Built it with AI? Let’s make it safe to launch.
AI builders get you to a working demo fast. What they skip is the part that matters once real users and real money arrive: access rules, secrets, payments, backups and monitoring. I review what you have, fix what is risky, and leave you able to keep building.
- → Written findings, ranked by risk
- → Fixed price before any fix
- → Audit fee credited toward fixes
- → Honest “rebuild or not” answer
- → A security review, not a pen test
Twelve things to check before launch.
Go through these yourself. If more than two are “not sure”, an audit will pay for itself. For a quick start, the free app safety check finds exposed keys, public source maps and a readable .env or .git in about half a minute.
- 01Row-level security is on for every tableIn Supabase, a table without RLS policies can be read or changed by anyone holding your public anon key — which ships in your frontend.
- 02No service-role or secret key in the browserSearch the built JavaScript for your service-role key, Stripe secret key and any API key that should stay on the server.
- 03Secrets are not in Git historyA key deleted in a later commit is still in the history. Rotate it, then remove it from the repo.
- 04Every API route checks who is askingGenerated endpoints often trust an ID from the request. Try fetching another user’s record by changing the ID.
- 05Payments are confirmed by webhook, not the browserMark an order paid only when a signed webhook from Stripe says so — never because the checkout page redirected back.
- 06Sign-in and sign-up are rate limitedWithout limits, one script can try thousands of passwords or create thousands of accounts.
- 07Input is validated on the serverClient-side checks are for convenience. The server must reject bad data on its own.
- 08Backups exist — and one has been restoredA backup you have never restored is a hope, not a backup. Restore one into a scratch database.
- 09Production and test data are separateExperiments and seed scripts should never run against your live database.
- 10Errors reach you before users doUptime checks and error alerts, so you hear about an outage from a tool rather than a customer.
- 11Dependencies have no known advisoriesRun npm audit (or your stack’s equivalent) and fix high-severity issues before launch.
- 12Logs don’t store passwords or personal dataGenerated code often logs whole request bodies. Check what ends up in your logs.
Audit, fix, then keep it running.
Security Audit & Hardening
A written-scope security review of one web app and its server, including apps built with Lovable, Bolt or Cursor on Supabase. I check access control, auth, secrets, database rules and server setup, then report findings with fixes and effort. From $650 ($500 founding) for the review, about 1–2 weeks. Email your stack to start.
AI-Built App Rescue
A new service: a fixed-price audit of an app built with Lovable, Bolt, Cursor or similar tools, covering data access rules, exposed keys, auth, backups and deploy, with a clear fix plan. I make the fixes on Laravel, Vue/Nuxt, Go, Python and Postgres/Supabase SQL. Audit from $390 ($300 founding), about a week.
Remediation Sprint
A fixed-price sprint that fixes an agreed list of findings from my audit or assessment (or another reviewer’s report, after a quick check). I fix, test and re-check each item and tell you what’s done. From $650 ($500 founding), delivered in 1–3 weeks. Email the report to start.
Backup & Disaster Recovery
I set up encrypted offsite backups of your database, uploaded files and configuration, then prove they work by restoring them into a scratch copy of your app. It’s for small SaaS teams and agencies who have never tested a restore. From $390 ($300 founding), done in 1–2 weeks. Email me your stack.
Monitoring & Alerting
I set up the checks that tell you your app is down, a certificate is about to expire, a disk is filling up or a nightly job silently stopped, and route those alerts to you. It’s for small teams without an ops person. From $260 ($200 founding), done in about a week. Email me your stack and hosting.
Monthly Care Plan
A fixed monthly plan that keeps your app patched, backed up and checked, with a monthly restore test, a short report and hours for small changes. It’s for small SaaS teams and founders without an ops person. From $195 a month, minimum 3 months. Working hours only, no 24/7. Email me to check fit.
Straight answers.
Which AI builders do you work with?
Apps built with Lovable, Bolt, Cursor, v0, Replit and similar tools. I audit any stack; I fix on stacks I support — Laravel, Vue and Nuxt, Go, Python and Postgres or Supabase SQL — and tell you straight when a rebuild or another specialist is the better call.
Have you rescued AI-built apps before?
This is a new service, so there is no case study for it yet. The work itself — security reviews, access rules, backups, tests and deploys — is what I have done on the systems in my case studies, and the fixed price and written scope protect you.
Is this a penetration test?
No. It is a hands-on security review with written findings and fixes. If an investor or customer needs a formal penetration test, I will tell you so.
Can I keep building with AI afterwards?
Yes. I leave notes on the patterns to keep — and the ones to stop the AI from generating again.
What does it cost?
The audit starts at a fixed price shown below. Fixes are quoted from the audit’s findings, and the audit fee is credited toward them.
Tell me what you built.
Which tool, which backend (Supabase, Firebase, your own), how many users and what worries you. I reply within one US business day.
3 of 3 client slots open · next start Oct 5