

8. 未來考量

• 可自定義的 PRD 模板： 允許使用者自定義機器人使用的 PRD 模板。
• 與專案管理工具的集成： 將機器人與 Jira 或 Asana 等專案管理工具集成，以自動從生成的子任務創建任務。
• 改進 AI 模型訓練： 通過使用者的回饋不斷訓練 AI 模型，以提高生成的 PRD 和子任務的品質和相關性。
• 支援其他語言： 將 PRD 生成和翻譯的語言支援擴展到英語和 Issue 的主要語言之外。
• 自動化測試： 實施自動化測試以確保機器人的功能和可靠性。

[kkdai]

@agent-fleet need_sub_task

[agent-fleet]

Generated Sub-tasks

Based on the PRD, here are the suggested sub-tasks:

• Develop GitHub App: Create a GitHub App with necessary permissions (Issues: Read & Write, Contents: Read-only). Obtain GITHUB_APP_ID, GITHUB_APP_NAME, GITHUB_WEBHOOK_SECRET, and GITHUB_APP_PRIVATE_KEY.

• Set up Google Cloud Functions (or equivalent): Deploy the bot as a serverless function on a chosen platform (Google Cloud Run, Heroku, Fly.io, etc.). Configure environment variables: GOOGLE_API_KEY, GITHUB_APP_ID, GITHUB_APP_NAME, GITHUB_WEBHOOK_SECRET, and GITHUB_APP_PRIVATE_KEY.

• Implement Webhook Endpoint (/webhook): Create the /webhook endpoint to receive GitHub events (new issue creation, issue comments).

• Implement PRD Generation (English): Develop the function to generate an English PRD using Google Gemini API based on Issue title, body, and project's README.md. Structure the PRD according to the specified format (Background, Goals, User Stories, Requirements, Success Metrics).

• Implement Language Detection and Translation: Integrate a language detection and translation service (e.g., Google Translate API) to translate the English PRD into the Issue's primary language.

• Implement PRD Posting: Develop the functionality to post the English and translated PRDs as a comment on the GitHub Issue.

• Implement Sub-task Generation: Develop the function to extract the PRD from Issue comments and generate a Markdown checklist of actionable sub-tasks using Google Gemini API.

• Implement Sub-task Posting: Develop the functionality to post the generated sub-task checklist as a comment on the GitHub Issue.

• Implement Error Handling: Implement robust error handling for all API calls (GitHub and Google Gemini), including informative error messages to the user and logging for debugging.

• Implement Manual Trigger Functionality: Implement command processing for @<bot-name> need_prd and @<bot-name> need_sub_task commands in issue comments.

• Develop Unit and Integration Tests: Write comprehensive unit and integration tests to cover all functionalities, including PRD generation, translation, sub-task generation, and error handling.

• Implement Security Best Practices: Securely store and manage API keys and other sensitive data. Adhere to GitHub's API rate limits and security best practices.

• Implement Logging and Monitoring: Implement comprehensive logging and monitoring to track bot performance, errors, and usage statistics.

• Develop Deployment Script: Create a deployment script to automate the deployment process to the chosen platform.

• Create Documentation: Create comprehensive documentation for the bot, including setup instructions, usage examples, and troubleshooting guides. This should include both English and Traditional Chinese versions.

• User Acceptance Testing (UAT): Conduct UAT with product managers, developers, and QA engineers to validate the functionality and usability of the bot.

• Performance Testing: Perform load testing to ensure scalability and responsiveness under expected usage conditions.

This breakdown provides a detailed list of actionable sub-tasks. The order might need adjustments based on dependencies and team preferences. Remember to prioritize tasks based on risk and dependency.