Cloud assembly

StoryShelf is cross-runtime: the core router (createShelfApp) uses only Web-standard APIs (Request, Response, fetch, ReadableStream, crypto, URL). There is no Node coupling in the core, so you can deploy on any platform that runs JavaScript.

Mix-and-match layers

Platform	Database	Storage	Testing	Auth	Server entry
Vercel	@storyshelf/db-sqlite/turso (Turso/libSQL)	@storyshelf/storage-s3 (R2/S3)	Hosting only (no testing)	@storyshelf/auth	Hono + @hono/vercel-edge
Cloudflare Workers	@storyshelf/db-sqlite/turso	@storyshelf/storage-s3 (R2)	Hosting only (no testing)	@storyshelf/auth	Hono + Workers entry
Azure Functions	@storyshelf/db-sqlite/turso	@storyshelf/storage-s3 (Azure Blob)	Hosting only (no testing)	@storyshelf/auth	Hono + Azure Functions handler
AWS Lambda	@storyshelf/db-sqlite/turso	@storyshelf/storage-s3 (S3)	Hosting only (no testing)	@storyshelf/auth	Hono + Lambda handler
Deno Deploy	@storyshelf/db-sqlite/turso	@storyshelf/storage-s3 (R2/S3)	Hosting only (no testing)	@storyshelf/auth	Hono + Deno entry
Bun	@storyshelf/db-sqlite or @storyshelf/db-sqlite/bun-sqlite	storage-local or storage-s3	testing (long-running host)	@storyshelf/auth	Hono + Bun.serve

Serverless: Storybook hosting only

Enabling testing requires a long-running server (a container, VM, Fly Machine or Kubernetes pod). It cannot run on serverless platforms that freeze or end the process after each request (Vercel Functions, AWS Lambda, Cloudflare Workers), because capture jobs run for minutes in the background and the job consumer must keep running. A serverless deployment can host Storybooks only, by leaving testing unset.

testing needs a process that keeps running between requests: capture jobs run for minutes in the background. Vercel Functions, AWS Lambda, Netlify Functions and Cloudflare Workers freeze or end the process after each request, so on those platforms StoryShelf hosts Storybooks only: leave the testing option out. See Hosting only.

• AWS Lambda. createShelfApp throws at boot when AWS_LAMBDA_FUNCTION_NAME is set and testing.runner is configured.
• Vercel and Netlify. Boot logs a warning when testing.runner is configured (VERCEL or NETLIFY is set).
• Override. STORYSHELF_ALLOW_SERVERLESS=1 silences both, for a host that looks serverless but keeps a process alive.
• Cloudflare D1. The @storyshelf/db-sqlite/d1 preset is a Workers-only binding, so it cannot back an instance with testing.runner; createShelfApp throws at boot. Turso over the network can.
• Uploads. With no consumer on the host, a zip over maxInlineUnzipSize is refused with 413, and storyshelf upload falls back to the file-by-file route by itself.

Minimal Turso + S3 recipe (long-running host with testing)

import { createShelfApp } from "@storyshelf/app";
import { createTursoDatabase } from "@storyshelf/db-sqlite/turso";
import { createS3Storage } from "@storyshelf/storage-s3";
import { createPlaywrightCaptureRunner } from "@storyshelf/runner-playwright";

const database = createTursoDatabase({
  url: process.env.TURSO_URL!,
  authToken: process.env.TURSO_AUTH_TOKEN!,
});
const storage = createS3Storage({
  bucket: process.env.S3_BUCKET!,
  region: process.env.S3_REGION,
  endpoint: process.env.S3_ENDPOINT,
  accessKeyId: process.env.S3_ACCESS_KEY_ID!,
  secretAccessKey: process.env.S3_SECRET_ACCESS_KEY!,
});
const capture = createPlaywrightCaptureRunner();

export const app = createShelfApp({
  database, storage, testing: { runner: capture, concurrency: 2 },
  config: { scratchDir: "/tmp/scratch" },
  auth: ...,
});
await app.lifecycle.setup();

Without the testing option the same assembly runs on any platform that supports Node-compatible fetch + crypto + sqlite/libsql (Storybook hosting only). With testing it needs a long-running host.

Postgres providers

@storyshelf/db-postgres uses postgres.js + Drizzle. Use createPostgresDatabase({ url }) for all providers — only the connection string and pooling change.

Provider	Connection shape	TLS	Pooling
Self-hosted	postgres://user:pass@localhost:5432/shelf	ssl: false	max: 10
AWS RDS	postgres://user:pass@<rds-endpoint>:5432/db?sslmode=require	ssl: true	prepare: true
GCP Cloud SQL	postgres://user:pass@<cloudsql-ip>:5432/db	ssl: { ca, cert, key }	Connector or injected client
Supabase	postgres://postgres.<ref>:<pass>@<pooler-host>:6543/postgres?pgbouncer=true	ssl: true	prepare: false on :6543
Neon	postgres://user:pass@<neon-host>/db?sslmode=require	ssl: true	max: 5; prepare: false on pooled
Azure DB for PG	postgres://user:pass@<server>.postgres.database.azure.com:5432/db?sslmode=require	ssl: true	Entra ID via injected client

See the @storyshelf/db-postgres package reference for full options.

Related

• AWS · Azure · GCP — reference Terraform stacks
• Docker Compose — single-host deploy