prompt-engineering

Installation
CLI
npx skills add https://github.com/martinholovsky/claude-skills-generator --skill prompt-engineering

Installez cette compétence avec la CLI et commencez à utiliser le flux de travail SKILL.md dans votre espace de travail.

Dernière mise à jour le 4/22/2026

Claude AI Skill Generator

A comprehensive template repository system for bootstrapping AI-assisted development projects with Claude Code integration. This repository provides production-ready templates, domain-specific skills, and automated setup scripts to quickly initialize new projects optimized for AI-assisted development.

🚀 Quick Start

# Clone this repository
git clone https://github.com/martinholovsky/claude-skills-generator.git
cd claude-skills-generator

# Run the setup script (interactive mode)
./setup-repo-multitype.sh /path/to/your/new-project

# Or use the original script
./init-new-repo.sh /path/to/your/new-project

The script will guide you through:

  • Project type selection (Web, Backend, Full-Stack, Desktop, CLI, Library)
  • Language selection (TypeScript, Python, Rust, etc.)
  • Feature selection and skill copying
  • Template file generation with your project details
  • Git repository initialization

📋 What's Included

1. Interactive Setup Scripts

  • setup-repo-multitype.sh - Enhanced setup with multi-type support
  • init-new-repo.sh - Original comprehensive setup script

Both scripts provide:

  • Interactive project configuration
  • Automatic skill selection based on project type
  • Template variable substitution
  • Git repository initialization
  • Claude Code configuration

2. Production-Ready Templates

All template files use placeholders (e.g., {{PROJECT_NAME}}) that get replaced during setup:

  • CLAUDE.md - Claude Code project instructions (auto-read by Claude)
  • README.md - Project documentation template
  • CONTRIBUTING.md - Contribution guidelines with AI-assisted workflows
  • SECURITY.md - Security policy with AI-specific considerations
  • CODE_OF_CONDUCT.md - Community code of conduct
  • .gitignore - Comprehensive ignore patterns
  • .env.example - Environment variable template

3. Claude Code Integration

The .claude/ directory contains:

Configuration:

  • settings.json - Hooks configuration (pre/post tool use, formatting)

Slash Commands (user-invoked):

  • /implement-feature - Feature implementation workflow
  • /run-tests - Automated test execution
  • /review-code - Code review checklist

4. GitHub Templates

The .github/ directory includes:

  • workflows/ci.yml - CI/CD pipeline (lint, test, build, security audit)
  • PULL_REQUEST_TEMPLATE.md - PR template with AI disclosure section
  • Issue Templates:
    • bug_report.md
    • feature_request.md

5. Domain-Specific Skills (67+ Skills)

Specialized knowledge modules for specific technologies. See SKILLS_REFERENCE.md for the complete catalog.

Categories:

  • Frontend/UI (11): Vue, Nuxt, Pinia, Tailwind, GSAP, Three.js, WebGL, Design Systems, UI/UX
  • Backend/API (15): FastAPI, GraphQL, REST, WebSocket, JSON-RPC, async, Celery, RabbitMQ, MCP
  • Languages (4): TypeScript, JavaScript, Python, Rust
  • Databases (5): SQLite, SQLCipher, SurrealDB, Graph Databases, Database Design
  • Security (6): AppSec, Security Auditing, Encryption, OS Keychain, Sandboxing, DevSecOps
  • DevOps (8): CI/CD, ArgoCD, Cilium, Harbor, Talos OS, Kanidm, Cloud APIs, Cross-Platform Builds
  • Desktop (3): Tauri, Browser Automation, Auto-Update Systems
  • Platform-Specific (7): macOS, Windows, Linux accessibility and automation
  • AI/ML (4): LLM Integration, Model Quantization, Prompt Engineering, Speech/Wake-Word
  • Multimedia (2): Web Audio API, Text-to-Speech/Speech-to-Text
  • Accessibility (2): WCAG, platform-specific accessibility

Each skill includes:

  • Core principles and best practices
  • Security considerations and CVE tracking
  • Code patterns and examples
  • Risk level classification (HIGH/MEDIUM/LOW)
  • Extended references for advanced patterns

🎯 Key Features

Security-First Development

  • OWASP Top 10 awareness built into all templates
  • CVE tracking in skill files
  • Credential management via OS keychain
  • Input validation emphasis
  • Security audit workflows

AI-Assisted Development Guidelines

  • TDD-first principles
  • Skill loading protocols
  • Security review requirements
  • AI code verification procedures
  • Conventional commit format enforcement

Template Variable System

All templates use placeholders that get replaced during setup:

  • {{PROJECT_NAME}} - Your project name
  • {{PROJECT_DESCRIPTION}} - Project description
  • {{ARCHITECTURE_DESCRIPTION}} - Architecture overview
  • {{TEST_COMMAND}} - Test command
  • {{BUILD_COMMAND}} - Build command
  • {{DEV_COMMAND}} - Development command
  • And many more...

📖 Documentation

🛠️ Supported Project Types

1. Web Application

  • Frontend SPAs (React, Vue, Svelte)
  • SSR frameworks (Next.js, Nuxt)
  • Relevant skills: vue-nuxt, tailwindcss, pinia, ui-ux-design

2. Backend/API

  • REST APIs (FastAPI, Express, NestJS)
  • GraphQL APIs
  • WebSocket servers
  • Relevant skills: fastapi, rest-api-design, graphql, websocket

3. Full-Stack

  • Monorepo setups
  • Combined frontend + backend
  • Relevant skills: Combination of web + backend skills

4. Desktop Application

  • Cross-platform apps (Tauri, Electron)
  • Native integrations
  • Relevant skills: tauri, browser-automation, auto-update-systems

5. CLI Tool

  • Command-line applications
  • Terminal utilities
  • Relevant skills: python, rust, typescript

6. Library/Package

  • Reusable libraries
  • npm/PyPI/crates.io packages
  • Relevant skills: Language-specific skills

🔧 How It Works

1. Setup Process

./setup-repo-multitype.sh my-new-project

The script will:

  1. Prompt for project details:

    • Project name and description
    • Project type(s) (multi-select supported)
    • Programming language(s)
    • Features and frameworks
  2. Copy relevant skills:

    • Automatically selects skills based on your choices
    • Copies selected skills to your project's skills/ directory
  3. Generate configuration files:

    • Replaces template variables with your values
    • Creates .claude/ configuration
    • Sets up .github/ workflows
  4. Initialize Git repository:

    • Creates initial commit
    • Sets up branch structure (optional)

2. Template Variable Replacement

During setup, all template files are processed:

Before:

# {{PROJECT_NAME}}

{{PROJECT_DESCRIPTION}}

After:

# My Awesome Project

A full-stack application for managing tasks with AI assistance.

3. Skill Loading

Skills are loaded based on your project configuration:

  • Automatic: Selected during setup based on project type
  • Manual: Copy additional skills from the skills/ directory
  • Claude Code: Reference skills in your CLAUDE.md or slash commands

🎓 Usage Examples

Example 1: Full-Stack Web Application

./setup-repo-multitype.sh ~/projects/task-manager

# Interactive prompts:
# - Project name: task-manager
# - Type: 3 (Full-Stack)
# - Language: 1 (TypeScript)
# - Framework: Nuxt 3 + FastAPI
# - Features: Database (SQLite), Authentication, API

Result:

  • Vue/Nuxt frontend skills
  • FastAPI backend skills
  • Database design and SQLite skills
  • TypeScript configuration
  • Full CI/CD pipeline
  • Claude Code integration

Example 2: Desktop Application

./setup-repo-multitype.sh ~/projects/my-desktop-app

# Interactive prompts:
# - Project name: my-desktop-app
# - Type: 4 (Desktop)
# - Framework: Tauri
# - Features: Auto-updates, System integration

Result:

  • Tauri skill with security patterns
  • Auto-update systems skill
  • Platform-specific skills (macOS, Windows, Linux)
  • Rust + TypeScript configuration
  • Cross-platform build workflows

