Audits for AI-built apps

Your AI-built app, checked before real users find the holes.

You built it with Lovable, Bolt, Cursor, Replit or Claude Code, and it works. Now it is about to hold real users' data and take real payments. We audit it in 48 hours, fix what matters, and stay on call.

  • Lovable
  • Bolt
  • Cursor
  • Replit
  • Claude Code
  • v0

What you get

An audit, a fix, and someone on call

Free scanners tell you something is wrong. We tell you what a stranger could do with it, fix it, and prove the fix.

The audit, in 48 hours

A written report, ranked by severity, in plain English. Each finding says where it is, what a stranger could do with it and exactly how to fix it, with a fixed quote for the fix work.

$149 to $349 by size

The fix package

We fix every critical and high finding at a fixed price and give you before-and-after proof for each one.

$1,500 to $3,500, half up front

The care plan

We check every release and stay on call for when something breaks.

$300 to $500 a month, cancel any time

What it finds

Problems you cannot see from the screen

In two open-source AI-built apps we audited in October 2026, both let any visitor give themselves the paid plan without paying. One accepted a negative number to add credits. One trusted whatever credit count the browser sent at checkout. None of it was visible from the user interface.

What we check

49 checks in 8 areas

Including your git history, where keys removed from the code still live.

  • Secrets and keysKeys in the code, the browser bundle or old commits
  • Data accessWho can read and change other people's data
  • Sign-in and sessionsWhether the server checks who is calling
  • Untrusted inputInjection, unsafe HTML, uploads
  • PaymentsWebhooks, prices and who gets paid features
  • Data loss and privacyBackups, deletes, personal data
  • Runaway costWho can run up your AI and email bills
  • ReliabilityErrors, tests, vulnerable dependencies

How it works

Agents read every file. An engineer signs.

  • AI agents read every file against the checklist, and a second set of agents tries to disprove each finding.
  • An engineer checks every finding before it reaches you and signs the report.
  • Your code is read, never run, and never used to train AI models.
  • We only audit an app with its owner's permission, and we never test your live site.

Price

Fixed prices, quoted before work starts

ServicePrice
Audit, up to 10,000 lines$149
Audit, up to 30,000 lines$249
Audit, up to 60,000 lines$349
Fix package$1,500 to $3,500
Care plan$300 to $500 a month

The first five audits are free, in exchange for a short testimonial and permission to describe the findings without naming you.

Book an audit

Send us the repository. Get the report in 48 hours.

Email us the repository link, where the app is hosted, roughly how many users it has and anything you are worried about. We reply with the price for its size and how to give us read-only access.