Prompt Generator

Enterprise-grade prompt generator for ChatGPT and Claude Code with web interface and CLI.

Generate structured, high-quality prompts with context, constraints, and deliverables for AI-powered development workflows.

Features

Core Functionality

• 🎯 Two Target AI Models: ChatGPT and Claude Code optimized prompts
• 🌐 Web Interface: Beautiful, responsive UI with dark/light theme
• 💻 CLI Tool: Command-line interface with interactive console mode
• 📝 Structured Prompts: Organize prompts with task, context, constraints, deliverables, and tone

NEW: Advanced Features 🚀

• 🎮 Prompt Playground: Test prompts with real AI models (Anthropic, OpenAI, OpenRouter)
• 📚 Prompt Library: Save, organize, and reuse your best prompts
• 🔑 Bring Your Own Key (BYOK): Securely store and manage API keys
• 🔌 Detached Mode: Generate prompts without API integration
• 🔐 Encrypted Storage: API keys encrypted at rest with Fernet
• 🏷️ Tag System: Organize prompts with custom tags
• ⭐ Favorites: Mark and filter your most-used prompts
• 📊 Usage Metrics: Track tokens used and response times

Enterprise Features

• 🔒 Production-Ready Security: HTTPS, CSP, CORS, rate limiting, input validation
• 📊 Monitoring & Metrics: Prometheus metrics, structured logging, health checks
• 🚀 High Performance: Redis caching, Gunicorn workers, auto-scaling ready
• 🐳 Containerized: Docker and Docker Compose for easy deployment
• ☸️ Kubernetes Ready: Complete K8s manifests with HPA, ingress, and StatefulSets
• 🔄 CI/CD Pipeline: Automated testing, linting, security scanning, and deployment
• 📈 Scalable Architecture: Load balancing, connection pooling, horizontal scaling
• 🛡️ Security Hardened: Non-root containers, read-only filesystem, security scanning
• 📚 Comprehensive Tests: Unit, integration, and API tests with >80% coverage
• 📖 API Documentation: OpenAPI 3.0 specification
• 🔧 Operational Runbook: Complete deployment and troubleshooting guide

Quick Start

Web Interface (Recommended)

# Using Docker Compose (includes Redis, Nginx)
docker-compose up

# Access at http://localhost

CLI Usage

# Install
pip install -r requirements.txt

# Generate a prompt
python -m prompt_gen --target chatgpt --task "Summarize this article" --context "..."

# Interactive console mode
python -m prompt_gen --console

Installation

Prerequisites

• Python 3.10+
• Docker (for containerized deployment)
• Redis (optional, for caching)

Development Setup

# Clone the repository
git clone https://github.com/yourorg/prompt-generator.git
cd prompt-generator

# Create virtual environment
python3 -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Run development server
python app.py

Access the web interface at http://127.0.0.1:5000

Production Setup

# Set required environment variables
export SECRET_KEY=$(python -c "import secrets; print(secrets.token_urlsafe(32))")
export FLASK_ENV=production
export REDIS_URL=redis://localhost:6379/0

# Run with Gunicorn
gunicorn -c gunicorn.conf.py "app:create_app()"

See docs/RUNBOOK.md for complete production deployment guide.

Usage

Web Interface

1. Navigate to http://localhost (or your deployment URL)
2. Select your target AI model (ChatGPT or Claude Code)
3. Fill in the required fields:
   • Task: What you want the AI to do (required)
   • Context: Background information, codebase details
   • Constraints: Restrictions, requirements, tech stack
   • Deliverables: Expected outputs
   • Tone: Desired communication style
4. Click "Generate" to create your prompt
5. Copy the generated prompt to use with your AI model

CLI Examples

# Basic usage
prompt-gen --task "Add error handling" --target claude_code

# With context and constraints
prompt-gen \
  --task "Implement user authentication" \
  --context "Flask API with PostgreSQL" \
  --constraints "Use JWT tokens, OAuth2" \
  --deliverables "Auth endpoints, tests, migration" \
  --tone "concise"

