@getvitops/create 0.2.0 → 0.3.0

package/README.md

@@ -43,6 +43,12 @@ with every relevant `@getvitops/*` package pre-wired:
 - **GitHub bootstrap** — `pnpm run init:github` creates a same-named GitHub
   repo via the `gh` CLI, pushes `main` + `dev`, and makes `dev` the default
   branch (idempotent; `--dry-run` / `--public` flags).
+- **Hosting seam** — astro.config resolves its adapter + EmDash
+  database/storage through `vitopsHosting()` from `@getvitops/emdash`
+  (`cloudflare` default; `node` for VPS/docker/k8s after installing
+  `@astrojs/node` + `better-sqlite3`), and `pnpm run init:hosting` provisions
+  the platform (cloudflare: dev D1 + R2 created via wrangler, `DEV_D1_ID` /
+  `CLOUDFLARE_ACCOUNT_ID` repo secrets set via `gh`; idempotent, `--dry-run`).
 - **dev → main promotion flow** — GitHub Actions deploying `dev` to an
   isolated `-dev` worker (`dev.<domain>` with a custom domain; own D1/R2 via
   `scripts/build-dev-wrangler.mjs`, which hard-fails without `DEV_D1_ID`) and

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@getvitops/create",
-  "version": "0.2.0",
+  "version": "0.3.0",
   "description": "Project templates for the Vitops design system — scaffold wired-up sites with `vp create @getvitops`.",
   "license": "MIT",
   "files": [

package/templates/emdash/.claude/settings.json

@@ -1,4 +1,7 @@
 {
+  "permissions": {
+    "allow": ["mcp__astro-docs__search_astro_docs", "mcp__emdash-docs__search_docs"]
+  },
   "hooks": {
     "SessionStart": [
       {

package/templates/emdash/AGENTS.md

@@ -22,6 +22,11 @@ styled by the Vitops design system.
   full generated class-vocabulary reference to this file.
 - The Worker cron trigger in `wrangler.jsonc` drives scheduled publishing —
   don't remove it.
+- Hosting (adapter + EmDash database/storage) comes from `vitopsHosting()`
+  (`@getvitops/emdash`) in astro.config.mjs — never hard-code an adapter or
+  d1/r2/sqlite there. Switch targets via `vitopsHosting({ target })` or the
+  `HOSTING` env var (see README "Hosting"); the `node` target needs
+  `@astrojs/node` + `better-sqlite3` installed first.
 
 ## Branch model
 
@@ -49,4 +54,6 @@ searching them over pre-trained knowledge for Astro/EmDash questions.
 - `pnpm deploy:dev` — build + deploy the isolated dev worker (needs `DEV_D1_ID`)
 - `pnpm run init:github` — one-time: create + push the GitHub repo (main + dev,
   dev default). Only run when the user asks to set up GitHub.
+- `pnpm run init:hosting` — one-time: provision the hosting platform (dev
+  D1/R2 + repo secrets on cloudflare). Only run when the user asks for it.
 - `npx vitops validate design-system.json` — schema-check the design config

package/templates/emdash/README.md

@@ -74,14 +74,15 @@ One-time setup:
    pushes `main` + `dev`, and makes `dev` the default branch. Idempotent;
    `-- --dry-run` previews. Then branch-protect `main` (require the promotion
    PR + review).
-2. Create the isolated dev resources:
-   `wrangler d1 create my-emdash-site-dev` and
-   `wrangler r2 bucket create my-emdash-site-media-dev`.
-3. Add repo secrets: `gh secret set CLOUDFLARE_API_TOKEN`,
-   `gh secret set CLOUDFLARE_ACCOUNT_ID`, and `gh secret set DEV_D1_ID`
-   (the id printed by `d1 create`). Optionally `PROMOTE_TOKEN` (a PAT allowed
-   to bypass protection) for hands-off promotion. The first workflow runs
-   fail until these exist.
+2. `pnpm run init:hosting` — provisions the hosting platform (after
+   `npx wrangler login`): creates the isolated dev D1 + R2 and sets the
+   `DEV_D1_ID` / `CLOUDFLARE_ACCOUNT_ID` repo secrets. Idempotent;
+   `-- --dry-run` previews.
+3. Create a Cloudflare API token (Workers + D1 + R2 edit —
+   <https://dash.cloudflare.com/profile/api-tokens>; tokens can't be created
+   via CLI) and `gh secret set CLOUDFLARE_API_TOKEN`. Optionally
+   `PROMOTE_TOKEN` (a PAT allowed to bypass protection) for hands-off
+   promotion. The first workflow runs fail until the secrets exist.
 
