project-scaffold

Project Scaffold

Unified project scaffolder for creating new projects or adding components to existing ones.

Purpose

This skill provides a unified way to scaffold:

• .agents/ folder structure for AI-first development
• Backend (NestJS) with MongoDB, Swagger, and best practices
• Frontend (NextJS) with Tailwind and @agenticindiedev/ui
• Mobile (Expo) with React Native and Expo Router
• Browser Extension (Plasmo) with React and Tailwind

When to Use

Use this skill when:

• Starting a new project from scratch
• Adding components to an existing project
• Setting up a monorepo or separate repositories
• Need a unified scaffolding experience across Claude, Codex, and Cursor

Usage

Interactive Mode (Recommended)

python3 ~/.claude/skills/project-scaffold/scripts/scaffold.py

The script will ask you:

1. Project name
2. Project root path
3. Repository structure (monorepo vs separate)
4. Components to scaffold (all optional):
   • .agent folder
   • Backend (NestJS)
   • Frontend (NextJS)
   • Mobile (Expo)
   • Extension (Plasmo)
5. Organization name (for monorepo packages)

From Claude

When user requests scaffolding, activate this skill and run:

python3 ~/.claude/skills/project-scaffold/scripts/scaffold.py

The script handles all interactive prompts.

Features

Flexible Structure

• Monorepo: All components in one repository with workspace configuration
• Separate repos: Each component in its own directory/repository
• Existing projects: Can add components to existing projects

Component Options

All components are optional - scaffold only what you need:

• .agents/ folder with full AI documentation structure
• Backend with NestJS, MongoDB, Swagger
• Frontend with NextJS 15, Tailwind, @agenticindiedev/ui
• Mobile with Expo Router, React Native
• Extension with Plasmo, React, Tailwind

Cross-Platform

Works from:

• Claude Code
• Codex
• Cursor

Generated Structure

Monorepo Example

myproject/
├── .agents/                  # AI documentation
├── package.json             # Workspace root
├── api/                     # NestJS backend
│   ├── .agents/
│   ├── apps/api/src/
│   └── package.json
├── frontend/                # NextJS apps
│   ├── .agents/
│   ├── apps/dashboard/
│   └── package.json
├── mobile/                  # Expo app
│   ├── .agents/
│   ├── app/
│   └── package.json
└── extension/               # Plasmo extension
    ├── .agents/
    ├── src/
    └── package.json

Separate Repos Example

myproject-api/               # Backend only
├── .agents/
├── apps/api/src/
└── package.json

myproject-frontend/          # Frontend only
├── .agents/
├── apps/dashboard/
└── package.json

Key Patterns Included

Backend (NestJS)

• Soft deletes: isDeleted: boolean
• Multi-tenancy: Always filter by organization
• Collection pattern: controllers → services → schemas
• Swagger documentation
• Dockerfile included

Frontend (NextJS)

• Path aliases: @components/, @services/, @hooks/
• Tailwind CSS with @agenticindiedev/ui
• TypeScript strict mode
• App Router structure

Mobile (Expo)

• Expo Router for navigation
• TypeScript configuration
• Platform-specific configurations

Extension (Plasmo)

• React + TypeScript
• Tailwind CSS with @agenticindiedev/ui
• Manifest configuration
• Popup component

Next Steps After Scaffolding

1. Install dependencies:

   cd [project-root]
   bun install
   

2. Start development:

   # Monorepo
   bun run dev:api
   bun run dev:frontend
   
   # Separate repos
   cd api && bun run start:dev
   cd frontend && bun run dev
   

3. Customize:

   • Update .agents/SYSTEM/RULES.md with your coding standards
   • Configure environment variables
   • Add your first features

Integration with Existing Skills

This skill integrates with:

• agent-folder-init: Uses its script for .agent folder scaffolding
• fullstack-workspace-init: Reuses component templates and patterns

Troubleshooting

Script not found: Ensure the skill is installed in ~/.claude/skills/project-scaffold/

Permission denied: Make script executable: chmod +x ~/.claude/skills/project-scaffold/scripts/scaffold.py

Python not found: Ensure Python 3 is installed and in PATH

Agent folder not created: The script tries to use agent-folder-init skill. If not found, it will skip .agent folder creation.