For people building with Lovable, Cursor, Replit, Bolt and v0
Shipped with AI.Checked for leaks.
AI tools write a lot of code fast, and they repeat the same few privacy mistakes: a database readable from the browser, a key in the frontend, a user record pasted into a model prompt, an email in the logs. Sentinel finds them and explains each fix in plain words you can paste straight back into your AI tool.
Sound familiar?
The problem, plainly.
Your database may be readable from the browser
Missing Supabase or Firebase rules are the most common way AI-built apps expose user data.
Your users’ emails are in your logs
console.log(user) feels harmless until the logs land in a third-party tool.
Keys ended up in the frontend
A service key in a client file is a key anyone can read in the browser.
Every finding names the file and line, says why it matters, and gives a short fix you can hand to Lovable, Cursor or Replit as a prompt.
Your code is read on a machine with the network off, never run, then shredded. No GitHub permissions needed.
One clean month earns Privacy-Ready Bronze, with a dated page anyone can open.
How a scan goes
Four steps, the way you’ll actually do it.
Zip the folder
Export from Lovable, Cursor, Replit, Bolt or v0, or just zip the project. No GitHub permissions to hand out.
It opens in a sealed machine
A fresh microVM with the network off. Your code is read, never run, then shredded.
Read it like a code review
Each finding names the file and line, says in plain words whose data could leak and where, and gives a short fix you can paste back into your AI tool.
Fix, re-check, show it
Your fix-check rescan is free. A clean month earns Privacy-Ready Bronze and a badge that links to a dated page anyone can check.
Sample findings
What a first scan tends to turn up.
Each row names the rule that finds it. What we check →
Illustrative first scan of an AI-built app. Yours will differ; the shape rarely does.
The plan that fits
Start with one scan and a free fix-check rescan.
Recommended
Prove it once. Get the badge.
one time
- Plus one free fix-check rescan
- One full Sentinel scan of your codebase
- Findings grouped by privacy risk, each with the exact fix
- Black Llama Privacy-Ready · Bronze badge on a clean result
- Badge valid for 3 months, then re-scan to keep it
$499
1 scan + 1 free fix-check rescan, badge valid 3 months.
Other plans: Path to Platinum $299, Sentinel for Programs $250. Compare all →
Privacy-Ready badges describe automated code scans for the stated period and scope. Not an audit, attestation or SOC 2 report; SOC 2 reports come only from licensed CPA firms.
Questions
Asked by ai builders.
Do I have to connect GitHub?
No. Upload a zip. Connecting a repo is optional and always read-only.
Will you run my app?
Never. Code is parsed and pattern-matched, not executed. No install, no build, no start.
Can I paste the fixes into Lovable or Cursor?
Yes. Each fix is a short, plain instruction tied to a file and line, which is the shape AI tools work best with. Check the change before you ship it, as you would any AI edit.
Does it check my Supabase or Firebase setup?
It reads what is in your code: Firebase and storage rules files (Beta) and keys that would bypass your rules. It does not log in to your live database.
What if the scan finds a lot?
Fix the criticals first; the report sorts them to the top. Warnings don’t block Bronze. Your first rescan is free, so you can confirm the fixes.
