Guide
You vibe coded an app. Now make it live.
The build was the fun part. Getting it online — with a real address, a database, and a login — is the part that stalls. Your agent can do it for you.
Why the last step is the hard one
When you vibe code, the app comes together fast — it runs on your machine and feels real. Then you want to show someone, and you hit a wall: it has no public address, no place to store data, and no way for people to sign in. Those are infrastructure jobs, and they are exactly what you were avoiding.
EdgeSpark removes that wall. Because it is agent-native, the same agent you built with can run the deploy. It provisions a database, sets up login, and ships the app to the edge — and hands you a link.
How to ship it
From your machine to a live link.
- 1
Give your agent the skill
Paste one line into the agent you built with — it installs the CLI and the EdgeSpark skill.
›Fetch and follow the guide on https://cdn.edgespark.dev/onboarding/install.txt - 2
Pick a workspace
Unlock the live URL your app deploys into — less than a coffee.
$1/mo · Starter plan
- 3
Tell it to deploy
Say it in plain words. You get back a real URL with a database, login, and storage.
›deploy this
What “live” actually gets you
A real, public link
Your app goes live at your-app.edgespark.app on a global edge network — a URL you can text to anyone.
Data that sticks
A managed database means sign-ups, posts, and saved settings persist — not reset on every reload.
Real users
Managed login lets people make accounts, so your app is more than a single shared demo.
Frequently asked questions
- I vibe coded an app — how do I publish it?
- Install the EdgeSpark skill in the agent you built with, pick a workspace, and tell the agent to deploy. It provisions a database, auth, and storage and returns a live URL.
- Do I need to understand servers or deployment?
- No. The agent runs the deploy through the skill. You describe what you want in plain language and get back a working link.
- Which agents does this work with?
- Claude Code, OpenAI Codex, Gemini CLI, Cursor, and GitHub Copilot, plus any other MCP-compatible agent via the generic skill.