Search Psychevus

12 results available.
↑ ↓ to navigate ↵ to open · esc to close

Lovable · Bolt · v0 · Supabase

Security checklist for Lovable, Bolt and v0 apps on Supabase

Twelve checks to run before you launch an app built with an AI builder on Supabase: RLS that's actually restrictive, keys, Edge Functions, storage, payments, AI features, auth settings and email.

Short answerBefore launch, confirm that every exposed table has RLS with policies tied to auth.uid() (not using (true)), no secret key is in the frontend, every Edge Function checks the caller, storage buckets are private where they should be, Stripe webhooks are signature-verified, AI features can't act without confirmation, and auth redirect URLs are restricted. Test with two real accounts trying to reach each other's data.

Data

  • RLS enabled on every table in public. Policies compare the row with (select auth.uid()) or check membership; no using (true) except on deliberately public content.
  • UPDATE policies have WITH CHECK, so a row can't be handed to another owner.
  • No policy reads user_metadata.
  • Views are security_invoker; SECURITY DEFINER functions are in a private schema with search_path set.

Keys and functions

  • Only the anon/publishable key reaches the browser. No service_role, Stripe secret or AI provider key in VITE_ variables.
  • Every Edge Function resolves the user from the Authorization header and checks they may touch the record they ask for.
  • Functions that call paid APIs are rate-limited per user.

Files, money and AI

  • Storage buckets with private documents are private; downloads use short-lived signed URLs; file names aren't guessable.
  • Stripe webhooks verify signatures and re-read objects from Stripe before granting access.
  • AI tools act with the user's permissions and need a click for anything consequential.

Auth and email

  • Redirect URLs are restricted to your domains; email confirmation is on.
  • Password reset doesn't reveal whether an account exists.
  • Your sending domain has SPF, DKIM and an enforcing DMARC policy, so nobody can email your users as you.

Then test like an attacker

Create two accounts in different workspaces. From account B's browser console, try to read, change and delete account A's rows, files and invoices, and call each Edge Function with A's IDs. Automated scanners can't do this step for you, because it needs two identities and knowledge of what should be private.

Simulate your RLS policies (free, in your browser)

Questions

My builder has a built-in security scan. Isn't that enough?

Use it, and treat a pass as a minimum. Checking that a policy exists is different from checking what it allows; a using (true) policy exists.

How long does this take?

An afternoon for a small app if you know where to look. The Launch Check does it for you in two business days, with evidence and fix prompts.

Updated 18 November 2026 · by

Reading about a flaw is the easy part. A Launch Check finds every one of these in your app, with evidence and a fix prompt for each, in two business days.

Get a fixed quote

Related

YOUR VISIT, YOUR CHOICE

Privacy, by intention.

You decide what stays on your device. You can change your choice at any time.

Essential

Saves your privacy choice. Administrator sign-in uses a secure session cookie.

Always on
Full privacy notice ↗