Qodeup Africa
Request an audit
Blueprint Audit · for apps built with AI builders · 48 hours

Your app works. Before customers, investors or a bigger client rely on it, find out what it's hiding. We take it through 90 checks across seven areas, prove every finding with the place in your code, and hand you the fixes in the order to do them.

90 checks · 7 areas · 48 hours · 1 signature

What we check.

90checks
7areas
48hours to the report
  • Security.

    Who can read what, who can call which function, and whether a secret has reached the browser.

  • Data.

    The rules in your database, what a stranger can see, and whether deleting something really deletes it.

  • Product.

    Whether every button does what it says, from sign-up to payment, on a laptop and on a phone.

  • Reliability.

    What happens when a payment provider, an email service or the AI model is slow or down, and whether anyone hears about it.

  • AI.

    Spending caps on the calls that cost you money, and whether a user can talk the model into something it shouldn't do. This area runs only when your app calls a model.

  • Performance.

    How fast it loads, how lists behave as your data grows, and whether search engines can read your public pages.

  • Shipping.

    How code reaches production, whether a bad release can be rolled back, and whether errors get recorded.

Some checks switch on only when your app needs them: sign-in, payments, uploads, webhooks, scheduled jobs. Apps on Supabase get a set of checks of their own.

How the audit runs.

  1. 01

    Access

    You give us read-only access to the code, and you run one read-only script on your database. It reads the structure and the access rules, never your users' rows. We pin the exact version we audit.

  2. 02

    Seven reviews

    One reviewer per area works through the checks in parallel and cites the file and line behind every verdict. Scanners run alongside for leaked secrets, packages with known holes, and page speed.

  3. 03

    The journey trace

    One reviewer follows a real user's path through every layer: sign up, pay, use the main feature, delete the account. Most serious problems sit where one part of the app hands over to the next.

  4. 04

    The live run

    We use your app as a visitor and, with your OK, as an ordinary signed-in user, on desktop and on phone. Every save, every delete and every email to your users is blocked while we do it.

  5. 05

    The challenge

    A second, independent reviewer tries to disprove every finding, and the urgent and serious ones are reviewed again, blind, by a separate system. Anything that doesn't hold comes out. A borderline grade gets a written ruling.

  6. 06

    The sign-off

    Before anything reaches you, every citation is checked to open where it says it does, and Orezi Emeotu reviews the evidence and signs the report.

  7. 07

    The handover

    The report, a walkthrough call, and a re-check of the urgent items at 30 days.

What you get.

Same day

Anything urgent

A leaked secret, data strangers can read, a way to run up your bill. You hear the day we find it, with the proof, so you can close it before the report lands.

48 hours

The report

One private report page with two views of the same findings: a founder view in plain words, and a developer view with the code.

Two PDFs: a short one for your board or investors, and the full one for whoever fixes it.

One fix file for your repository, with a paste-ready prompt for every finding.

What already works, and every check with its result, so you can see what passed as well as what failed.

30 mins

The walkthrough

We go through the findings with you. Then you choose who fixes them: your developer, your AI builder with our fix file as its prompt, or us.

Day 30

The re-check

We confirm the urgent problems are closed.

What a finding looks like.

From our sample report, written for an invented invoicing app.

Anyone can read every customer's invoices without signing in

Where: the invoices table, which has no access rules.

What it means: a stranger with the public app key can download every invoice, client name and amount. That is a data breach you would have to report.

How bad: urgent. Fix before anything else. Under an hour of work.

The fix: turn on row-level security for invoices, and add rules so each signed-in user reads and writes only their own rows.

Prompt for your AI builder: “In Supabase, enable and force row-level security on the public.invoices table. Add select, insert, update and delete policies that allow a row only when owner_id = auth.uid(). Before changing anything, list any page or function that reads invoices without a signed-in user, and tell me.”

See the whole sample report, or download it as the founder PDF or the developer PDF.

What it costs.

$750

Per app

One web app on one domain and its subdomains: front end, backend, database and server functions. Paid by card through Stripe, in US dollars.

