Proxy Deployment
Deploy the AIProxyGuard proxy for self-hosted prompt injection detection.
Don’t want to manage infrastructure? Use the Cloud API directly with our SDKs - no proxy deployment required.
Deployment Options
| Platform | Best For | Guide |
|---|---|---|
| Docker | Local development, simple deployments | Quick start |
| DigitalOcean | Small-medium production | App Platform |
| AWS Lightsail | AWS ecosystem | Container Service |
| Azure Container Apps | Azure ecosystem | Serverless containers |
| Google Cloud Run | GCP ecosystem | Serverless |
| Railway | Startups, quick deploys | One-click deploy |
Health Endpoints
All deployments should configure health checks:
| Endpoint | Purpose | Response |
|---|---|---|
GET /healthz |
Liveness | {"status": "healthy"} |
GET /readyz |
Readiness | {"status": "ready", ...} |
GET /metrics |
Prometheus | Prometheus text format |
Resource Sizing
| Traffic | Memory | Notes |
|---|---|---|
| < 100 req/min | 512 MB | Development |
| 100-1000 req/min | 1 GB | Small production |
| 1000-10000 req/min | 2 GB | Medium production |
| > 10000 req/min | 2+ GB × N instances | Scale horizontally |