 `pnpm deploy:dev` reproduces the dev deploy locally: it rewrites the
 adapter-generated wrangler config (`scripts/build-dev-wrangler.mjs`) so the
@@ -89,6 +90,30 @@ dev worker never touches prod data — it hard-fails without `DEV_D1_ID`,
 on purpose. Schema/content changes live in the EmDash database, not git, so
 promote them separately (via the prod admin or its MCP server).
 
+## Hosting
+
+The adapter and EmDash database/storage come from one call in
+`astro.config.mjs` — `vitopsHosting()` from `@getvitops/emdash` — so hosting
+is a configurable dimension of the project, not a rewrite:
+
+- **`cloudflare`** (default) — Workers + D1 + R2. `wrangler.jsonc`,
+  `src/worker.ts` (fetch + scheduled handlers), and the deploy workflows
+  belong to this target.
+- **`node`** — `@astrojs/node` (standalone) + SQLite file + local uploads,
+  for a VPS / docker-compose / k8s host. Not installed by default; to switch:
+  `pnpm add @astrojs/node better-sqlite3`, then `vitopsHosting({ target:
+'node' })` (or set `HOSTING=node` — the env var wins). Run the built site
+  with `node ./dist/server/entry.mjs`; data lives in `./data/` (gitignored) —
+  persist it as a volume. Scheduled publishing runs in-process on Node, so
+  the worker/cron pieces aren't needed. For production, move storage to
+  S3-compatible and/or the database to Postgres via `vitopsHosting({ node:
+{ database, storage } })`. CI deploy for Node hosts is TBD — the shipped
+  workflows are Cloudflare-specific.
+
+Moving an existing site between targets (either direction) is a data
+migration: content lives in the database (D1 export / EmDash seed round-trip)
+and media in the storage backend (bucket/directory copy).
+
 ## AI tooling
 
 - **MCP servers** (`.mcp.json`): the Astro docs (`mcp.docs.astro.build`) and

package/templates/emdash/_gitignore

@@ -12,6 +12,9 @@ worker-configuration.d.ts
 # wrangler local state (local D1/R2 data lives here during `astro dev`)
 .wrangler/
 
+# node hosting target: SQLite db + local uploads
+data/
+
 # generated by the getvitops() integration on dev/build
 src/styles/
 public/vitops/

package/templates/emdash/astro.config.mjs

@@ -1,14 +1,21 @@
-import cloudflare from '@astrojs/cloudflare';
 import react from '@astrojs/react';
-import { d1, r2 } from '@emdash-cms/cloudflare';
 import getvitops from '@getvitops/astro';
-import { vitopsEmdash } from '@getvitops/emdash';
+import { vitopsEmdash, vitopsHosting } from '@getvitops/emdash';
 import { defineConfig } from 'astro/config';
 import emdash from 'emdash/astro';
 
+// Hosting seam: adapter + EmDash database/storage for the configured target.
+// Default 'cloudflare' (Workers + D1 + R2 — see wrangler.jsonc / src/worker.ts).
+// To move to a Node host (VPS / docker-compose / k8s) later:
+//   pnpm add @astrojs/node better-sqlite3
+// then pass { target: 'node' } here (or set the HOSTING env var, which wins).
+// Content lives in the database and media in storage, so switching hosts is a
+// data migration, not a rewrite — see README "Hosting".
+const { adapter, database, storage } = await vitopsHosting();
+
 export default defineConfig({
   output: 'server',
-  adapter: cloudflare(),
+  adapter,
   image: {
     layout: 'constrained',
     responsiveStyles: true,
@@ -17,8 +24,8 @@ export default defineConfig({
     // The EmDash admin UI (/_emdash/admin) is a React app.
     react(),
     emdash({
-      database: d1({ binding: 'DB', session: 'auto' }),
-      storage: r2({ binding: 'MEDIA' }),
+      database,
+      storage,
       // Vitops design-system blocks (image compare, copy snippet, banner,
       // disclosure, carousel) in the editor slash menu. Default 'integration'
       // script delivery: Layout.astro renders <Head /> from @getvitops/astro,