$600 back

If we find nothing serious

If the report has no urgent or serious finding that survived our second reviewer, we refund $600 and you keep the report. The $150 we keep pays for the review.

$500

The first five audits

In exchange for permission to publish what we found, with your app, your company and every person in it anonymized. You see the text before anything goes out. If we find nothing serious, $400 comes back.

From $5,000

If you want us to fix it

A fix sprint, quoted from your fix plan, and only if you choose us. The audit stays separate so it stays impartial. A care plan at $500 a month keeps it fixed.

What we need, and what we never ask for.

  • Read-only access to your code repository, as a collaborator.
  • The result of one read-only script on your database.
  • Your app's address, and a staging address if you have one.
  • A test account. With your OK, we create one through your normal sign-up. If sign-up is invite-only, you create one for us.
  • About 20 minutes of your time, for the kickoff.

We never ask for production passwords, your admin login, payment keys, or your users' data.

Questions.

Is my code safe with you?

Access is read-only, so we can't change anything. Our copy sits in a separate workspace, used only for your audit, under confidentiality terms you accept before we start. If we find personal data in your code, we name the file and never copy what's in it. Our copy of the code is deleted when the audit closes. The report and its evidence are kept for three months, then deleted.

Will you break my live app?

No. We use it the way a visitor or an ordinary user would. Saves, deletes and emails to your users are blocked. We don't load-test, we don't attack, and we stop at your checkout unless it runs in test mode.

Which apps can you audit?

Web apps, whether Lovable, Bolt, Cursor, Replit or v0 built them, or a developer did: Next.js, Vite and React, Python with FastAPI, or a TypeScript service, on Postgres. Apps on Supabase get extra checks of their own. We don't audit native mobile apps. If your mobile app talks to a web backend, we can audit the backend.

What counts as one app?

One domain and its subdomains. app., api. and admin. on the same domain are one audit.

We're an agency. Can we order an audit for a client?

Yes. You order it and sign the terms as the buyer, for your client's app. The report carries Qodeup's name and Orezi Emeotu's signature, so it can't be resold or rebranded.

What if you find nothing serious?

We refund $600, and you keep a report showing everything we checked and what passed. That's worth having when an investor or a client asks.

What if I disagree with a finding?

Every finding shows the exact place in your code. Tell us which one within 14 days, and we go through it with you in a follow-up, with screenshots. We recorded the version we audited, so we're both looking at the same code. One revised report is included. A finding whose evidence doesn't hold comes out, and if that leaves nothing serious, the refund applies.

Can I show the report to investors?

Yes. The short PDF is written for that. It shows what we checked and found on the day we audited. It isn't an investment guarantee, and no audit finds every defect.

Is this a penetration test?

No. A penetration test takes weeks, costs five figures, and attacks your live app. This reads how your app is built, in 48 hours, at a fixed price, and it's aimed at the mistakes AI builders make.

Isn't the scan in my AI builder enough?

It checks the patterns it was written to catch, one at a time. It can't tell you that an ordinary user can make themselves an admin, because each piece of that looks fine on its own. We run scanners too. The difference is everything around them.

Do you use AI to do the audit?

Yes, the way a careful engineering team does. AI reviewers do the reading at speed, a separate system checks the serious findings blind, and a person signs every report. Your code goes to those providers only for your audit, under account settings that stop it being used to train their models. The privacy notice names every provider.

Does it cover compliance?

It flags the gaps, for example around personal data. A lawyer or a compliance auditor signs off on compliance. We don't.

How do I pay?

By card, through Stripe, in US dollars. Work outside Nigeria is contracted through digiskillsng Solutions LLC, a US company under the same management as Qodeup Africa. Prices exclude any taxes that apply to you.

What time zone do you work in?

West Africa Time, UTC+1. We book calls in your working hours, US time zones included.

How soon can we start?

As soon as you've paid and access is in. The 48 hours start then, counted Monday to Friday. If we're waiting on something from you, we tell you exactly what, and the clock waits with us.

Know what's in your app before someone else finds it. Request an audit by email, or book a 20-minute call. You pay when you order, we start when access is in, and the report lands 48 hours later.