Find the cracks before your users do.

With a specialized code audit for your AI-built app

White marble bowl with cracks repaired in gold, in the kintsugi style

I help you find the cracks when fixing them is still cheap.

Hands tying a broken white plate back together with twine, beside a pile of shards
Kintsugi (金継ぎ, “golden joinery”) is the Japanese art of repairing broken pottery with urushi lacquer and powdered gold, silver, or platinum.

These cracks came from a real app.

Excerpts from a real audit of an app built with Lovable, anonymised.

Launch advice

NO-GO: fix the P0 showstopper first, then re-test.

ready-made fix prompts
7
P0 showstopper
1
P1, fix before launch
5
P2 improvement
1
  • P0

    An admin feature skips the database’s security rules

    Impact: If its checks fail or are fooled, the whole database is exposed: a full data leak.

    Fix: Validate every request strictly and enforce the admin checks on the server.

  • P1

    The daily usage limit only exists in the browser

    Impact: Anyone can get around it and make unlimited AI requests, paid for by the app’s owner.

    Fix: Count and enforce usage on the server.

  • P1

    Password reset links can point to the wrong place

    Impact: Resets fail at random, and reset emails can carry links that don’t lead back to the real site.

    Fix: Build reset links from one fixed, trusted address, and check the session before a new password is saved.

  • P1

    The server accepts requests from any website

    Impact: Any site can call the app’s server functions, and some calls may break once it goes live.

    Fix: Only accept requests from the app’s own domain, and check what every request contains.

Every crack found, at a price you know up front.

What I check

  • Login, roles and database security
  • Login protection and abuse prevention
  • File uploads and input validation
  • Errors, logging and vulnerable packages
  • Risks specific to your app
  • Code quality and maintainability

What you get

  • A clear go or no-go launch verdict
  • A copy-paste fix prompt for every finding
  • A risk report ranked by priority, down to the exact line
  • A checklist to confirm the fixes worked

From €490

€70 per hour, 1 hour per 1,000 lines, report included

Report in 3-5 business days

lines

Not sure? Ask your AI tool to count the lines of code
  • Over 30,000 lines? Ask for a quote, with price and timing agreed together
  • The price I confirm before we start is the price you pay

  1. Book a free call

    We go through your app and what worries you.

  2. I review it

    By hand: your code, database rules and dependencies.

  3. You fix it, or I do

    Paste the prompts into your AI tool and work through the checklist, or have me fix it for you.

Book a free call
Lucas Kaas on a bench, with a ring-tailed lemur sitting next to him

Hi, I'm Lucas.

One of the worst feelings is being confident in something you built, only to find out it was full of cracks. Pretty much every developer has been there, me included.

AI tools let anyone build an app now, which is great. But AI isn't perfect, and I'm here to find what it missed.

Book a callorReach out

Experience

  1. Business Owner

    – Present

    KAAS Software

    Freelance software engineer and advisor for businesses.

  2. Full-stack Engineer

    – Present

    KAAS Software

    Responsible for everything behind BlueGemBot, a tool that tracks CS2 skin listings across online marketplaces. bluegembot.com

  3. Backend Engineer

    –

    Mozo

    Built new features for both business partners and end users.

  4. Full-stack Developer

    –

    Folderz.nl

    Built new features for the website and fixed problems, working closely with the rest of the business.

  5. Software Development Intern

    –

    ABN AMRO

    During my graduation internship, I built a dashboard that brings important system information together in one place, with charts that make errors and warnings easy to spot.

Good to know.

Do I need to be technical?

No. The report explains each risk in plain language. I fix the issues for you, or the fixes are provided as highly specific prompts that you can apply yourself.

How much does an audit cost?

€70 per hour, with the report and fix prompts included, from €490 for the smallest audit. To see your price:

  1. Count your app's lines of code. Ask your AI tool: "How many lines of code does this project have? Count only code files, not configuration, images or libraries."
  2. Or, if you use a terminal, run npx cloc . --exclude-dir=node_modules,dist,build in the project folder and take the total under "code".
  3. Enter that number in the price calculator.

Before we start, I count your code myself and tell you the full price, so you know it up front. The report is ready in 3-5 business days, or on an agreed date for apps over 30,000 lines.

Which tools and stacks do you cover?

Apps built with any of these:

  • Lovable
  • Bolt.new
  • v0 by Vercel
  • Replit
  • Base44
  • Firebase Studio
  • Cursor
  • Windsurf
  • Claude Code
  • GitHub Copilot
  • OpenAI Codex

Including the services they run on, such as Supabase and Firebase.

Do you need access to my code?

Yes, the audit is a review of your actual code and database rules. You share the code through a read-only GitHub invite, which you can remove as soon as the audit is done. I don't need access to your users' data. If your code is confidential, I'm happy to sign an NDA beforehand.

Do you fix the issues too?

Yes. I can fix the issues for you if you wish, this is optional.

Do you do other work besides audits?

Yes. I also take on general development projects, and run standalone security scans for apps that don't need a full audit.

Reach out for more information.

Book a call

Book a free intro call