package/templates/emdash/package.json

@@ -10,6 +10,7 @@
     "deploy": "astro build && wrangler deploy",
     "deploy:dev": "astro build && node scripts/build-dev-wrangler.mjs && wrangler deploy --config dist/server/wrangler.dev.json",
     "init:github": "node scripts/init-github.mjs",
+    "init:hosting": "node scripts/init-hosting.mjs",
     "typecheck": "astro check",
     "build:css": "vitops generate -i design-system.json -f tailwind -o src/styles",
     "prepare": "vp config"
@@ -19,7 +20,7 @@
     "@astrojs/react": "^6.0.0",
     "@emdash-cms/cloudflare": "^0.31.0",
     "@getvitops/astro": "^0.4.0",
-    "@getvitops/emdash": "^0.1.0",
+    "@getvitops/emdash": "^0.2.0",
     "astro": "^7.1.2",
     "emdash": "^0.31.0",
     "react": "^19.2.4",
@@ -30,6 +31,7 @@
     "@cloudflare/workers-types": "^4.20260305.1",
     "@getvitops/cli": "^0.4.0",
     "typescript": "^6.0.3",
+    "vite": "catalog:",
     "vite-plus": "catalog:",
     "wrangler": "^4.111.0"
   },
@@ -40,7 +42,7 @@
       "onFail": "download"
     }
   },
-  "packageManager": "pnpm@11.9.0",
+  "packageManager": "pnpm@11.17.0",
   "emdash": {
     "seed": "seed/seed.json"
   }

