genai-scaffold

genai-scaffold is an interactive Python CLI tool that bootstraps production-ready Generative AI project structures with customizable tech stacks, best practices, and modular organization.

---

✨ Features

• 🎯 Interactive CLI: Choose your tech stack interactively with a beautiful terminal UI
• 🧠 Multiple LLM Providers: OpenAI, Anthropic (Claude), Azure OpenAI, or Ollama (local)
• 🔧 Orchestration Frameworks: LangChain, LlamaIndex, DSPy, or raw Python
• 💾 Vector Databases: Pinecone, ChromaDB, Qdrant, or PostgreSQL with pgvector
• 🎨 UI Frameworks: Streamlit, Gradio, FastAPI, or headless
• 📦 Dependency Management: Poetry or pip (requirements.txt)
• 🐳 Docker Support: Automatic docker-compose configuration for local services
• 📊 Observability: Optional LangSmith or Weights & Biases integration
• 🧪 Testing: Pre-configured pytest setup with example tests
• 📝 Prompt Management: YAML-based prompt templates with versioning
• 🚀 Production-Ready: Makefile, .env management, logging, and best practices

---

📦 Installation

You can install it via PyPI:

pip install genai-scaffold

Or using pipx:

pipx install genai-scaffold

---

🚀 Usage

Interactive Mode (Recommended)

Simply run the create command with the --interactive flag or without any arguments:

genai-scaffold create --interactive

Or just:

genai-scaffold create

This will launch an interactive wizard that guides you through selecting:

• Project name
• LLM provider (OpenAI, Anthropic, Azure, Ollama)
• Orchestration framework (LangChain, LlamaIndex, DSPy, None)
• Vector database (ChromaDB, Pinecone, Qdrant, pgvector)
• UI framework (Streamlit, Gradio, FastAPI, None)
• Dependency manager (pip or Poetry)
• Docker configuration
• Observability tools

Command-Line Mode

For automation or quick scaffolding, specify all options via flags:

genai-scaffold create my-rag-app \
  --provider openai \
  --orchestrator langchain \
  --vector-db chromadb \
  --ui streamlit \
  --deps pip

Example: Create a RAG App with LangChain and Streamlit

genai-scaffold create my-chatbot \
  --provider anthropic \
  --orchestrator langchain \
  --vector-db pinecone \
  --ui streamlit

Example: Create a DSPy App with Local Models

genai-scaffold create local-ai-app \
  --provider ollama \
  --orchestrator dspy \
  --vector-db chromadb \
  --ui gradio \
  --no-docker

Generated Project Structure

my-rag-app/
├── src/
│   ├── llm/              # LLM client implementation
│   │   ├── __init__.py
│   │   └── client.py
│   ├── prompts/          # Prompt templates and management
│   │   ├── __init__.py
│   │   ├── loader.py
│   │   └── templates.yaml
│   ├── utils/            # Utility functions (logging, etc.)
│   ├── config.py         # Configuration management
│   ├── vector_store.py   # Vector database interface
│   └── rag_pipeline.py   # RAG implementation (if orchestrator selected)
├── tests/                # Pytest test suite
│   ├── conftest.py
│   └── test_example.py
├── data/                 # Data directories
│   ├── cache/
│   ├── outputs/
│   └── embeddings/
├── notebooks/            # Jupyter notebooks (optional)
├── app.py                # UI application (Streamlit/Gradio/FastAPI)
├── docker-compose.yml    # Docker services (if enabled)
├── Dockerfile            # Application container
├── .env.example          # Environment variables template
├── Makefile              # Common tasks (setup, test, run, etc.)
├── requirements.txt      # Python dependencies (or pyproject.toml)
├── pytest.ini            # Pytest configuration
└── README.md             # Project documentation

---

🎯 Quick Start with Generated Project

After scaffolding your project:

# 1. Navigate to your project
cd my-rag-app

# 2. Set up environment variables
cp .env.example .env
# Edit .env with your API keys

# 3. Install dependencies
make setup
# or: pip install -r requirements.txt

# 4. Start services with Docker (if enabled)
docker-compose up -d

# 5. Run the application
make run

# 6. Run tests
make test

---

🛠️ Available Commands

Create Command

genai-scaffold create [PROJECT_NAME] [OPTIONS]

Options:

• --provider: LLM provider (openai, anthropic, azure, ollama)
• --orchestrator: Framework (langchain, llamaindex, dspy, none)
• --vector-db: Vector database (pinecone, chromadb, qdrant, pgvector)
• --ui: UI framework (streamlit, gradio, fastapi, none)
• --deps: Dependency manager (pip, poetry)
• --docker/--no-docker: Enable/disable Docker configuration
• --interactive, -i: Use interactive mode

Version Command

genai-scaffold version

---

🔧 Tech Stack Options

LLM Providers

• OpenAI: GPT-4, GPT-3.5, and embedding models
• Anthropic: Claude 3 (Opus, Sonnet, Haiku)
• Azure OpenAI: Enterprise-grade OpenAI models
• Ollama: Local models (Llama 2, Mistral, etc.)

Orchestration Frameworks

• LangChain: Full-featured LLM framework with chains and agents
• LlamaIndex: Data framework for LLM applications
• DSPy: Declarative language model programming
• None: Raw Python with custom implementation

Vector Databases

• ChromaDB: Easy-to-use, local-first vector store
• Pinecone: Managed vector database service
• Qdrant: High-performance vector search engine
• pgvector: PostgreSQL extension for vector operations

UI Frameworks

• Streamlit: Fast way to build data apps
• Gradio: Quick ML model interfaces
• FastAPI: Modern, fast API framework
• None: Headless/CLI application

---

📚 Generated Features

Prompt Management

The generated projects include a sophisticated prompt management system:

from src.prompts import load_prompt

# Load and format a prompt template
prompt = load_prompt("rag_query", context="...", question="...")

Prompts are stored in src/prompts/templates.yaml with versioning support.

Configuration Management

Environment-based configuration with validation:

from src.config import Config

# Access configuration
api_key = Config.OPENAI_API_KEY
model = Config.OPENAI_MODEL

Logging

Pre-configured logging utilities:

from src.utils import get_logger

logger = get_logger(__name__)
logger.info("Processing request...")

Observability (Optional)

If enabled, automatic tracing with LangSmith or W&B:

from src.observability import trace_llm_call

@trace_llm_call
def my_llm_function():
    # Automatically traced
    pass

---

🧪 Testing

Generated projects include a complete test setup:

# Run tests
make test

# Run with coverage
make test-coverage

# Run specific test
pytest tests/test_example.py -v

---

🐳 Docker Support

When Docker is enabled, projects include:

• docker-compose.yml with service definitions
• Dockerfile for the application
• Automatic configuration for:
  • ChromaDB server (if selected)
  • Qdrant server (if selected)
  • PostgreSQL with pgvector (if selected)

Start all services:

docker-compose up -d

---

🔄 Makefile Commands

Generated projects include a Makefile with common tasks:

make setup          # Install dependencies
make test           # Run tests
make test-coverage  # Run tests with coverage
make run            # Run the application
make format         # Format code (black, isort)
make lint           # Run linter (ruff)
make clean          # Clean build artifacts
make docker-up      # Start Docker services (if Docker enabled)
make docker-down    # Stop Docker services

---

🤝 Contributing

Pull requests are welcome! For major changes, open an issue first to discuss what you'd like to change.

---

📄 License

This project is licensed under the MIT License.

---

🙌 Acknowledgements

Built for developers who want to quickly scaffold production-ready GenAI applications with best practices and flexibility.