简体中文 | English

aicoding-cookbook 是一个面向 AI Coding 工作流的实用资料仓库，收集了本地配置资产、可复用技能和辅助脚本。它的定位不是应用、SDK 或框架，而是一套围绕真实操作场景整理出来的 cookbook，用于配置工具、复用技能，以及标准化日常的 AI 辅助开发流程。

• 使用 Claude Code、Codex CLI、Augment Context Engine 等 AI 编码工具的开发者
• 希望直接复用现成配置资产，而不是每次从零搭建环境的个人或团队
• 需要维护 Codex 技能、任务跟踪流程和技能打包流程的操作者

augment-mcp-config/ 目录提供了一套面向 Augment Context Engine MCP 的配置资料：

• Augment-MCP配置教程.md：中文配置指南，面向 AI 辅助执行场景，覆盖 Augment MCP 和 Codex MCP 在 Windows、macOS、Linux 上的配置流程
• augment.mjs：随指南一起提供的 augment.mjs 文件。根据指南说明，这是一个经过修改的 Auggie 入口文件，包含 codebase-retrieval 和 prompt-enhancer 能力

这个目录适合直接交给 AI 助手执行配置任务，减少手工逐步操作。

skills/codex/ 目录当前包含 3 个可复用技能包：

• taskmaster：多步骤任务跟踪技能，带任务规格、进度日志和基于 CSV 的里程碑管理
• todo-list-csv：轻量级 CSV 任务跟踪技能，用于把待办文件和 agent 计划保持同步
• skill-creator：用于创建、校验和打包新技能的说明与脚本集合

仓库里不仅有说明文档，也包含可直接复用的脚本和模板：

• skills/codex/todo-list-csv/scripts/todo_csv.py：Python CLI，用来创建和推进任务 CSV，支持 path、init、add、start、done、todo、advance、plan、status、cleanup 等命令
• skills/codex/skill-creator/scripts/init_skill.py：从模板初始化一个新技能目录
• skills/codex/skill-creator/scripts/package_skill.py：先校验技能，再把技能目录打包成 zip
• skills/codex/skill-creator/scripts/quick_validate.py：对技能 SKILL.md 的元数据做轻量校验
• skills/codex/taskmaster/assets/：可复用的 SPEC.md、PROGRESS.md、TODO.csv 模板

aicoding-cookbook/
├── README.md
├── augment-mcp-config/
│   ├── Augment-MCP配置教程.md
│   └── augment.mjs
└── skills/
    └── codex/
        ├── skill-creator/
        │   ├── LICENSE.txt
        │   ├── SKILL.md
        │   └── scripts/
        ├── taskmaster/
        │   ├── SKILL.md
        │   └── assets/
        └── todo-list-csv/
            ├── SKILL.md
            └── scripts/

1. 打开 augment-mcp-config/
2. 按 Augment-MCP配置教程.md 中的步骤执行
3. 只有在你的本地环境与文档假设一致时，才替换已安装环境中的 augment.mjs
4. 按文档说明在 Claude Code 或 Codex CLI 中配置对应 MCP 服务

1. 浏览 skills/codex/ 下的各个技能目录
2. 阅读每个 SKILL.md，确认其触发条件和工作流
3. 如果你需要复用其中的既定流程，优先直接使用随技能附带的脚本和模板

如果你只需要自动化部分，也可以直接使用这些脚本：

• todo_csv.py：CSV 任务推进
• init_skill.py：技能脚手架初始化
• package_skill.py：技能打包
• quick_validate.py：技能元数据快速校验

• Augment/Auggie 相关配置流程默认需要 Node.js 和 npm
• skills/codex/ 下的辅助脚本默认需要 Python 3
• Augment 配置指南默认你能够编辑本地工具配置文件，例如 ~/.claude.json 或 ~/.codex/config.toml

• 当前顶层 README 默认使用中文，便于中文读者直接阅读
• 仓库中仍然保留部分中文原始资料，例如 augment-mcp-config/Augment-MCP配置教程.md
• 向下滚动即可查看完整英文版说明

• 不要把真实 API token、私钥或机器相关凭据提交进这个仓库
• 当前 .gitignore 已忽略一个本地敏感路径：augment mcp setting/key.txt

• J3n5en 提供了指南中引用的修改版 augment.mjs
• Augment ACE MCP
• OpenAI Codex CLI

aicoding-cookbook is a practical repository of setup assets, reusable skills, and helper scripts for AI coding workflows. It is not an application, SDK, or framework. It is a working cookbook for configuring tools, reusing skill packages, and standardizing day-to-day AI-assisted development routines.

