Deploy your site
by asking for it.
One index.html or a whole SaaS with its own backend. Say “deploy this” in Claude Code, Cursor, Codex or any MCP editor and it is live on HTTPS in about a minute, with a URL you can send to someone.
Free for 7 days, then $3/mo. No card to start.
ONE COMMAND · FREE HTTPS URL · YOUR DOMAIN WHEN READYA folder in.
A working URL out.
A page with no server is served straight from disk. An app with a backend runs on the port it asks for. You do not have to know which one you built, or tell us.
Say “deploy this site”. Build output is found on its own, so dist/ or out/ works without you saying so.
Every app is live at its own subdomain with a certificate. Nothing to configure and nothing to pay for.
Two DNS records and it serves on your hostname instead. The certificate is issued for you.
The bit you are
actually wondering.
Do I need a build step or a config file?
No. A folder with an index.html in it deploys as is. If you did run a build, we find the output on our own — dist, build, out, public and _site are all detected, so a Vite or Astro project works without you saying which.
Can I host more than a static site here?
That is the point. The same command deploys your Discord bot, your worker and your cron job, and a page that needs a real backend rather than just files on a CDN. They all run in one place, on one bill, and your agent manages every one of them the same way.
What about the backend and the database?
The backend is just your app listening on process.env.PORT, in Node or Python. The database is yours to bring — put the connection string in an environment variable, encrypted and write-only.
You do not have to
know which you built.
A page
Plain HTML, or the build output of Vite, Astro or a Next export. Served straight from disk, so there is nothing to install and nothing to start.
An app
Express, FastAPI, Next in server mode, anything that listens. It binds the port we give it and we route your URL to it.
Whatever it built,
this runs it.
Deploy a folder and get a URL that works. A plain page is served straight from disk; an app with a server of its own runs on the port it asks for. Both get HTTPS and both take one command.
You never open a panel.
Connect once, then deploy, read logs, restart and roll back by asking. No API key to copy.
Priced per app,
not per request.
No metering, no surprise invoice. Seven days free to start, and no card until you keep it.
SOLO for sites & web apps. A static site uses almost no memory, so the smallest plan is the honest answer. Move up when you want more than one.
What people ask
before they deploy.
Can I really deploy one HTML file?
Yes. A folder with an index.html in it is a valid deploy. There is no build step to configure, no framework to declare, and no server file to write.
What about a Vite, Astro or Next export?
Ship the build output and we find it. dist/, build/, out/, public/ and _site/ are all detected, so you do not have to tell us which one your framework used.
Does client-side routing work?
Yes. Unknown paths fall back to index.html so React Router and friends work, unless your site ships its own 404.html, in which case we serve that instead.
What if my app has a backend?
Then it is a normal app: listen on process.env.PORT and it is served on the same URL. Node and Python both work, and the database is yours to bring.
It is one command.
Go and send the link.
$3 a month, always on. First one free for 7 days, no card.
Cursor and VS Code open and install it themselves. For Claude Desktop or claude.ai, paste the URL under Settings → Connectors → Add custom connector.