I Will fix the app you built with Lovable, Bolt, Cursor or v0 and get it ready for real users

AI builders get an app to the demo stage fast. The gaps show up later: a login that fails for some users, a payment that goes through but never unlocks anything, data that any logged-in user can read, a secret key sitting in the browser bundle, or a deploy that only works from the builder's own preview. I work on exactly that layer, in the code your tool produced, and I don't rebuild what already works.

Starter is a go-live check: I read the code, run the app, and send a written list of problems ranked by what hurts first, each with where it lives and what the fix involves. Standard adds up to five fixes from that list, made in your repository. Advanced takes up to twelve fixes, adds tests around login and payments, and sets up a deploy you can repeat without the builder.

The check starts with what generated code tends to miss. First come access rules on the database: without them, users can read each other's records. Many of these apps run on Supabase, which is PostgreSQL underneath; there the access rules are row-level security policies, and I write and test them. Next, I look for secrets that belong on the server but end up in the browser, such as a payment provider's secret key or Supabase's service key. Then I check that payment webhooks verify the provider's signature, that forms don't trust whatever the browser sends, and that a failed request shows an error instead of a blank screen.

Stack: React, Next.js and Vite front ends, Node.js and TypeScript on the server, PostgreSQL. I use AI coding tools myself every day, and I read and test everything they produce.

If your tool syncs with a GitHub repository, I work in that repository, so you can keep building with the tool afterwards. Otherwise I work from an export and hand the code back in a repository you own. For 14 days after delivery, I fix anything that does not work as we agreed, free of charge.

No calls. Scope is agreed in writing before I start; after that, you get a written update at the end of each working day and a note per fix saying what was wrong and how to check it.

Steps for completing your project
1. Access and a first run - You share the repository or an export and a test account. I get the app running the same way you do before touching anything.
2. Go-live check - I go through login, data access, payments, secrets, error handling and the deploy, and write each problem down with its location and the size of the fix.
3. You choose the fixes - The list comes ranked. You pick what goes in, within your package, and anything beyond it is quoted before I start.
4. Fixes with proof - Each fix lands as its own commit with a short note. Where a rule protects data, a test shows that one user cannot reach another user's records.
5. Handover - A written summary of what changed, what is still open, and how to deploy again. On Advanced, the deploy runs from your repository without the builder.