# Output to file
prompt-gen --task "Refactor database layer" --out prompt.txt

# JSON output (for scripting)
prompt-gen --task "Fix bug in checkout" --json

# Interactive console
prompt-gen --console

Console Commands

In console mode (prompt-gen --console):

prompt-gen> help
Available commands:
  new              Reset all fields
  fields           List available fields
  set <field> <value>    Set a field value
  target [name]    Get/set target AI model
  show             Display current configuration
  generate         Generate the prompt
  copy             Copy generated prompt to clipboard
  save <path>      Save generated prompt to file
  export <path>    Export configuration to JSON
  load <path>      Load configuration from JSON
  exit/quit        Exit console

Prompt Playground 🎮

Test your prompts with real AI models directly in the browser:

1. Navigate to Playground tab

2. Select Provider:

   • Detached Mode: Generate prompts without API calls (default)
   • Anthropic Claude: Claude 3.5 Sonnet, Haiku, Opus
   • OpenAI: GPT-4o, GPT-4o-mini, GPT-4 Turbo
   • OpenRouter: Access to multiple models

3. Add API Key (for non-detached mode):

   • Go to "API Keys" tab and add your keys
   • Or enter key inline (not saved)

4. Test Your Prompt:

   • Enter prompt text
   • Click "Execute"
   • View response with token usage and timing

Example Use Cases:

• Test different prompt variations
• Compare model responses
• Verify prompt effectiveness before production use
• Debug prompt issues in real-time

Prompt Library 📚

Save and organize your prompts for reuse:

Saving Prompts:

1. Generate a prompt in the Generator tab
2. Click "Save to Library"
3. Prompt is automatically saved with metadata

Managing Library:

• Search: Find prompts by name, description, or content
• Filter: Show only favorites
• Tags: Organize with custom tags
• View Details: Click any prompt to see full content
• Copy: One-click copy to clipboard
• Delete: Remove prompts you no longer need

Example Workflow:

1. Create prompt: "Add authentication to Flask API"
2. Save to library with tags: ["authentication", "flask", "api"]
3. Mark as favorite for quick access
4. Later: Search "auth" → Find prompt → Copy → Use

API Key Management 🔑

Securely store API keys for use in Playground:

Adding Keys:

1. Go to "API Keys" tab
2. Select provider (Anthropic, OpenAI, OpenRouter)
3. Enter key name (e.g., "My Claude Key")
4. Paste API key
5. Click "Add Key"

Security:

• Keys encrypted at rest using Fernet (AES)
• Keys never logged or exposed in API responses
• Only you can access your keys
• Delete keys anytime

Using Stored Keys:

• In Playground, select from "Stored API Key" dropdown
• No need to re-enter keys each time
• Switch between multiple keys easily

Environment Variables:

# Set encryption key (important!)
export ENCRYPTION_KEY=$(python -c "import secrets; print(secrets.token_urlsafe(32))")

# Database location
export DATABASE_URL=sqlite:///./prompt_generator.db

API Documentation

RESTful API for integration with other tools.

Endpoints

POST /api/v1/generate

Generate a prompt from JSON input.

Request:

{
  "target": "chatgpt",
  "task": "Summarize this article",
  "context": "Technical blog post about distributed systems",
  "constraints": "Keep it under 3 paragraphs",
  "deliverables": "Key points and action items",
  "tone": "concise"
}

Response:

{
  "prompt": "You are an expert assistant...",
  "cached": false
}

GET/POST /api/v1/library/prompts

List or create saved prompts.

Query Parameters (GET):

• search: Search by name/description/task
• tag: Filter by tag
• favorites: Show only favorites (true/false)
• limit: Max results (default: 50)
• offset: Pagination offset

Request (POST):

{
  "name": "Auth Implementation",
  "task": "Add JWT authentication",
  "target": "claude_code",
  "tags": ["auth", "security"],
  "is_favorite": false
}

GET/PUT/DELETE /api/v1/library/prompts/:id

Get, update, or delete a specific prompt.

POST /api/v1/library/prompts/:id/favorite

Toggle favorite status of a prompt.

GET /api/v1/library/tags

Get all unique tags from saved prompts.

GET/POST /api/v1/keys

List or create API keys.

Request (POST):

{
  "provider": "anthropic",
  "key_name": "My Claude Key",
  "api_key": "sk-ant-..."
}

DELETE /api/v1/keys/:id

Delete an API key.

GET /api/v1/playground/providers

List available AI providers and their models.