package/templates/emdash/scripts/init-hosting.mjs

@@ -0,0 +1,150 @@
+// One-time hosting-platform setup for the target configured in
+// astro.config.mjs (vitopsHosting from @getvitops/emdash).
+//
+//   pnpm run init:hosting                 # cloudflare (the default target)
+//   pnpm run init:hosting -- --dry-run
+//   pnpm run init:hosting -- --target node
+//
+// cloudflare: creates the isolated DEV resources (D1 database + R2 bucket —
+// prod ones are auto-provisioned by wrangler on the first deploy) and, when a
+// GitHub repo is wired up (see init-github.mjs), sets the repo secrets the
+// deploy workflows need. The Cloudflare API token cannot be created via CLI,
+// so that one secret stays manual.
+//
+// Idempotent: safe to re-run — existing resources and secrets are just
+// updated/skipped. Requires wrangler auth (`npx wrangler login`).
+
+import { execSync, spawnSync } from 'node:child_process';
+import { readFileSync } from 'node:fs';
+
+const args = process.argv.slice(2);
+const dryRun = args.includes('--dry-run');
+const targetFlag = args[args.indexOf('--target') + 1];
+const target = (args.includes('--target') && targetFlag) || process.env.HOSTING || 'cloudflare';
+
+const name = JSON.parse(readFileSync('package.json', 'utf8')).name.replace(/^@[^/]+\//, '');
+
+function quiet(cmd) {
+  try {
+    return execSync(cmd, { encoding: 'utf8', stdio: ['ignore', 'pipe', 'ignore'] }).trim();
+  } catch {
+    return null;
+  }
+}
+
+function run(cmd, { capture = false } = {}) {
+  if (dryRun) {
+    console.log(`[dry-run] ${cmd}`);
+    return '';
+  }
+  console.log(`$ ${cmd}`);
+  try {
+    if (capture) {
+      const out = execSync(cmd, { encoding: 'utf8', stdio: ['ignore', 'pipe', 'inherit'] });
+      process.stdout.write(out);
+      return out;
+    }
+    execSync(cmd, { stdio: 'inherit' });
+    return '';
+  } catch {
+    console.error(
+      `\nERROR: \`${cmd}\` failed (see output above). Fix and re-run — the script is idempotent.`,
+    );
+    process.exit(1);
+  }
+}
+
+if (target === 'node') {
+  console.log(`Hosting target 'node' has no cloud resources to provision. To switch this
+project to a Node host (VPS / docker-compose / k8s):
+
+  1. pnpm add @astrojs/node better-sqlite3
+  2. In astro.config.mjs: vitopsHosting({ target: 'node' })  (or set HOSTING=node)
+  3. Build and run:  pnpm build && node ./dist/server/entry.mjs
+     Data lives in ./data/ (SQLite db + uploads) — persist it as a volume.
+  4. CI deploy for VPS/docker/k8s is not scaffolded yet — the deploy-*.yml
+     workflows are Cloudflare-specific; replace or disable them.
+
+For production on Node, move storage to S3-compatible and/or the database to
+Postgres via vitopsHosting({ node: { database, storage } }) — see the
+@getvitops/emdash README.`);
+  process.exit(0);
+}
+
+if (target !== 'cloudflare') {
+  console.error(`ERROR: unknown hosting target '${target}'. Valid targets: cloudflare, node.`);
+  process.exit(1);
+}
+
+// --- Preflight: wrangler auth ------------------------------------------------
+const whoami = spawnSync('npx', ['wrangler', 'whoami'], { encoding: 'utf8' });
+if (whoami.status !== 0) {
+  console.error('ERROR: wrangler is not logged in. Run `npx wrangler login` first.');
+  process.exit(1);
+}
+// "Account Name | <32-hex account id>" table row in `wrangler whoami` output.
+const accountId = (whoami.stdout.match(/\b([0-9a-f]{32})\b/) ?? [])[1] ?? null;
+
+// --- Dev D1 database ---------------------------------------------------------
+const devDb = `${name}-dev`;
+let devDbId = null;
+const d1List = quiet('npx wrangler d1 list --json');
+if (d1List) {
+  devDbId = JSON.parse(d1List).find((d) => d.name === devDb)?.uuid ?? null;
+}
+if (devDbId) {
+  console.log(`D1 database ${devDb} already exists (${devDbId}).`);
+} else {
+  const out = run(`npx wrangler d1 create ${devDb}`, { capture: true });
+  devDbId = (out.match(/"database_id":\s*"([0-9a-f-]{36})"/) ?? [])[1] ?? null;
+  if (!dryRun && !devDbId) {
+    // Output format drifts across wrangler versions — the list is authoritative.
+    const relist = quiet('npx wrangler d1 list --json');
+    devDbId = relist ? (JSON.parse(relist).find((d) => d.name === devDb)?.uuid ?? null) : null;
+  }
+  if (!dryRun && !devDbId) {
+    console.error('ERROR: could not determine the database_id of the created dev database.');
+    process.exit(1);
+  }
+}
+
+// --- Dev R2 bucket -----------------------------------------------------------
+const devBucket = `${name}-media-dev`;
+const bucketList = quiet('npx wrangler r2 bucket list');
+if (bucketList && new RegExp(`name:\\s+${devBucket}\\s*$`, 'm').test(bucketList)) {
+  console.log(`R2 bucket ${devBucket} already exists.`);
+} else {
+  run(`npx wrangler r2 bucket create ${devBucket}`);
+}
+
+// --- GitHub repo secrets (when the repo is wired up) -------------------------
+const hasGh = spawnSync('gh', ['auth', 'status'], { encoding: 'utf8' }).status === 0;
+const hasOrigin = quiet('git remote get-url origin') !== null;
+if (hasGh && hasOrigin) {
+  if (devDbId) run(`gh secret set DEV_D1_ID --body ${devDbId}`);
+  if (accountId) run(`gh secret set CLOUDFLARE_ACCOUNT_ID --body ${accountId}`);
+} else {
+  console.log(
+    '\nSkipping GitHub secrets (no gh auth or no origin remote — run init:github first).',
+  );
+}
+
+console.log(`
+Cloudflare dev environment ready:
+  D1:  ${devDb}${devDbId ? ` (${devDbId})` : ''}
+  R2:  ${devBucket}
+${
+  hasGh && hasOrigin
+    ? ''
+    : `  Set repo secrets once GitHub is wired up:
+    gh secret set DEV_D1_ID --body ${devDbId ?? '<id>'}
+    gh secret set CLOUDFLARE_ACCOUNT_ID --body ${accountId ?? '<account id>'}
+`
+}
+Still manual (CLI can't create API tokens):
+  1. Create a token with Workers + D1 + R2 edit permissions:
+     https://dash.cloudflare.com/profile/api-tokens
+  2. gh secret set CLOUDFLARE_API_TOKEN
+
+Prod D1/R2 are provisioned automatically by wrangler on the first \`pnpm deploy\`.
+Local \`pnpm deploy:dev\` runs need:  export DEV_D1_ID=${devDbId ?? '<id>'}`);