🤖 ai-agent-init

Bootstrap any project with AI-agent-ready documentation in seconds.

Stop wasting time explaining your project structure to every new AI Agent. Run one command, point your agent to one file, and get a fully documented, standardized project.

The Problem

Every time you start a new AI coding session, you waste tokens and time re-explaining:

• What tech stack you're using
• What architecture patterns to follow
• What naming conventions to use
• Where files should go

The Solution

ai-agent-init drops a smart bootstrap file into your project. You point your AI Agent to it, and the agent does the rest — generating tailored documentation for your specific codebase, then cleaning up after itself.

Usage

# 1. cd into any project (new or existing)
cd my-project

# 2. Run the initializer
npx ai-agent-init init

# 3. Tell your AI Agent (Cursor, Copilot, Cline, Gemini, etc.):
# "Read docs/ai/ai_bootstrap.md and set up this project."

That's it. The agent will:

• 🔍 Analyze your codebase and detect the tech stack
• 📝 Generate docs/ai/rules.md — conventions tailored to your project
• 🗺️ Generate docs/ai/project_map.md — a map of your folder structure
• 🗑️ Generate docs/ai/uninstall.md — guide to remove the project and free up system resources
• 📄 Create or update README.md with real project info
• ⚙️ Create .cursorrules and .clinerules at the root
• 🧹 Delete all template files automatically

Supported Tech Stacks

The bootstrap templates cover conventions for:

| Stack | Detected By | |---|---| | Flutter / Dart | pubspec.yaml | | Python | requirements.txt, pyproject.toml | | Node.js / TypeScript | package.json | | Go | go.mod | | Generic | Fallback for any other project |

What Gets Generated

your-project/
├── docs/
│   └── ai/
│       ├── rules.md          ← Code conventions & architecture for this project
│       ├── project_map.md    ← Folder structure map with descriptions
│       └── uninstall.md      ← Guide to remove project and free resources
├── README.md                 ← Updated with real project info
├── .cursorrules              ← For Cursor IDE
└── .clinerules               ← For Cline / Roo-Code

Why This Works

The key insight is: AI Agents are great at reading context and adapting. Instead of writing complex detection logic in the CLI, we give the agent a clear instruction file (ai_bootstrap.md) and let it figure out the specifics. This makes the tool:

• ✅ Adaptive — works with any language or framework
• ✅ Accurate — agent reads actual code, not just file names
• ✅ Lightweight — CLI has zero dependencies
• ✅ Self-cleaning — no template clutter after setup

Contributing

Contributions welcome! To add a new tech stack template:

1. Add a templates/rules_<tech>.md file
2. Reference it in templates/ai_bootstrap.md's detection table

License

MIT © ngocbaomobile