Skip to content
ScanAnt

Lovable security check: is your Lovable app safe to launch?

Lovable gets you a working app in an afternoon, and it usually puts Supabase behind it. The generated front end talks to your database directly, so your row-level security policies are the only lock. ScanAnt recognises Lovable apps, scans what they publish and tells you what a stranger could read.

Free, no signup. We only read what your site already shows every visitor.

What we check

Each item is tagged with the scan that reports it. The quick scan is free and needs no signup. The others run only on sites you have proved you own.

  • Lovable and Supabase detection

    Quick scan

    We recognise lovable.app and custom-domain Lovable sites, find your Supabase project URL and anon key and label them as normal to expose.

  • service_role and AI keys in the bundle

    Quick scan

    The admin service_role key, OpenAI, Anthropic, Groq, Stripe and other keys that a generated front end sometimes receives. Masked, never reused.

  • Trust placed in the browser

    Quick scan

    Prices decided in the browser, admin access decided by a client-side flag, and paid features unlocked by a local setting.

  • AI features called from the browser

    Quick scan

    An AI provider called directly with a key, an AI endpoint that looks open to anyone and a system prompt shipped in the page.

  • Tables readable without signing in

    Verified scan

    The most common Lovable problem: a table with RLS off or a policy that lets everyone through. We save names and counts, never the values.

  • Storage and sign-up settings

    Verified scan

    Listable storage buckets, open sign-ups, sign-ups without e-mail confirmation and anonymous sign-ins.

  • Edge Functions and RPC

    Deep scan

    Functions your app calls without checking the caller, and database functions callable without login.

  • RLS in your project files

    Code scan

    Upload an export of the project to check migrations for missing RLS, always-true policies and users who can change their own role.

  • Quick scan: No signup. Reads only what every visitor receives.
  • Verified scan: After you prove you own the site. Read-only.
  • Deep scan: After you prove ownership and tick the deep option.
  • Code scan: You upload your project as a ZIP.

No scan can promise to find every problem. A clean result means no known problems were found on what the scan could see.

Common mistakes

  • Shipping with RLS off because the preview worked

    Without policies the anon key in your front end can read, and often write, whole tables.

  • Asking the chat to "fix permissions" without checking

    A generated policy may use USING (true), which looks like a fix but allows everyone.

  • Putting a payment or AI key straight into the app

    Lovable front ends are public; the key is visible to every visitor.

  • Admin and paid features guarded only in the interface

    Hiding a button is not access control; the database call still works.

  • Forgetting that CVE-2025-48757 is about this pattern

    Ineffective row-level security put many Lovable apps at risk, and it is easy to repeat in a new one.

How to fix it

  1. Run a scan, then copy the fix prompt from each finding into Lovable. Prompts are written for Lovable's chat.

  2. Enable RLS on every table and write owner-based policies, then retest with a second account.

  3. Move secret keys into Edge Function secrets and call the function from the front end.

  4. Check roles and prices on the server side, not in the interface.

  5. Rescan after every large prompt, since a new prompt can change policies again.

Questions

Can I scan before publishing?

The live scan needs a reachable URL. Before that, export or sync the project and upload it as a ZIP for the code scan, which checks your SQL migrations.

Do I need Lovable access?

No. ScanAnt works from outside, from the public site. For the full scan you prove you own the site by adding a token file or a meta tag, which fits sites where you cannot edit DNS.

Does Lovable already check security?

Platforms add their own checks, and they help. An outside scan sees what a stranger sees, and it can be repeated after each change, which is why it is a useful second opinion.

Will the fix prompts break my app?

They are concrete, narrow instructions, but any change can have side effects. Apply one at a time and test the main flows before you move on.

Is the full scan free?

Yes. The quick scan is free without signup, and the verified full scan is free once you prove the site is yours.

Run the scan now

Free for any public site, no signup. Need the verified, deep or code scan? Choose the level on the scan page.

Choose a scan level →