composable-svelte-deployment_skill

Overview

This skill documents production deployment patterns for Composable Svelte SSR applications, focused on Fly.io and Docker-based workflows. It provides concrete Docker multi-stage builds, Fly.io configuration, security hardening, performance tuning, and guidance for integrating Composable Rust backends. The content is actionable and aimed at keeping images small, secure, and production-ready.

How this skill works

It prescribes a 3-stage Docker build (deps → builder → runner) to produce minimal, non-root runtime images and reliable ES module resolution. It shows Vite SSR configuration for separate client and server builds, Fly.io Apps V2 configuration including internal .internal networking to connect to a Rust backend, and runtime safeguards like health checks, secrets validation, and HTTP security headers. It also covers caching, compression, rate limiting, and monitoring for scalable SSR.

When to use it

• Deploying a Composable Svelte SSR app to Fly.io or any Docker-friendly cloud platform
• You need <150MB production images with no devDependencies
• When integrating a Rust backend over Fly.io internal networking (.internal)
• When you require secure defaults: non-root runtime, CSP, HSTS, and rate limits
• When you need clear SSR caching and compression strategies for performance

Best practices

• Use a 3-stage Dockerfile: deps (prod) → builder → runner (minimal, non-root user)
• Run health checks and validate required secrets at startup; set secrets via Fly CLI, not fly.toml
• Configure Vite for dual builds (client + server) and don't externalize workspace packages needed at runtime
• Enable compression (Brotli), response caching (in-memory for single instance, Redis for multi-instance), and code-splitting for heavy components
• Harden HTTP with strict CSP, HSTS, X-Frame-Options, and per-route rate limiting

Example use cases

• Build and deploy a single-repository SSR app to Fly.io with private Rust backend via my-backend.internal:8080
• Package a monorepo: install workspace deps, build core and example packages, then copy only dist and production deps into runner image
• Local workflow: build Docker image, run with COMPOSABLE_RUST_BACKEND_URL pointing to localhost for integration testing, validate /health
• Scale on Fly.io: start with min memory 512mb, then scale horizontally or vertically and enable auto_scaling for traffic-driven scaling