Example 3: Python CLI Tool

./setup-repo-multitype.sh ~/projects/cli-tool

# Interactive prompts:
# - Project name: cli-tool
# - Type: 5 (CLI)
# - Language: 3 (Python only)
# - Features: Configuration, Logging

Result:

  • Python skill with best practices
  • CLI patterns and examples
  • Testing configuration (pytest)
  • PyPI packaging setup

🤖 Claude Code Integration

After project creation, Claude Code automatically:

  1. Reads CLAUDE.md at session start
  2. Applies project-specific rules and guidelines
  3. Uses skills when invoked via commands or instructions
  4. Enforces security and code quality standards
  5. Runs hooks for formatting and validation

Working with Claude Code

# In your new project
cd my-new-project

# Claude Code automatically reads CLAUDE.md
# Use slash commands:
/implement-feature     # Load feature implementation workflow
/run-tests            # Run all tests automatically
/review-code          # Run code review checklist

# Skills are referenced in CLAUDE.md and loaded as needed

📦 Project Structure After Setup

my-new-project/
├── CLAUDE.md                          # Claude's project instructions
├── README.md                          # Your project documentation
├── CONTRIBUTING.md                    # Contribution guidelines
├── SECURITY.md                        # Security policy
├── CODE_OF_CONDUCT.md                 # Code of conduct
│
├── .claude/
│   ├── settings.json                  # Claude Code configuration
│   └── commands/                      # Slash commands
│       ├── implement-feature.md
│       ├── run-tests.md
│       └── review-code.md
│
├── .github/
│   ├── workflows/
│   │   └── ci.yml                     # CI/CD pipeline
│   ├── PULL_REQUEST_TEMPLATE.md
│   └── ISSUE_TEMPLATE/
│       ├── bug_report.md
│       └── feature_request.md
│
├── skills/                            # Relevant domain skills
│   ├── typescript/
│   ├── fastapi/
│   ├── database-design/
│   └── ...
│
├── src/                               # Your source code
└── tests/                             # Your tests

🔒 Security Considerations

Built-in Security Features

  • OWASP Top 10 awareness in all templates
  • CVE tracking in security-sensitive skills
  • Credential management via OS keychain (never plaintext)
  • Input validation emphasis throughout
  • Security audit workflow in CI/CD
  • Sandboxing guidelines for file operations
  • Security policy template for responsible disclosure

Security Skills

The repository includes specialized security skills:

  • appsec-expert - Application security patterns
  • security-auditing - Security review checklists
  • encryption - Cryptography best practices
  • os-keychain - Secure credential storage
  • sandboxing - File system isolation
  • devsecops-expert - Security automation

🤝 Contributing

This is a template repository designed to bootstrap new projects. Contributions are welcome!

How to Contribute

  1. Add new skills:

    • Create skill directory: skills/your-skill/
    • Add SKILL.md with patterns and security considerations
    • Include risk level classification
    • Add references if needed
  2. Improve templates:

    • Enhance existing template files
    • Add new template variables
    • Improve Claude Code integration
  3. Enhance setup scripts:

    • Add new project types
    • Improve interactive prompts
    • Add validation and error handling
  4. Update documentation:

    • Improve usage examples
    • Add troubleshooting guides
    • Document new features

Contribution Guidelines

  • Follow security-first principles
  • Include examples and documentation
  • Test setup scripts thoroughly
  • Maintain template variable consistency
  • Follow conventional commit format

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

  • Built for use with Claude Code by Anthropic
  • Inspired by best practices from the open-source community
  • Security patterns from OWASP and CVE databases
  • Framework-specific patterns from official documentation

📞 Support

  • Documentation: See USAGE.md for detailed usage instructions
  • Skills Reference: See SKILLS_REFERENCE.md for all available skills
  • Issues: Report issues via GitHub Issues
  • Questions: Open a discussion in GitHub Discussions

Built with ❤️ for AI-assisted development