POST /api/v1/playground/execute

Execute a prompt with an AI provider.

Request:

{
  "provider": "anthropic",
  "model": "claude-3-5-sonnet-20241022",
  "prompt": "Explain quantum computing",
  "key_id": 1  // or "api_key": "sk-..."
}

Response:

{
  "success": true,
  "response": "Quantum computing...",
  "model": "claude-3-5-sonnet-20241022",
  "tokens_used": 250,
  "duration_ms": 1523
}

GET /api/v1/playground/history

Get playground execution history.

GET /health

Health check endpoint for load balancers.

GET /ready

Readiness check with dependency validation.

GET /metrics

Prometheus metrics (restrict in production).

Full API specification: openapi.json

Architecture

┌─────────────────┐
│   Load Balancer │
│   (Nginx/ALB)   │
└────────┬────────┘
         │
    ┌────▼────┐
    │  Nginx  │ (Reverse Proxy, Rate Limiting)
    └────┬────┘
         │
    ┌────▼────────┐
    │   Gunicorn  │ (WSGI Server, 4 workers)
    │   + Flask   │
    └────┬────────┘
         │
    ┌────▼────┐
    │  Redis  │ (Caching, Rate Limiting)
    └─────────┘

Key Components

• Flask: Lightweight WSGI web framework
• Gunicorn: Production WSGI server with gthread workers
• Redis: In-memory cache and rate limit store
• Nginx: Reverse proxy, SSL termination, rate limiting
• Prometheus: Metrics collection and alerting
• Structlog: Structured JSON logging

Configuration

All configuration via environment variables:

Variable	Required	Default	Description
SECRET_KEY	Yes*	-	Flask secret key (required in production)
FLASK_ENV	No	production	Environment: production, development, testing
REDIS_URL	No	redis://localhost:6379/0	Redis connection URL
CACHE_ENABLED	No	true	Enable response caching
RATE_LIMIT_ENABLED	No	true	Enable rate limiting
RATE_LIMIT_PER_MINUTE	No	60	Requests per minute per IP
LOG_LEVEL	No	INFO	Log level: DEBUG, INFO, WARNING, ERROR
LOG_FORMAT	No	json	Log format: json or text

Generate SECRET_KEY:

python -c "import secrets; print(secrets.token_urlsafe(32))"

See config.py for full configuration reference.

Development

Running Tests

# Install dev dependencies
pip install -r requirements.txt

# Run all tests with coverage
pytest --cov=prompt_gen --cov=. --cov-report=html

# Run specific test file
pytest tests/unit/test_core.py

# Run with verbose output
pytest -v

# Open coverage report
open htmlcov/index.html

Code Quality

# Lint with ruff
ruff check .

# Format code
ruff format .

# Type check
mypy app.py prompt_gen/

# Security scan
safety check --file requirements.txt

Local Development with Docker

# Build and run all services
docker-compose up

# Rebuild after code changes
docker-compose up --build

# Run in background
docker-compose up -d

# View logs
docker-compose logs -f web

# Stop all services
docker-compose down

Deployment

Docker

# Build image
docker build -t prompt-generator:v1.0.0 .

# Run container
docker run -d \
  -p 5000:5000 \
  -e SECRET_KEY=your-secret-key \
  -e REDIS_URL=redis://redis:6379/0 \
  --name prompt-generator \
  prompt-generator:v1.0.0

Kubernetes

# Deploy to Kubernetes
kubectl apply -f k8s/deployment.yaml
kubectl apply -f k8s/redis.yaml

# Check rollout status
kubectl rollout status deployment/prompt-generator -n prompt-generator

# View logs
kubectl logs -f deployment/prompt-generator -n prompt-generator

# Scale deployment
kubectl scale deployment/prompt-generator --replicas=5 -n prompt-generator

Cloud Providers

See platform-specific guides:

• AWS ECS/EKS: docs/aws-deployment.md
• Google Cloud Run/GKE: docs/gcp-deployment.md
• Azure Container Apps/AKS: docs/azure-deployment.md

Monitoring

Health Checks

# Health endpoint (fast, no dependencies)
curl http://localhost:5000/health

# Readiness endpoint (checks dependencies)
curl http://localhost:5000/ready

Metrics

Prometheus metrics available at /metrics:

curl http://localhost:5000/metrics

Key metrics:

