Vibe Coding — Deploy Guide

The Process

8 steps to your first deploy

Prepare your environment variables

Do this before touching any deploy platform

15 min

Every secret in your .env file needs to be set on the deploy platform as an environment variable — it will never see your .env file. This is the step that catches most first-time deployers off guard.

Open your .env file locally. Make a list of every variable in it.

Identify which values need to change for production (database URLs, API keys that have different prod/dev versions).

If you use Supabase: create a new project for production — don't use your dev project in production.

Critical: never commit .env to git

Before pushing to GitHub, run: git status and verify .env is not in the staged files. If it already was committed: rotate every secret immediately, then clean the git history.

Push your code to GitHub

A private repo is fine for everything below

10 min

If you haven't set up git yet

# In your project folder
git init
git add .
git status  # verify .env is NOT listed
git commit -m "Initial commit"

# Create repo on github.com, then:
git remote add origin https://github.com/your-name/your-repo.git
git push -u origin main
        

AI prompt for git help

"Walk me through pushing my project to a new GitHub repo. My project is in [folder]. I've never done this before. Show me every command."

Connect to Vercel (or Railway)

Takes about 5 minutes

5 min

Go to vercel.com, sign up with GitHub.

Click "Add New → Project". Select your repository from the list.

Vercel detects your framework automatically — verify it's correct.

Before clicking Deploy: go to "Environment Variables" and add every variable from your .env list.

Click Deploy. First deploy takes 2–5 minutes.

The most common first-deploy failure

Environment variables not set. Your app builds fine but crashes at runtime because it can't find the database URL or API key. Always set env vars before hitting Deploy.

Read the build logs — don't skip this

Where 80% of deployment problems are diagnosed

10 min

Every deploy platform shows build logs. Read them. If the deploy fails, the error is in there — often near the bottom. Don't guess; read.

✓

Look for lines marked ERROR or Failed — these are the actual problem.

✓

If you see "Module not found" — a package isn't in your package.json (maybe installed globally locally).

✓

If you see "Environment variable X is not defined" — add it in the platform's env var settings.

AI prompt for build log errors

My deploy on Vercel failed. Here is the build log:
[paste the full log]

What is causing this error and how do I fix it?
        

Test your live URL thoroughly

Production behaves differently from development

20 min

Your app is live but production is different from your local environment. Test every critical path before sharing the URL with anyone.

✓

Open the live URL in an incognito window (clears any cached state).

✓

Walk through your done-test from the Intent Worksheet — the exact user journey.

✓

Test on a real phone, not just a resized browser.

✓

Open browser DevTools → Console. Any red errors? Investigate before sharing.

Test with a slow network: DevTools → Network tab → throttle to "Slow 3G". Is it still usable?

Add a custom domain (optional but recommended)

yourname.com instead of your-project-xyz.vercel.app

20 min

A custom domain makes your project feel real and shareable. Domains cost ~$12/year on Namecheap or Cloudflare. The connection process takes 20 minutes, but DNS propagation can take up to 24 hours.

Buy a domain on Namecheap or Cloudflare Registrar (cheapest options).

In Vercel: Project Settings → Domains → Add your domain.

Vercel shows you DNS records to add. Copy them.

In your domain registrar: go to DNS settings, add the records Vercel provided.

Wait 10–60 minutes. Check vercel.com/dashboard — it shows when the domain is verified.

AI prompt for DNS help

"I bought a domain on [Namecheap/Cloudflare]. Vercel is telling me to add these DNS records: [paste records]. Walk me through exactly where to add them in [registrar name]."

Set up automatic deploys

Push to git → live in 2 minutes, every time

5 min

Vercel and Railway both watch your GitHub repository by default. When you push new code, they deploy automatically. This is already set up from Step 3 — but confirm it works.

Make a tiny change locally (add a comment, change some text).

git add . && git commit -m "Test auto-deploy" && git push

Watch Vercel dashboard — a new deploy should start within 30 seconds.

If it doesn't: check Vercel project settings → Git → confirm the correct branch is connected.

Set up monitoring — the 48-hour window

The first two days after launch are the most revealing

15 min

The first 48 hours after launch reveal problems your testing didn't find: real users do unexpected things with real data on real devices. Set up the minimum viable monitoring before sharing your URL.

✓

Error monitoring: Install Sentry (free tier). One script tag or npm package — shows you every error real users hit, in real time.

✓

Uptime monitoring: Add your URL to UptimeRobot (free) — emails you if the site goes down.

Analytics: Add Plausible or Fathom (privacy-friendly) or Google Analytics — see who visits and what they do.

The most important post-launch action

Watch Sentry for the first hour after sharing your URL. The first real errors will appear within minutes. Most are minor — a missing image, a failed API call in an edge case — but catching them immediately builds trust with early users.

Troubleshooting

The 5 most common deploy failures

Build fails: Module not found

A package installed locally isn't in package.json. Run npm install [package-name] --save locally, commit the updated package.json, push again.

App crashes: env variable is undefined

You forgot to set an env var on the platform. Check the platform's environment variables settings. Add every variable from your .env file. Redeploy.

API routes return 404

Usually a routing config issue. The AI may have configured routes that work in dev but need a different config in production. Search your framework + "production API routes" for the exact fix.

Database connection fails in production

Your production database URL is different from your local one. Check that DATABASE_URL (or equivalent) is set to the production database, not the local one. If using Supabase, use the connection string from the production project.

CORS errors in browser console

Your backend is not allowing requests from your production frontend URL. Update your CORS config to include the production domain. Never use wildcard (*) in production — add your specific domain.

When in doubt: ask the AI

Paste the full error message, the relevant config file, and the deploy platform you're using. "My Vercel deploy fails with this error. Here are my files. What's wrong and how do I fix it?"

Choose your deploy platform

8 steps to your first deploy

Pre-launch checklist

The 5 most common deploy failures