You built something with Cursor, Lovable, Bolt, v0, or Replit and it works — until it doesn't. AI writes code that runs; it doesn't write code that's safe to put in front of paying users. I'll go through your codebase and tell you exactly what will hurt you, ranked by how badly.

This is triage, not a rewrite. You get a clear answer to one question: what has to be fixed before this goes live?

Security pass: exposed API keys, secrets committed to the repo, missing auth on endpoints, unprotected admin routes

Database access review: direct client-side DB access, missing row-level security, over-permissive policies

Features that look finished but aren't — hardcoded values, stubbed handlers, mocked responses left in production paths

Money and data risks: unbounded AI API calls, no rate limiting, no backups, runaway cost paths

Deployment blockers that will fail on first real traffic

Written report with every issue ranked P0 / P1 / P2, each with the exact file and line

30-minute walkthrough call

Scope: One codebase, up to ~15,000 lines. Findings and recommendations only, implementation isn't included.