• Developers who use AI coding tools such as Claude Code, Codex CLI, and Augment Context Engine
• Individuals or teams who want reusable local setup assets instead of rebuilding toolchains from scratch
• Operators who maintain Codex skills, task-tracking flows, and skill packaging workflows

The augment-mcp-config/ directory contains a setup package for Augment Context Engine MCP:

• Augment-MCP配置教程.md: a Chinese-language setup guide for AI-assisted execution, covering Augment MCP and Codex MCP setup across Windows, macOS, and Linux
• augment.mjs: a bundled augment.mjs file referenced by the guide. The guide describes it as a modified Auggie entry file with codebase-retrieval and prompt-enhancer support

This folder is designed to be handed directly to an AI assistant so the assistant can perform most of the setup process.

The skills/codex/ directory currently contains three reusable skill packages:

• taskmaster: a multi-step task tracking skill with task specifications, progress logging, and CSV-backed milestone management
• todo-list-csv: a lightweight CSV task tracker that keeps a to-do file synchronized with agent planning
• skill-creator: a guide and script bundle for creating, validating, and packaging new skills

The repository includes reusable scripts and templates in addition to documentation:

• skills/codex/todo-list-csv/scripts/todo_csv.py: a Python CLI for creating and advancing task CSV files, with commands such as path, init, add, start, done, todo, advance, plan, status, and cleanup
• skills/codex/skill-creator/scripts/init_skill.py: initializes a new skill directory from a template
• skills/codex/skill-creator/scripts/package_skill.py: validates and packages a skill directory into a zip archive
• skills/codex/skill-creator/scripts/quick_validate.py: performs lightweight validation on SKILL.md metadata
• skills/codex/taskmaster/assets/: reusable templates for SPEC.md, PROGRESS.md, and TODO.csv

aicoding-cookbook/
├── README.md
├── augment-mcp-config/
│   ├── Augment-MCP配置教程.md
│   └── augment.mjs
└── skills/
    └── codex/
        ├── skill-creator/
        │   ├── LICENSE.txt
        │   ├── SKILL.md
        │   └── scripts/
        ├── taskmaster/
        │   ├── SKILL.md
        │   └── assets/
        └── todo-list-csv/
            ├── SKILL.md
            └── scripts/

1. Open augment-mcp-config/
2. Follow the steps in Augment-MCP配置教程.md
3. Replace the installed augment.mjs only if your local environment matches the assumptions in the guide
4. Configure the MCP server in Claude Code or Codex CLI as described in the document

1. Browse the folders under skills/codex/
2. Read each SKILL.md to understand its trigger conditions and workflow
3. Reuse the bundled scripts and templates when you want the exact packaged workflow

If you only want the automation pieces, the bundled scripts can also be used directly:

• todo_csv.py for CSV-backed task progression
• init_skill.py for skill scaffolding
• package_skill.py for skill packaging
• quick_validate.py for lightweight metadata validation

• Node.js and npm are expected for the Augment/Auggie-related setup flow
• Python 3 is expected for the helper scripts under skills/codex/
• The Augment guide assumes you are comfortable editing local tool configuration files such as ~/.claude.json or ~/.codex/config.toml

• This README defaults to Chinese for the primary reading experience
• Some bundled source materials are still Chinese-language documents, especially augment-mcp-config/Augment-MCP配置教程.md
• The English section mirrors the same repository facts for global readers

• Do not commit real API tokens, private keys, or machine-specific credentials into this repository
• The current .gitignore already excludes a local secret path: augment mcp setting/key.txt

• J3n5en for the modified augment.mjs referenced by the bundled guide
• Augment ACE MCP
• OpenAI Codex CLI