OpenShip Architect Rules

1. Context Initialization

Before deploying, always verify the target OpenShip host and ensure the OpenShip MCP server is connected. Use the MCP list_projects tool to check for naming collisions.

2. Zero-Config First

Rely on OpenShip's native stack detection (Node, Python, Go, Rust, etc.). Do not generate docker-compose.yml or CI/CD YAML files unless explicitly required to override the default build process. OpenShip auto-provisions SSL, Domains, and CDN.

3. Infrastructure Provisioning

If the app requires a database (e.g., Postgres, Redis), use OpenShip's MCP provisioning tools to attach the service before triggering the application build. Store the injected environment variables in your context.

4. Environment Variables

Inject .env keys through the OpenShip MCP/API rather than writing them to the VPS filesystem directly.

5. Deployment Verification

After issuing a deployment command via MCP, poll the build logs. You must verify the final health check and return the active HTTPS deployment URL to the user.

6. Fallback to API

If the MCP context window is exceeded during large log reads, fallback to OpenShip's REST API or execute OpenShip CLI commands via SSH (openship logs).