crafyne

CLI to deploy static sites, edge functions, and fullstack workers to Crafyne. Zero dependencies for static sites (needs Node 18+; edge functions need esbuild in your project).

Install

Install once, globally, so the crafyne command is available everywhere:

npm install -g crafyne

Or skip the install — replace crafyne with npx -y crafyne in any command below.

Sign in (once, via the browser)

crafyne login

Your browser opens on crafyne.app — click Allow, done. The token is stored in ~/.crafyne/config.json. No token copy-pasting.

Your sites are served from crafyne.site — a separate registrable domain from the dashboard, so your JavaScript never runs on the same origin as the Crafyne session cookie. A custom domain always wins if you've attached one.

Deploy (auto-build)

From a project folder (one with a package.json), Crafyne detects the framework, runs the build, and uploads the output:

crafyne deploy

→ • Detected: Vite → → Build: npm run build → ↑ Uploading … → ✓ Live at https://namek.crafyne.site.

No project yet? You don't need the dashboard. If the slug doesn't exist, the CLI offers to create it (using the folder name as a first guess) and saves it to crafyne.json so the next deploy doesn't ask again. In CI (no TTY), pass --create so nothing hangs.

Supported: Vite, Astro, Next (export), Nuxt, SvelteKit, CRA, Vue, Angular, Gatsby.

Already have a built folder (or a plain static site)? Skip the build:

crafyne deploy ./dist        # upload the folder as-is
crafyne deploy --no-build    # use ./dist without building

If output detection guesses wrong, say so: --output <dir>.

Edge functions (Pro plan)

Add a functions/ folder in your project root — each file becomes a serverless route, while everything else stays static (and free to serve):

functions/api/contact.js      →  /api/contact
functions/api/rsvp/[id].js    →  /api/rsvp/[id]   ([id] matches one path segment)

Each file exports an onRequest handler:

export function onRequest({ request, env, params }) {
  return Response.json({ ok: true, id: params.id });
}

Deployed automatically on crafyne deploy. Requires esbuild in your project: npm i -D esbuild.

Fullstack / SSR (Pro plan)

Ship a real server worker. If your build produces a _worker.js in the output folder (the Cloudflare adapter convention used by Astro SSR, SvelteKit, Qwik, and others), Crafyne deploys it as a fullstack worker — every request runs your code, and the rest of the folder is served as fast, free static assets via env.ASSETS:

export default {
  async fetch(request, env) {
    const url = new URL(request.url);
    if (url.pathname === "/api/hello") return Response.json({ ok: true });
    return env.ASSETS.fetch(request);   // static files
  },
};

No _worker.js at the root? Point to it in crafyne.json: { "type": "worker", "main": "dist/_worker.js" }.

Preview before going live

crafyne deploy --preview

The deploy is still stored (immutable, with its own hash) but does not replace your live site. You get https://namek--a1b2c3.crafyne.site to check first. Happy with it? Promote it:

crafyne deploys                      # version history (● = live)
crafyne rollback --deploy a1b2c3     # make that version live

Commands

| Command | What it does | |---|---| | crafyne login | sign in via the browser, store the token locally | | crafyne logout | remove the saved credentials | | crafyne whoami | show the signed-in account | | crafyne deploy [folder] | build (optional) + upload; deploys functions/ and _worker.js too; creates the project if missing | | crafyne deploy --preview | upload without going live | | crafyne projects | list your projects and their URLs | | crafyne deploys | version history of this project | | crafyne rollback --deploy <hash> | make an older version live again | | crafyne delete --project <slug> | delete a project and all its deploys (permanent) | | crafyne open | open the site in your browser | | crafyne --version | installed CLI version |

Deleting a project

crafyne delete --project blog7

Deleting takes the site down and erases every deploy in its history — there is no undo. That's why the confirmation asks you to retype the slug rather than press y. In CI, pass --yes (and think twice).

CI/CD (no browser)

Use a token from the dashboard (Account → API & CLI → Regenerate). Add --create so the project is created automatically when it doesn't exist — nobody can answer a prompt in CI:

CRAFYNE_TOKEN=crfy_live_xxx npx crafyne deploy ./dist --project namek --create

crafyne.json (optional)

{ "project": "namek", "output": "./dist" }

Then plain crafyne deploy is enough. The CLI writes this file for you when it creates a project.

Publishing (Crafyne maintainers)

cd cli
npm version patch
npm publish
npm view crafyne version    # confirm it went live — npm may hold it in staging for approval

How login works

1. POST /api/cli/start → returns a code + an authorization URL (/cli?code=…).
2. The browser opens that page; your logged-in session (cookie) calls POST /api/cli/approve, which mints a token and ties it to the code.
3. The CLI polls POST /api/cli/poll → receives the token once, then stores it locally.