Intelligent code generation platform that transforms natural language ideas into production-ready applications through AI-powered development teams.
β’ AI-powered multi-agent code generation system
β’ Complete full-stack application development
β’ Natural language to production-ready code transformation
AI AGENTS β’ DEVELOPMENT β’ PYTHON β’ REACT β’ DJANGO
CodeCraft AI is an intelligent code generation platform that takes your project ideas in plain English and crafts complete applications with:
- Python backend modules with complete business logic
- Frontend HTML/CSS/JS interfaces
graph TB
A[React Frontend<br/>Port 3000] -->|HTTP Requests| B[Django Ninja API<br/>Port 8000]
B -->|Kickoff Crew| C[CrewAI Orchestrator]
C --> D[Engineering Lead Agent<br/>ποΈ System Design]
D --> E[Backend Engineer Agent<br/>π Python Code]
E --> F[Frontend Engineer Agent<br/>π HTML/CSS/JS]
D -->|Creates| G[Design Document<br/>π .md file]
E -->|Creates| H[Python Module<br/>π .py file]
F -->|Creates| I[Web Interface<br/>π .html file]
G --> J[Output Directory<br/>π Generated Files]
H --> J
I --> J
J -->|File Contents| B
B -->|JSON Response| A
subgraph "Docker Container"
C
D
E
F
end
subgraph "Generated Output"
G
H
I
end
The system uses three specialized AI agents working in sequence:
- Role: System architect and design lead
- Task: Analyzes requirements and creates detailed technical design
- Output: Design document with architecture and specifications
- Role: Python developer
- Task: Implements the design as a complete Python module
- Output: Production-ready Python code with classes and methods
- Role: Web developer
- Task: Creates responsive HTML/CSS/JS interface
- Output: Self-contained HTML file with modern styling
Backend:
- Django 5.1+ with Django Ninja API
- CrewAI for AI agent orchestration
- Google Gemini 1.5 Flash LLM
- SQLite database
Frontend:
- React 18 with modern hooks
- Tailwind CSS for styling
- Axios for API communication
- Lucide React icons
- β¨ Natural Language Input: Describe your project in plain English
- π Real-time Generation: Watch as AI agents create your code
- π± Responsive UI: Works on desktop and mobile devices
- π₯ Download Code: Get generated files instantly
- π― Production Ready: Generated code follows best practices
- π Fast API: High-performance Django Ninja endpoints
- Python 3.10+
- Node.js 16+
- Docker (Required for CrewAI agents)
- Google Gemini API key
β οΈ Docker is mandatory - CrewAI agents run in Docker containers for safe code execution
# Make sure Docker is running
docker --version
# Start Docker Desktop or Docker daemoncd app
pip install -r requirements.txt
python manage.py migrate
python manage.py runservercd frontend
npm install
npm startCreate app/.env:
MODEL=gemini/gemini-1.5-flash
GEMINI_API_KEY=your_gemini_api_key_here- Start Backend: Django API runs on http://localhost:8000
- Start Frontend: React app runs on http://localhost:3000
- Open http://localhost:3000 in your browser
- Enter your project requirements
- Specify module and class names
- Click "Generate Code"
- Download your generated files
Backend (Django API):
cd app
python manage.py runserver
# API available at: http://localhost:8000/api/
# API docs at: http://localhost:8000/api/docsFrontend (React App):
cd frontend
npm start
# Web app available at: http://localhost:3000Note: Both backend and frontend must be running simultaneously for the application to work properly.
Input:
Build me a todo web app in cyberpunk style
Generated Code:
index.html (Cyberpunk styled)
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="UTF-8">
<title>Cyberpunk To-Do</title>
<style>
body {
background-color: #111;
color: #0f0;
font-family: 'Courier New', monospace;
}
.container {
max-width: 600px;
margin: 20px auto;
padding: 20px;
border: 1px solid #0f0;
box-shadow: 0 0 10px rgba(0, 255, 255, 0.5);
}
h1 {
color: #0ff;
text-shadow: 2px 2px 4px #000;
}
/* More cyberpunk styling... */
</style>
</head>
<body>
<div class="container">
<h1>Cyberpunk To-Do</h1>
<input type="text" id="newTask" placeholder="Enter new task">
<button id="addTaskBtn">Add Task</button>
<ul id="taskList"></ul>
</div>
<script>
// Interactive JavaScript for task management
</script>
</body>
</html>
Output Files:
todo_app.py- Complete Python module with TodoApp classindex.html- Cyberpunk-styled responsive web interfacetodo_app_design.md- Technical documentation and specifications
POST /api/run- Generate code from requirementsGET /api/health- Health check endpointGET /api/docs- Interactive API documentation
MIT License - see LICENSE file for details
- CrewAI for the multi-agent framework
- Django Ninja for the fast API
- Google Gemini for the language model
CodeCraft AI - Where ideas become code β¨