• flask_http_request_total - Total requests
• flask_http_request_duration_seconds - Latency
• flask_http_request_exceptions_total - Errors
• process_resident_memory_bytes - Memory usage

Logging

Structured JSON logs for easy parsing:

{
  "event": "request_completed",
  "method": "POST",
  "path": "/api/v1/generate",
  "status_code": 200,
  "request_id": "abc123",
  "timestamp": "2026-01-10T12:00:00Z"
}

Security

Security Features

• ✅ HTTPS enforced with HSTS
• ✅ Content Security Policy (CSP)
• ✅ CORS protection
• ✅ Rate limiting (60 req/min per IP)
• ✅ Input validation and sanitization
• ✅ Request size limits (1MB max)
• ✅ Security headers (X-Frame-Options, X-Content-Type-Options)
• ✅ Non-root container execution
• ✅ Read-only root filesystem
• ✅ Regular dependency vulnerability scans

Security Best Practices

1. Never run with FLASK_DEBUG=true in production
2. Generate a strong SECRET_KEY (32+ random bytes)
3. Keep dependencies updated with pip-audit or safety
4. Restrict /metrics endpoint to internal networks
5. Use HTTPS with valid TLS certificates
6. Monitor logs for suspicious activity
7. Scan containers with Trivy or similar tools

Troubleshooting

Common Issues

Problem: High latency (>200ms)

• Solution: Check Redis connection, enable caching, scale replicas

Problem: Rate limit errors (429)

• Solution: Adjust RATE_LIMIT_PER_MINUTE, implement authentication

Problem: Memory usage growing

• Solution: Check for leaks, adjust worker max_requests, restart pods

Problem: Cache misses

• Solution: Verify Redis connectivity, check REDIS_URL configuration

See docs/RUNBOOK.md for complete troubleshooting guide.

Performance

Benchmarks

Tested on AWS t3.medium (2 vCPU, 4GB RAM):

Metric	Value
p50 latency	35ms
p95 latency	78ms
p99 latency	145ms
Throughput	1,200 req/sec
Cache hit rate	85%
Memory per worker	80MB
CPU utilization	45% @ 1000 req/sec

Scaling Guidelines

• < 100 req/sec: 2-3 replicas
• 100-500 req/sec: 3-5 replicas
• 500-1000 req/sec: 5-8 replicas
• > 1000 req/sec: 8+ replicas with Redis cluster

Auto-scaling configured for CPU >70% and Memory >80%.

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

Development Workflow

1. Fork the repository
2. Create a feature branch: git checkout -b feature/amazing-feature
3. Make your changes
4. Run tests: pytest
5. Run linters: ruff check . && ruff format .
6. Commit: git commit -m "Add amazing feature"
7. Push: git push origin feature/amazing-feature
8. Open a Pull Request

License

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

Support

• 📖 Documentation: docs/
• 🐛 Bug Reports: GitHub Issues
• 💬 Discussions: GitHub Discussions
• 📧 Email: support@example.com

Changelog

v1.0.0 (2026-01-10)

Major Improvements:

• ✨ Complete rewrite with enterprise-grade architecture
• 🔒 Production-ready security (HTTPS, CSP, CORS, rate limiting)
• 📊 Monitoring and metrics (Prometheus, structured logging)
• 🚀 High performance (Redis caching, Gunicorn, auto-scaling)
• 🐳 Docker and Kubernetes deployment
• 🔄 CI/CD pipeline with automated testing and security scans
• 📚 Comprehensive tests (unit, integration, API)
• 📖 OpenAPI documentation
• 🔧 Operational runbook

Breaking Changes:

• API endpoints now versioned: /api/generate → /api/v1/generate
• Configuration now via environment variables (no more hardcoded values)
• Requires Redis for full feature set (optional, graceful degradation)

Migration Guide:

• Update API calls to use /api/v1/generate
• Set SECRET_KEY environment variable
• Configure REDIS_URL for caching and rate limiting

Acknowledgments

• Built with Flask
• Monitored with Prometheus
• Deployed on Kubernetes
• Inspired by best practices from 12-Factor App

---

Made with ❤️ for developers who love high-quality prompts