RLS that lets everyone read everything
Row level security switched off, or switched on with a policy of USING (true), which still shows as enabled while every row is public. The most common hole in Lovable apps by far.
Security review
We do security reviews of Lovable apps: $350 for a written audit in 2 days, or from $750 with every fix applied and re-tested (prices from October 2026). Lovable's own docs say its Quick and Deep security scans "do not replace a thorough security review" and suggest a professional review for apps that handle sensitive data. Its Aikido pentest ($100 a test) is automated; we test by hand, logging in as two real users to try every Supabase RLS policy, then fix what we find in your repo.
Audit $350, report in 2 days. Audit with fixes from $750.
Sound familiar?
Row level security switched off, or switched on with a policy of USING (true), which still shows as enabled while every row is public. The most common hole in Lovable apps by far.
Service role keys, Stripe secret keys or OpenAI keys shipped to the browser, where anyone can copy them from dev tools.
Admin checks done in React instead of the database, so changing one value in the browser unlocks the admin panel.
Invoices, ID documents and private uploads sitting in buckets marked public, reachable by URL.
Prices
Know where you stand
$350
Found, fixed, re-tested
from $750
Prices from October 2026. Large apps (many roles, many tables or several edge functions) get a fixed quote before we start. Many audits elsewhere stop at the report; ours can include the fixes.
Want escrow? Payment through an Upwork contract is possible on request.
What you get
Fixed price, agreed after a free call. No hourly surprises.
How it goes
Tell us what the app does and who uses it. We agree a fixed price.
Invite us to your Lovable project and Supabase organization using our two-minute guides.
We test, write up and fix. Nothing changes in production without your OK.
You get the report, the fixes and a clean re-test.
Sample report
Every finding comes with a severity, proof that it is real, and the exact fix. This excerpt is a made-up example for this page, built from the kinds of problems we find most often in Lovable apps. Your report covers your own tables, functions and buckets.
| Severity | Finding | Proof | Fix |
|---|---|---|---|
| Critical | Every order is readable by any signed-in user | RLS on orders is enabled with a SELECT policy of USING (true). Test account B listed 312 orders belonging to account A through the REST API with its own login. | Replace the policy with USING (auth.uid() = user_id), add a matching policy for admins via a role check in the database, re-test with both accounts. |
| High | Users can make themselves admin | profiles has an UPDATE policy for the row owner on all columns, and the admin panel only checks profiles.role in React. Account B set its own role to admin with one PATCH request. | Move role to a table users cannot write, or restrict UPDATE to safe columns, and check the role in RLS and edge functions, not only in the UI. |
| High | Uploaded invoices are public | The invoices storage bucket is marked public, so any file URL opens without a login, including URLs pasted into support emails. | Make the bucket private, serve files through signed URLs that expire, and update the stored links. |
| Medium | Email function can be called by anyone | send-email has verify_jwt turned off and no rate limit, so a script can send mail from your domain through it. | Turn verify_jwt on or check a shared secret, validate the recipient, and add a per-user rate limit. |
Reports are in plain English and ranked by severity, so you can fix the critical items first. With the Audit + fixes package we apply every fix and re-test the same way.
“Lovable 2 Live has a real focus on quality. They're the team we trust on Klick.now when something has to work in production, and they're always straight with us about what's worth building and what isn't. Easy recommendation.”
Questions
We do, as a fixed-price job: $350 for a written security audit of your Lovable app in 2 days (Supabase RLS policies tested with two real accounts, exposed keys, auth, edge functions and storage buckets, ranked by severity with proof), or from $750 with every fix applied in your repo and re-tested. Lovable's own security docs say its Quick and Deep scans do not replace a thorough security review and recommend a professional review for apps with sensitive data; this is that review. Checked October 25, 2026.
We use OWASP ASVS 5.0 (published May 30, 2025) as the checklist for the parts of a Lovable app that break most often. The audit covers V8 Authorization (Supabase RLS tested with two real accounts), V6 Authentication, V4 API and Web Service (edge functions and the Supabase REST API), V5 File Handling (storage buckets) and V13 Configuration (keys and secrets shipped to the browser). It is not a full ASVS Level 1 verification and not a certified pentest. If a customer or an auditor asks for one of those, we say so on the free call instead of selling you the wrong report. Checked November 5, 2026.
It is a hands-on test of the attacks that matter most in Lovable apps: a second account trying to read and change the first account's rows through the Supabase API, users making themselves admin, edge functions called without a login, and private files reachable by URL. It is not a certified pentest report for compliance; Lovable puts a traditional pentest firm at $5,000 to $50,000. Lovable's Aikido pentest is automated and costs $100 a test. We read the findings of Lovable's Deep scan too, then test what a scanner cannot know: which user should see which row.
A written audit is $350: RLS, exposed keys, auth, edge functions and storage, ranked by severity with proof. The audit with every fix applied and a re-test starts at $750. If you start with the audit and then want the fixes, the $350 is credited.
You get the written report 2 days after we have access to your Lovable project and Supabase. If you choose the fixes too, we agree the timeline with the fixed price, then re-test once they are in.
Not necessarily, but security depends on what the AI generated for your data. Lovable runs a Quick scan on every publish and offers a Deep scan of your code, which are a good start. They cannot know your business rules, like which user should see which row. That is where most real problems are.
The April 2026 incident exposed chat histories of public projects on the free plan. Your app's own database security is a separate question, and it is the one this audit answers.
No. We test access rules with throwaway accounts and only look at data where a finding needs proof.
Fixes are tested against your real user flows first, and we go through each change with you before it ships.
Fifteen minutes is usually enough to tell you what's wrong and what it costs to fix.