1. Home
  2. Secure code review
  3. Pre-launch security audit

Pre-launch security audit for your AI-built app

A pre-launch security audit finds and fixes the problems in your app while nobody is affected yet. Here is what a proper one covers, what you get and how to know whether you need it.

A pre-launch security audit is the last check before real users and real data reach your app. It finds the problems you can still fix quietly, before anyone is affected. This page explains what an audit covers, when you need one and what you should have in hand when you start. If you only want a quick look first, read the vibe coding security checklist.

Do you need a pre-launch audit?

Not every project does. Use this as a guide.

  • A self-check is enough if the app holds no personal data, takes no payments and has no logins. Work through the vibe coding security checklist.
  • Get a review if you store customer or user data, take payments, let users upload files, connect to a language model, or sell to businesses that will ask about your security.
  • Get one urgently if the app is already live and you have never looked at its security. Ask us about a code review now rather than after launch.

If you handle personal data of people in the EU, GDPR requires appropriate technical measures. A documented review is evidence that you took reasonable care.

What a good audit covers

A useful audit goes beyond running a tool. Here is what we look at in an AI-built app, and how it maps to the OWASP Top 10, the standard list of web risks.

Access control and authentication

Can a user reach another user's data? Are admin functions protected on the server? This covers broken access control and broken authentication, the top items on the OWASP list. We test with several account types and read every route that touches data.

Secrets and configuration

We scan the code and its git history for hardcoded secrets and exposed API keys, check environment variable handling and look at debug flags, CORS and security headers.

Injection and input handling

SQL injection, cross-site scripting and server-side request forgery all start with input that was trusted too much. We trace user input from the form to the database and the server's outgoing calls.

Database rules

For Supabase and similar backends we review row-level security policies against your business rules. See Supabase RLS and common AI-code mistakes for the common failures.

Dependencies

We check every package for known vulnerabilities and confirm that none are hallucinated dependencies that an attacker could have registered.

Business logic

This is where people beat scanners. Can a user change the price in the request? Apply a coupon twice? Skip a payment step? A senior engineer asks these questions because they know how real attacks work.

AI features

If your app sends user text to a model, we look for prompt injection paths, over-broad model permissions and leaked provider keys.

What you get at the end

  • A written report in plain language, with findings ranked by how much they matter.
  • For each finding, what it means for your business and how to fix it.
  • In the paid review, the fixes themselves, applied to your code and explained, so you can see what changed and why.

That last part is the difference from a scanner or a report-only audit. You do not get a list of homework. You get a fixed app.

How to prepare

  1. Give us read access to the repository, or a copy of it.
  2. Tell us the stack and what the app does with money and personal data.
  3. Share the live or staging address, if there is one. Never send us live passwords.
  4. List the user roles in your app, such as customer and admin.

What it costs and how long it takes

Our code review starts from 500 euro, which is indicative. The price depends on the size and risk of the app. We agree the scope and price with you before any work starts. Timing depends on the size of the codebase, so we agree it with you up front.

Before you book

Read is your vibe-coded app secure if you are not yet sure you need an audit, and securing apps built with Lovable, Cursor, Bolt or Replit if you built with an AI builder. When you are ready, talk to us. The code review service page explains the whole process.

Common questions

What is a pre-launch security audit?

A pre-launch security audit is a review of your app's code, configuration and dependencies for security problems before real users arrive. Findings are ranked and fixed so you launch with known risks closed.

When should I do a security review of my app?

Before launch, and again after large changes such as adding payments, user accounts or a new data source. Earlier is cheaper, since fixes made before users arrive involve no cleanup.

What is the difference between a code review and a penetration test?

A code review reads your source to find flaws and fix them. A penetration test attacks the running app from outside to see what breaks. For a young app, a code review usually finds more per euro.

How much does a security review of an AI-built app cost?

Our code review starts from 500 euro, which is indicative. The final price depends on the size of the app. We agree the scope with you first, so you know what you are paying for.