Skip to content

// Overview by tool

Security by AI build tool: which to review first

7 minUpdated June 30, 2026By Ler em português

The gap between prototype and product

Tools like Lovable, Bolt, v0, Cursor, Replit and Firebase Studio make the MVP work fast. The problem appears at the boundary between "works in the demo" and "holds real customers": server-side authorization, per-customer isolation, input validation and control over what the AI can do usually get left for later, and "later" arrives with the first paying customer.

The tool is not the problem. But each one repeats a different dominant hole, and knowing yours directs the review. This overview summarizes; each tool has a dedicated page with the real incident, the CVE and how to review.

Where each tool fails (dedicated page under Keep reading)

Each row has its own page under /tecnologias with the detail, the incident and the review steps.

Dominant hole by tool
ToolDominant holeAnchored in
LovableRLS off on Supabase, anon_key dumps the tableCVE-2025-48757 (170+ apps)
Bolt.newservice_role and paid key in the client bundleScrapers find it in hours
v0dangerouslySetInnerHTML and key in client componentXSS at integration
CursorIDOR (checks login, not the owner), unvalidated JWTCVE-2026-50548 (RCE)
Replitagent with production access, endpoint with no authThe wiped-database freeze case
Firebase Studiopermissive Security Rule (if true), open bucketID-photo leak in 2025

How to use this overview

Find your tool in the table, open the dedicated page under "Keep reading" and run the RET Promptbook for a first paid review of the real flows (access, data, checkout, uploads) before more users enter the risk. Serious signals go to Risk Review.

Want to know whether this happens in your system?

Manual pentest for web applications, APIs and apps, from R$9,500, with the Sprint run within 7 business days. We refund 50% if we find no flaw.

See price and timeline for my case

Frequently asked questions

Is an app built in Lovable/Bolt/v0/Replit insecure by nature?

No. The generated code can be good. The risk is shipping sensitive flows without reviewing the boundary (authorization, isolation, secrets) the generator does not close automatically.

Do I need to migrate tools to be secure?

Usually not. The review inspects the product flow, not the code’s origin. Migrating without reviewing just moves where the same mistake lives.

I use Supabase; does RLS solve everything?

RLS helps a lot, but only if it is enabled and correct. Many projects leave policies permissive or disabled. Confirm the rules and combine them with application-level checks.

Keep reading