AI Support System
Deloney-code/ai-support-system
Sistema de tickets com agente IA, dashboard WebSocket e automação segura
Ver no GitHubSistema de suporte ao cliente alimentado por IA que gerencia tickets de forma inteligente usando Claude. Automatiza classificação, resumo de conversas, polimento de respostas e resolução automática de problemas simples — tudo integrado com email, dashboard em tempo real e controle de acesso baseado em papéis.
Instalação
npx convez-ai Deloney-code/ai-support-system1 comando no terminal: detecta o formato do repositório (marketplace ou plugin avulso) e instala via Claude Code. Requer Node 18+ e o Claude Code instalado.
Métricas
Casos de uso
- Automatizar resposta a emails de suporte com Claude
- Classificar e priorizar tickets automaticamente
- Resumir conversas longas de clientes em um clique
- Escalar problemas complexos para agentes humanos
- Dashboard em tempo real com WebSockets para equipe
Veredito

Solução madura e pronta para produção que combina Django, Claude API e infraestrutura moderna (PostgreSQL, Redis, Celery). Excelente referência para quem quer implementar automação de suporte ao cliente com IA — 49 testes e segurança auditada demonstram qualidade industrial.
README
A tradução para o português deste README ainda está na fila.
AI Support System
A production-grade AI-powered customer support ticket system built with Django. Deployed live on Railway with PostgreSQL, Redis, Celery, and WebSockets.
Features
Ticket Management
- Role-based access control (Customer, Agent, Admin)
- Full ticket lifecycle — open, in progress, resolved, closed
- Priority and category classification
- Real-time dashboard updates via WebSockets
- Comment threads on tickets
AI Features (Anthropic Claude API)
- Summarize — condense long ticket conversations in one click
- Polish Reply — transform rough agent drafts into professional responses
- Auto-classify — automatically assign category and priority using AI
- Auto-resolve — AI resolves simple tickets automatically
Email AI Agent
- Customers email support@mg.simedelonney.com
- AI agent picks up the email automatically
- Generates and sends a professional reply within seconds
- Escalates complex issues to human agents
- No human involvement for simple requests
Security
- IP-based rate limiting on authentication endpoints
- CSRF protection on all forms
- XSS prevention with bleach sanitization
- Role-based access control
- Mailgun webhook signature verification
- Environment variables for all secrets
- 49 automated security tests — all passing
Tech Stack
| Layer | Technology |
|---|---|
| Backend | Django 5.x |
| Database | PostgreSQL |
| Cache/Queue | Redis |
| Background Tasks | Celery |
| Real-time | Django Channels + WebSockets |
| AI | Anthropic Claude API |
| Mailgun | |
| Frontend | Bootstrap 5 |
| Containerization | Docker |
| Deployment | Railway |
| Testing | Django TestCase (49 tests) |
Architecture
Browser → Daphne (ASGI) → Django → Django Channels → Redis (WebSockets) → Celery Worker → Redis (Task Queue) → PostgreSQL (Data) → Anthropic API (AI) → Mailgun (Email)
Live Demo
https://web-production-4e3df.up.railway.app
Local Development
Prerequisites
- Python 3.11+
- Redis running locally
- PostgreSQL database
Setup
# Clone the repository
git clone https://github.com/Deloney-code/ai-support-system.git
cd ai-support-system
# Create virtual environment
python -m venv venv
source venv/bin/activate # Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# Create .env file
cp .env.example .env
# Edit .env with your credentials
# Run migrations
python manage.py migrate
# Create superuser
python manage.py createsuperuser
# Start Redis (WSL on Windows)
redis-server
# Start Celery worker
celery -A core worker --loglevel=info --pool=solo
# Start development server
daphne -p 8000 core.asgi:application
Environment Variables
SECRET_KEY=your-secret-key DEBUG=False ALLOWED_HOSTS=your-domain.com DATABASE_URL=postgresql://... REDIS_URL=redis://... ANTHROPIC_API_KEY=your-anthropic-key MAILGUN_API_KEY=your-mailgun-key MAILGUN_SMTP_KEY=your-smtp-key MAILGUN_SMTP_LOGIN=postmaster@mg.yourdomain.com DEFAULT_FROM_EMAIL=support@mg.yourdomain.com EMAIL_BACKEND=django.core.mail.backends.smtp.EmailBackend
Running Tests
python manage.py test tickets accounts --verbosity=2
49 tests covering:
- Model validation and XSS sanitization
- View access control and permissions
- AI endpoint authorization
- Mailgun webhook security
- Authentication flows
Project Structure
ai_support_system/ ├── accounts/ # Custom user model, auth views ├── tickets/ # Tickets, comments, AI features │ ├── models.py # Ticket, TicketComment, InboundEmail │ ├── views.py # All views including webhook │ ├── ai_service.py # Anthropic API integration │ ├── tasks.py # Celery background tasks │ ├── consumers.py # WebSocket consumers │ └── tests.py # 49 automated tests ├── core/ # Settings, URLs, Celery config ├── templates/ # Bootstrap 5 templates └── start.py # Production startup script
Author
Built by Sime Delonney as part of a Professional Masters in Cybersecurity project.
⭐ Star this repo if you find it useful!